It’s just physics, but it’s so much more.

The original description of a complete mathematical model for estimating cycling power output has to go to Martin. But it was Antoine Vayer’s snarling bulldog sound bites and his collaboration with Fred Portoleau on Not Normal Doping that brought the power estimates and the debate over what the numbers meant into the mainstream’s view.

Although the debate over the interpretation of the numbers is truly just getting started, the method itself is not quite the pseudo-science that some would prefer you believe. Mainly the method is based on relatively straight forward physics; the total power is the sum of the power to overcome gravity, wind resistance, rolling resistance, and drive train drag. Yes that’s basically it.

The details, of slightly simplified version of the equation, look like this:

What is most important though is how the math holds up in the real world. To find out Vetoo collected 250 power meter files from 36 Pro-tour riders over 250 climbs with adequate video footage to mathematically estimate power-outputs. I then took this data set and ran some basic statistics in R and the results were quite good.

As you can see in the figure above, the mathematically estimated power correlated well with the directly measured power meter data.

Statistically the estimate are “likely” to be within +/- 1.6% (75% confidence interval) of power meter measurements and “very likely” to be within +/- 2.7% (95% confidence interval). For perspective, the “gold standard” SRM power-meters are only reported to be accurate to within +/- 1-2% themselves.

A quick quality check, shows that the residuals are have a reasonably normal distribution without much weirdness going on.

So unsurprisingly to me anyways, it turns out that there is nothing magical about the physics of pro-cyclists. They too follow at least some of the fundamental laws of nature.

Now it is true strong consistent head or tail wind can throw an estimate off by 5-10% or more. But as we can see from the 250 climbs above, this scenario looks like a rare event in Pro-tour races.

The post Intro to Power Estimate Modelling appeared first on veloclinic.

Cycling related pain and overuse injury is historically very common among cyclists. From a 1 year recall period cyclists reported complaints related to the neck (48.8 %), knees (41.7 %), groin/buttocks (36.1 %), hands (31.1 %), and back (30.3 %) (Wilber et al, 1995). Similarly high rates of complaints were found during bicycle tours related to the buttocks (32.8%), knee (patellar and IT band) (20.7%), neck-shoulder (20.4%), groin (10%), palm (10%) (Weiss et al, 1985), buttocks (42%), crotch (34%), upper leg (25%), neck (24%), knee (24%), hand or fingers (19%), shoulder (17%), foot or toes (17%), back (16%), wrist (8%), and lower leg (6%) (Dannenberg et al, 1996).

Percent of Cyclists With Complaints by Body Region

The primary clinical intervention to treat/prevent cycling related pain is the bike fit. Bike fit is the process of optimizing the 3 dimensional configuration of the cleats, seat, and handlebars for the individual cyclist. Of the bike fit parameters, multiple authors agree that correct saddle height is not only important for knee injury prevention but for performance maximization as well (Obrien 1991, Asplund et al 2004, Holmes et al 1994, Shanner and Halloran 2000, Wanich 2007, Bini et al 2011). At the time of this writing, the most recent recommendations from a comprehensive review of the scientific literature recommends assessing appropriate saddle height by measuring knee angle at bottom dead center (BDC) and achieving a target range of 25-30 degrees flexion (Bini et al 2011). This recommendation is in line with the major commercial fit system recommendations of a 25-35 degrees knee angle at BDC (SICI 2007, Swift and Schoenfeldt). However, these recommendations are based on incomplete evidence (Bini et al 2011).

The initial recommendations for assessing appropriate saddle height are based adjustments to saddle height to match percentages of lower extremity (LE) measurements found to maximize performance parameters (Hamley and Thomas 1967, Shennum and deVries 1976, Nordeen-Snyder, 1977). Experimental evidence also demonstrates that changes in seat height relative to these LE measurements affects both tibiofemoral forces as well as patellofemoral forces (Ericson and Nissell (1986 and 1987). However, a limitation of this early work is that anthropometric measurement based methods result in a wide range of kinematic responses in terms of the knee joint angle achieved at the BDC (Peveler et al, 2005). More recent studies improve upon performance maximization using a target knee angle of 25 degrees BDC as compared to the traditional method (Peveler et al 2007, and Peveler 2008, Peveler and Green 2011). The use of a target knee angle at BDC to assess appropriate seat height is further supported by studies describing strong kinematic relationships between knee angle at BDC and large 4 – 10% changes in seat heights (Nordeen-Snyder 1977, Price and Donne 1997, Sanderson and Amoroso 2009). A recent 3D study, has confirmed the effect of smaller 2 cm seat height changes on hip, knee, and ankle kinematics (Puchowicz et al. 2013).

The major commercial fit methods of SICI and BIKEFIT also primarily use the 25 -35 degree knee angle at BDC to assess saddle height as part of a whole body optimization (SICI 2007, Swift and Schoenfeldt). Cyclists undergoing a bike fit are further instructed that they may need to adjust the seat up or down by up to 1 cm or approximately plus minus 1% of seat height to accommodate for individual comfort and functional status. While any change greater than this suggests there position is out of range and needs reassessment (SICI 2007). Interestingly, while the goal of seat height adjustment has been to alter knee flexion, the only kinematic study comparing cyclists with and without knee pain found no difference in knee flexion (Bailey et al. 2003).

Another focus of bike fit  theory is that knee pain is the result of  repetitive strain from shank abduction and subsequent medial motion of the knee in response to maximal loading during the powerphase of the pedal stroke(Francis 1988, Sanner and O’Halloran 2000, Hannaford et al 1986, Asplund et al 2004, Obrien 1991, Wanich et al 2007, SICI 2007, Swift and Schoenfeldt). Varus wedges mounted under the cleat are recommended as an intervention to support the medial column, decrease shank and knee motion during the power phase, and thereby reduce knee stresses and injury risk (Francis 1988, Sanner and O’Halloran 2000, Wanich et al 2007, SICI 2007, Swift and Schoenfeldt). Due to the presence of forefoot varus in 87% of normal people (Garbalosa et al 1994), it is has been suggested that the normal foot will collapse medially when called upon to act as a rigid lever in cycling (Swift and Schoenfeldt) and that the majority of cyclists would benefit from a  varus cleat wedge (Sanner and O’Halloran 2000, Swift and Schoenfeldt). Bike Fit Systems and Specialized both market plastic cleat wedges for this purpose (bikefit.com, specialized.com).

Francis (1988) put forward the the initial theory linking foot pronation, shank abduction, and knee injury. Based on theoretical modeling, his theory states that as the foot is loaded pronation results in shank abduction during the power phase of the pedal stroke, 30 to 150 degrees (Cavanagh et al, 1988), followed by the shank returning to a neutral position in the recovery phase as the foot is unloaded and pronation is no longer a factor. The power phase shank abduction is seen clinically as medial motion of the knee towards the top tube during the down stroke. The implication is that the normal non-driving moments acting on the knee (Davis et al, 1981, Ericson et al, 1984) would be increased. Effects would then be compounded by the interaction of varus/valgus moments with axial moments to produce more knee stress than either alone (Mills et al, 1991). This theory is supported by the finding that joint moments are significantly increased by forefoot varus, and that the shank follows the expected sequence of abduction during the power phase and adduction through the recovery phase (Ruby, 1992). Unfortunately, simply allowing multi degree freedom at the pedal is not effective in reducing joint moments (Boyd et al, 1997). Instead the authors conclude that pedal parameters needed to be adjusted on an individual by individual basis. Studies on injury are limited but increased medial lateral knee motion (Hannaford et al 1986) and increased shank abduction have been described in subjects with a history of knee pain versus controls (Bailey et al 2003) lending further support to the theory.

Attempts to correct shank abduction during the power phase using cleat wedges have been mixed. Sanderson (1994) found a position shift in lateral extreme of motion but no clear effect on range of medial lateral motion of the knee in response to varus and valgus wedges. Pilot data utilizing 3D analysis did confirm the positional shift, but from the frontal plane positional shift of the ankle center rather than changes in shank ab/adduction (Puchowicz et al. 2011).

In contrast to reductionist approaches to manipulating bike fit for the prevention or management of injury, the kinematic data as a whole suggests an integrated view is likely necessary. Although the seat is a fixed point relative to the bottom bracket the cyclists interaction with this contact point is not. During normal pedalling the hip translates forward and downward (Sauer et al 2007). Changes in hand position have been shown to affect pelvic tilt (Sauer et al 2007) and trunk lean is connected to ankle PF (Dingwell et al 2008). Additionally, knee ROM, knee flexion at BDC, ankle plantar flexion (PF) at BDC (Nordeen-Snyder 1977, Sanderson and Amoroso 2009, Price and Donne 1997, Puchowicz et al. 2013), hip rocking, and pelvic tilt (Price and Donne 1997) have all been shown to alter with changes in seat height. Similarly, kinematics are altered during static versus dynamic measurements, as well as with increasing load, and with fatigue (Peveler et al. 2012). Taken together, it is evident that the effect of any bike fit change, or musculoskeletal dysfunction is likely to be spread over several body segments functioning and dependent upon the dynamic state at time of measurement.

The interconnectedness of cycling kinematics as it relates to injury is reflected in the multi-point and overlapping bike interventions that are suggested for the variety of overuse injuries found in cyclists. Limiting this discussion to the lower extremity, 3 recent in depth reviews the following table illustrates the interventions suggested for each injury:

B = (Bini et al 2011), S = (Sanner et al 2000),  W = (Wanich et al, 2007). A letter indicates a suggested interventions, a “?” preceding a letter indicates an intervention to be evaluated.

Unfortunately, the vast majority of bike fit interventions suggested are supported only by anecdotal experience and have not been tested in a scientific manner. Experimental testing is of particular importance in areas of competing theories. For example, Sanner (2000) suggests that a high seat position increases medial lateral knee tracking due the greater difference in medial and lateral femoral condyle circumference engaged near full extension putting cyclists at risk for knee injury in contrast to the “common knowledge” that too low of a seat causes anterior knee pain due to increase patellofemoral compressive forces. Sanner (2000) goes on to note that his experience has shown VMO overdevelopment in cyclist with patellofemoral dysfunction perhaps as a compensation for poor bike fit/mechanics. Similarly, pilot data on the effect of cleat wedges failed to show changes in shank abduction during the powerphase with varus or valgus wedges (Puchowicz et al. 2011). Instead, the positional shift of the proximal shank previously found (Sanderson et al 1994) appeared to be the result of a symmetric shift of the entire shank across the pedal stroke (Puchowicz et al. 2011). This finding challenges the theory behind the use of wedges to decrease powerphase shank abduction (Francis 1988, Sanner and O’Halloran 2000, Wanich et al 2007, SICI 2007, Swift and Schoenfeldt). Such practice carries a risk of inappropriate varus wedging and potential exposure of the knee to increased varus moments (Wolchock et al 1998, Gregersen et al 2006).

The ultimate goal of bike fit should be to make fit interventions that reduce the risk of injury in a preventive manner as well as treat active disease. However, the current paucity of clinically relevant studies makes it difficult to know where to begin testing. At this point, the most logical point is with a thorough prospective investigation to identify cycling kinematic dysfunctions that correlate with injury risk. This information would allow the targeted testing of bike fit interventions ultimately setting the groundwork future clinical trials.

Conclusion:

Cyclists, due to the repetitive nature of the activity, are at extremely high risk for overuse injuries. A variety of bike fit interventions have been suggested to treat and prevent injury based on often competing and untested theories of kinematic dysfunction.

The post What is the Evidence Behind Bike Fit ? appeared first on veloclinic.

In the search for a competitive edge athletes often look to the practices of top athletes, or the latest theory from a popular guru. By chance, this approach will sometimes lead to a true performance improvement. But usually, the fad ultimately proves to make no difference or even hurts performance. An alternate approach is to look at evidence from research studies where ideas were more effectively tested. The evidence based approach will always seem behind the latest trends. It often takes 2 or more years from an idea to make it through the process of research design, testing, analysis, and publication. But in those same 2 years, many of the once hot trends are ultimately abandoned.

The purpose of this guide is to help athletes and coaches develop nutrition and hydration strategies based on quality evidence. Most of the information in this guide comes from the position statements from the American College of Sports Medicine on nutrition and fluid replacement. These statements were published in 2009 and 2007. After spending some time reviewing the literature, these two position statements still represent some of the most comprehensive reviews available on this topic. They provide a good foundation to build a strategy to guide food and drink choices for endurance athletes. Also, incorporated into this guide is the 2010 review article published in Sports Medicine (O’Reilly et. al) on the topic of the effects of glycemic index on performance and metabolism.

To help illustrate how this information translates into practice, an example 150 (70kg) “typical” cyclist is used for calculations. When the recommendations seem to stretch the limits of practicality I offer suggested modified strategies.

Some important things to keep in mind when using this guide:

All recommendations will need to be tailored to your own personal needs.

Nutrition and hydration deficits will cause decreases in performance but excess intake will not improve it.

Stick to familiar nutrition or hydration on race day.

Use training rides to try out new foods and drinks.

When experimenting, don’t be fooled by a particularly good or bad day. Instead, use what works most consistently.

Eat real food. If it looks like something that could be found in the wild it is probably a better choice than the processed wonder food equivalent.

Daily Needs

Carbohydrates, Protein, and Fat:

The endurance athlete should take in 6-10 gm/kg of carbohydrates and 1.2 – 1.7 gm/kg of protein. For a typical 70 kg (150 lb) athlete that would be about 420 – 600 gm of carbs per day and 84 – 119 grams of protein. Fat should make up 25 – 30% of the total calories. (A quick way to convert your own weight in pounds to kilograms subtract 10% then divide by 2, for example 155 lb, 155-15 = 140, 140 / 2 = 70).

For perspective of what these numbers mean, one popular sandwich restaurant chain claims a foot-long oven roasted chicken sandwich with cheese, veggies, and a bag of baked chips would get you 143 grams of carbohydrates, 50 grams of protein, and 39 grams of fat (or 20% of total calories). Notice that eating an equivalent meal 3 times a day would get you to the lower range of carbs, quite a bit over for protein, and a little short on your fat intake.

From a realistic standpoint, tracking every gram you take in probably won’t happen. A simpler method that works out reasonably well is a 50/25/25 approach. When you sit down to eat, fill half of your plate with vegetables and fruits, a quarter with a lean protein, and the other quarter with a complex carb such as whole grain pasta or a baked potato. The extra carbs you need will come from intake during and around workouts. Rounding out this diet with “healthy” fats by cooking with olive oil, snacking on nuts, or having a fatty fish as your protein choice etc. should get you fairly close to the recommended intake.

Once every couple of weeks or so you can spot check your diet on a typical day to make sure your balance of foods is appropriate. Comparing your actual intake to the general recommendations should give you a sense of any major deficiencies.

Vitamins or supplements shouldn’t be necessary unless you have diet restrictions or are actively trying to lose weight.

Periodically, the cycling press has made the high fat or low carbohydrate diets a hot topic. The concept is that you can train your body to burn more fats during exercise resulting in better endurance by sparing carbohydrate (glycogen) stores. There are some studies that support this idea. However, such an approach carriers a fair amount of risk because of how important carbohydrates are as a fuel source in the typical diet.

Pre-Race Nutrition

Ideally, a high carbohydrate meal should be eaten 3-4 hours prior to the event. Eating 200 – 300 grams prior to events has been shown to improve performance in some studies, and 3 to 4 hours is usually enough time for the stomach to empty. To get this amount of carbohydrate for 9 am start an athlete would need to eat the equivalent of 4-5 bagels at 5 in the morning. Such a large meal might be a bit tough all on its own let alone so early in the morning.

The practical approach is to simply eat a reasonable amount of carbohydrates as early you can. Adjust the size of the meal based on how much time you have before the race to avoid an upset stomach. If you need to eat close to start of the race it may also be a good idea to avoid foods high in fiber, fat, and protein. Foods high in these nutrients will slow down the rate that your stomach can empty leaving you uncomfortable during fast starts.

When choosing pre-race carbohydrates many experts tout the benefit of complex carbohydrates, or low glycemic index carbs. (For simplicity complex carbohydrates are generally starchy foods, while low glycemic index carbohydrates can be starchy foods or foods that contain protein or fat which lower the glycemic index). The complex/low glycemic carbohydrate strategy appears to effective if the only carbohydrate intake will be the pre-race meal. Taking in sufficient carbohydrates during the ride eliminates the performance advantage of the complex/low glycemic carbohydrates. O’Reilly et al make the point that the type of carbohydrate doesn’t really matter as long as you can optimize the amount of carbohydrate before and during your event.

However, I do still recommend complex carbohydrates and low glycemic index foods whenever possible because they tend to be in more nutritious foods. Also, in the real-world setting carbohydrate intake is often not maximized during sports. This reality leaves open the possibility to gain some performance benefit of complex and low glycemic index pre-race carbohydrates.

Notes:

It is ok to eat right up until the start of the event as long as your stomach can tolerate it. Studies looking at the effect of eating within an hour of an event do not show any decrease in performance. This evidence goes against the popular teaching that eating within the hour or two before an event should be avoided.

Pre-Race Hydration

Drink 5 – 7 ml/kg of water (a little less than a 24 oz water bottle) 4 hours before the event. At two hours before the event if you are not urinating clear or very light yellow (dark yellow urine can be a sign of dehydration) drink another bottle. Drinking this amount should correct any minor dehydration. If you are significantly dehydrated, more water will not necessarily help as the  body can not correct deficits much faster. It is important to drink far enough in advance of the event to give your body plenty of time urinate out excess water. Avoid drinking plain water in the hour before the event as you are likely to find yourself with a full bladder just as you roll off from the start line.

Drinking fluids with salt or protein can help your body hold on to more fluid than drinking water alone. Logic would imply that this may be an effective strategy for events where drinking is difficult or for particularly hot humid days. However, hyper-hydrating has not been shown to improve performance. Attempting to hyper-hydrate with glycerol is specifically not recommended.

Although there may be no performance advantage, a sports drink is likely a better option than plain water in the hour before the start to reduce the chance of needing to urinate at the start of the race. After the race is underway the rate of urine production slows significantly making the need to urinate less of an issue.

Race Nutrition

Carbohydrates

Take in 30-60 grams (1-1.5 bananas, 1-2 gels, or 1-2 bottles of sports drink) of carbohydrate every hour during events that are greater than 1 hour. This amount represents the upper end of what most athletes can absorb during exercise. Carbohydrates that are not absorbed can lead to abdominal discomfort and diarrhea. Experiment during training to figure out how much carbohydrates intake you can tolerate during exercises. When possible choose real foods over processed options

It is important to realize that your body can not absorb enough carbohydrate to keep up with the demand of moderate to intense exercise. Your intake of carbohydrates simply extends the time you have before you run out.

For sports drinks, concentrations of greater than 8% for drinks will empty slower from your stomach. If you stick with the mixing instructions, or using just enough to add some flavor, you will be fine.

Many sports drinks make the claim that maltodextrin, an other polymers, are complex carbohydrates. And as a complex carbohydrate, they have an advantage in increasing your ability to take in large quantity without slowing down stomach emptying or faster absorption. These claims are somewhat misleading as maltodextrin is an chain of glucose molecules connected end to end, i.e. a chain of simple sugars. This chain is broken easily by digestive enzymes and the process starts as soon as it hits your mouth. (Maltodextrin actually has a higher glycemic index than table sugar.)

The type of complex carbohydrate that will not get broken down as quickly have multiple branches rather than a chain. These types of carbohydrates will not taste sweet and are found in typically in starchy foods.

The type of carbohydrate/sugar doesn’t seem to matter from a performance standpoint as long as it is not just all fructose. Fructose by itself is not taken up as fast as glucose or glucose fructose combinations. I have not come across any sports products that contain straight fructose. High-fructose corn syrup, which is found in soft drinks and some less expensive sports drinks, probably isn’t as bad as its current reputation as the cause of the obesity epidemic. It is actually fairly equivalent in fructose and glucose balance to sucrose (cane and beat sugar), and honey. Rice syrup contains some complex carbohydrates in addition to simple sugars.

The advantages to different carbohydrates/sugars is mostly the sweetness/flavor and cost.

Protein

Protein intake does not immediately improve performance unless your intake of carbohydrates is suboptimal. The original studies that showed improved performance of 4:1 carbohydrate to protein compared a suboptimal intake of carbohydrates to carbohydrates plus protein. Followup studies that made the total calorie intake the same did not show a performance advantage. However, there may still be a theoretical advantage to taking in protein during exercise as it may spare muscle breakdown and promote muscle growth after exercise. The theoretical discussion is beyond the scope of this guide.

Fats

Fats represent the largest fuel reserve in your body. It is not necessary to take in additional fat during exercise. At low intensity the majority of your energy can come from burning fat. However the rate at which you can burn fat is limited and dependent on carbohydrates. As intensity increases your body is forced to burn a higher percentage of carbohydrate. When your body runs out of carbohydrates you not only lose your ability to sustain high intensity exercise but you also lose your ability to efficiently burn fats making even low-moderate levels difficult to sustain.

Strategies to improve fat utilization, i.e. spare carbohydrates, may provide a performance advantage. These strategies include improving aerobic fitness, intelligent pacing, and diet manipulation. Improving fitness and good pacing are logical and easy to endorse. Diet manipulation however, is difficult to recommend because of the significant potential to cause more harm than good.

Electrolytes

Electrolytes (salts) have not been shown to improve performance. Typical western diets contain plenty of electrolytes to replace anything lost during exercise.

The advantages of salt containing sports drinks are for taste, and to lessen the risk of hyponatremia. Taste is self explanatory. Hyponatremia, or low salt concentration, usually occurs in less experienced endurance athletes who drink too much plain water during an event. As exercise slows kidney function, the body loses its ability to get rid of excess water. In this state an athlete can potentially drink enough water to cause a low salt level caused by dilution. Theoretically, salt containing sports drinks decrease this risk.

Conventional wisdom is that the loss of salt and fluids causes cramping. However, replacing fluids and salts has not been shown to prevent cramping. It has been shown that improving fitness and acclimatization to hot environments reduce cramping AND excessive salty sweating. An alternate explanation may be that both cramping and excessive salty sweating are caused by over-reaching or suboptimal fitness/acclimatization.

Water

Studies show that dehydration (losing 2% of your body-weight) leads to decrease aerobic performance. Aerobic performance continues to fall off as dehydration worsens. Anaerobic performance is less affected by dehydration.Water intake should roughly match your rate of water loss through sweat and evaporation. For endurance athletes this can range anywhere from 0.5 – 2 L per hour depending on temperature, intensity, and individual physiology.

For the hypothetical 155 lb (70kg) rider, a loss of about 1.4kg (about 3 lbs) or 1.4 liters would equal 2% dehydration. Saving a little margin of error it is probably a good idea for the average rider to not to lose more than 1 liter (a little more than a 24 oz water-bottle).

On hot days a good strategy might be to assume a high sweat rate of 1.5 – 2 liters per hour and try to drink enough to avoid losing more than 1 liter. Thinking about hydration in these terms, replacement fluids need to be progressive as the length of the ride increases. For example you might lose:

.75 – 1 L in 30 min

1.5 – 2 L in 1 hr

3 – 4 L in 2 hr

4.5 – 6 L in 3hr

Assuming you have about 1 liter to lose comfortably without affecting performance your minimum intake to prevent dehydration would need to be:

0 L (0 bottles) for 30 min

.5 – 1 L (1 bottle) for 1 hr

2 – 3 L (3 bottles) for 2 hr

3.5 – 5 L (6 bottles*) for 3 hr

*Notice that the 3 hr estimate of 6 bottles is a very large amount of water. Less experienced athletes should gain experience optimizing their hydration with exercise bouts in the 2 hour range to decrease the risk of hyponatremia and dehydration.

The calculations above are likely to overestimate water intake needs for longer events and for well conditioned, properly acclimatized athletes. Also, rate of sweating tends to fall off with time and as the body becomes dehydrated.

Ultimately, trial and error is needed to figure out what your body will need during any given circumstances. Spot checking yourself with pre and post-ride weights can tell you if you are drinking enough, see below.

Recovery

Carbohydrate

Carbohydrate intake should be your priority following hard rides. Within the first 30 min take in 1 – 1.5 grams/kg, about 100 grams for an average rider. Take in another 100 grams every 2 hours for the next 4 – 6 hours or until you have a main meal.

There appears to be mixed data for the best type of carbohydrate to eat after exercise. Without a clear answer my default recommendation is again complex carbohydrates and low glycemic index carbohydrates.

Protein

Protein intake is also a good idea, especially if you are trying to gain muscle mass or have difficulty maintaining it. Studies do show that taking in protein after exercise may promote muscle repair and growth. For endurance athletes the ideal post-exercise intake is not entirely clear. But given that the daily intake is about 1:5 protein to carbohydrate you can probably use this a starting point for recovery intake as well, i.e. about 20 grams of protein in the first 30 min followed by 20 grams every 2 hours until your main meal assuming you are taking in 100 grams of carbohydrates.

Electrolytes

The typical western diet contains enough salt to replace losses during exercise.

Water

The water replacement recommendation is to drink 1 – 1.5 L for every kilogram (16-24 oz for every pound) of body weight lost during your event.

Going on the assumption that most people will not bring scale with them to events calculating your water needs is not entirely practical. Instead, start with the post-race hydration of drinking a bottle of water immediately after the event. Drink another bottle of water every 2 hours until your urine is consistently clear or very light yellow. To check your fluid status compare your pre and post exercise weights. Every 2 pounds of weight loss equals about 1 liter or water deficit (or a little more than 1 bottle).

Recovery Drinks

If real food is not available, recovery drinks are reasonable way to make sure that you get some carbohydrate and protein in immediately after an event. Just check the label to make sure that there is a reasonable ratio of protein and carbohydrate. For endurance sports err on the side of more carbohydrates as their is strong evidence for the benefits of carbohydrates for recovery while the benefit of protein is still theoretical at this time.

References

American Dietetic Association; Dietitians of Canada; American College of Sports Medicine, Rodriguez NR, Di Marco NM, Langley S. American College of Sports Medicine position stand. Nutrition and athletic performance. Med Sci Sports Exerc. 2009 Mar;41(3):709-31.

American College of Sports Medicine, Sawka MN, Burke LM, Eichner ER, Maughan RJ, Montain SJ, Stachenfeld NS. American College of Sports Medicine position stand. Exercise and fluid replacement. Med Sci Sports Exerc. 2007 Feb;39(2):377-90.

O’Reilly J, Wong SH, Chen Y. Glycaemic index, glycaemic load and exercise performance. Sports Med. 2010;40(1):27-39. Med Sci Sports Exerc. 2009 Mar;41(3):709-31.

The post Evidence Based Sports Nutrition appeared first on veloclinic.

Velo Clinic Bike Fit

Bike fit

Bike fit is the process of optimizing the three dimensional configuration of the cleats, seat, and handlebars for the individual cyclist.

Permissive Bike Fit

The key to the veloclinic bike fit is a concept called permissive bike fit. Permissive fit means that the goal is to configure the equipment to allow the cyclist to function within a range of motion that is biomechanically sound for the individual. Although the fitter should attempt to accommodate any biomechanical issues, the fitter should be realistic that the bike fit is unlikely to “fix” underlying biomechanical problems and understands that attempts to do so may put the cyclist at greater risk of injury and diminished performance. First and foremost, the fitter should strive to do no harm.

Background

Cyclists are at particular risk for overuse injuries with 85% percent of cyclists surveyed reporting an overuse injury from a 1 year recall period (Wilber et al. 1995). Neck pain was most frequently reported by (49%), followed by pain in the knees knees (42%), groin/buttocks (36%), hands (31%), and back (30%). Similarly, high rates of overuse injury have also been reported during multi day bicycle tours (Weiss et al, 1985, Dannenberg et al, 1996). Given this risk, the prevention of injury must be of utmost importance to the bike fitter.

Unfortunately, no bike fit intervention has yet been scientifically shown to prevent or treat overuse injuries. Instead, current recommendations are largely based on “expert opinion.” These opinions are typically most heavily influenced by personal experience or personal theories extrapolated from limited scientific evidence. Although attempts to put the scientific evidence first are made, the sober reality is that in most cases solid evidence is lacking. This lack of evidence is the reason behind the permissive bike fit concept. The permissive bike fit is essentially a do no harm approach. In it, the fitter should strive to allow the cyclist to function within a sound biomechanical range, avoid the extremes of range of motion, and do not attempt to force the cyclist into an “ideal position or movement pattern.”

Below, some of the evidence behind saddle height and cleat wedges will be discussed. The point of the discussion is to illustrate the issues with even these basic fit parameters, and why a healthy skepticism for current bike fit systems is warranted.

Saddle Height

Of the bike fit parameters, multiple authors agree that correct saddle height is not only important for injury prevention but for performance maximization as well (Obrien 1991, Asplund et al 2004, Holmes et al 1994, Shanner and Halloran 2000, Wanich 2007, Bini et al 2011). At the time of this writing, the most recent recommendations from a comprehensive review of the scientific literature recommends assessing appropriate saddle height by measuring knee angle at bottom dead center (BDC) and achieving a target range of 25-30 degrees flexion (Bini et al 2011). However, these authors concede that current recommendations are based on incomplete evidence (Bini et al 2011).

The Initial recommendations for assessing appropriate saddle height are based on adjusting saddle height to match percentages of lower extremity (LE) measurements found to maximize performance parameters (Hamley and Thomas 1967, Shennum and deVries1976, Nordeen-Snyder, 1977). Experimental evidence also demonstrates that changes in seat height relative to these LE measurements affects both tibial femoral forces as well as patellofemoral forces (Ericson and Nissell (1986 and 1987). However, a limitation of this early work is that anthropometric measurement based methods result in a wide range of kinematic responses in terms of the knee joint angle achieved at the BDC (Peveler et al, 2005). More recent studies improve upon performance maximization using a target knee angle of 25 degrees BDC as compared to the traditional method (Peveler et al 2007, and Peveler 2008, Peveler and Green 2011). The use of a target knee angle at BDC to assess appropriate seat height is supported by studies describing strong kinematic relationships between knee angle at BDC and large 4 – 10% changes in seat heights (Nordeen-Snyder 1977, Price and Donne 1997, Sanderson and Amoroso 2009).

The major commercial fit methods of SICI and BIKEFIT also primarily use the 25 -35 degree knee angle at BDC to assess saddle height as part of a whole body optimization (SICI 2007, Swift and Schoenfeldt). Cyclists undergoing a bike fit are further instructed that they may need to adjust the seat up or down by up to 1 cm or approximately plus minus 1% of seat height to accommodate for individual comfort and functional status. While any change greater than 1 cm suggests that their position is out of range and needs reassessment (SICI 2007). In this author’s experience, cyclists are unlikely to tolerate any saddle height change of more than 2 cm from an otherwise optimized position.

Although knee angle at BDC is the current standard of practice, no study has demonstrated the superiority of this measure over other parameters.  A more recent study using smaller changes of 3% produced inconsistent changes at the knee for high and  low conditions relative to a reference condition (Tamborindeguy et al, 2011).  These authors speculate that their effect at the knee may have been lost due to accommodation at another joint. This hypothesis is supported by studies showing that knee ROM, ankle plantar flexion (PF) at BDC (Nordeen-Snyder 1977, Sanderson and Amoroso 2009, Price and Donne 1997), hip rocking, and pelvic tilt (Price and Donne 1997) have all been shown to increase with increasing seat heights. Further variability  which may obscure the effect measurable at a single joint is introduced by the normal hip forward and downward translation with pedalling (Sauer et al 2007). Interestingly, a retrospective study of cyclists with knee pain showed an association with increased ankle dorsiflexion (9 degrees versus 4 degrees) but failed to find an association with knee flexion (Bailey et al 2003). Taken together these study raises doubt as to whether knee angle at BDC in isolation is the most the appropriate measure to assess 1- 2 cm (1 – 2%) changes that are typically employed during a clinical bike fit.

Cleat Wedges

One prominent theory in bike fit is that knee pain is the result of  repetitive strain from shank abduction and subsequent medial motion of the knee in response to maximal loading during the powerphase of the pedal stroke(Francis 1988, Sanner and O’Halloran 2000, Hannaford et al 1986, Asplund et al 2004, Obrien 1991, Wanich et al 2007, SICI 2007, Swift and Schoenfeldt). Varus wedges mounted under the cleat are recommended as an intervention to support the medial column, decrease shank and knee motion during the power phase, and thereby reduce knee stresses and injury risk (Francis 1988, Sanner and O’Halloran 2000, Wanich et al 2007, SICI 2007, Swift and Schoenfeldt). Due to the presence of forefoot varus in 87% of normal people (Garbalosa et al 1994), it has been suggested that the normal foot will collapse medially when called upon to act as a rigid lever in cycling and that the majority of cyclists would benefit from a  varus cleat wedge (Sanner and O’Halloran 2000, Swift and Schoenfeldt). Bike Fit Systems and Specialized both market plastic cleat wedges for this purpose (bikefit.com, specialized.com). However, the kinematic effects of the commercially available wedges have not been evaluated.

Francis (1988) first put forward the the initial theory linking foot pronation, shank abduction, and knee injury. Based on theoretical modeling, his theory states that as the foot is loaded pronation results in shank abduction during the power phase of the pedal stroke, 30 to 150 degrees (Cavanagh et al, 1988), followed by the shank returning to a neutral position in the recovery phase as the foot is unloaded and pronation is no longer a factor. The power phase shank abduction is seen clinically as medial motion of the knee towards the top tube during the down stroke. The implication is that the normal non-driving moments acting on the knee (Davis et al, 1981, Ericson et al, 1984) would be increased. Effects would then be compounded by the interaction of varus/valgus moments with axial moments to produce more knee stress than either alone (Mills et al, 1991). This theory is supported by the finding that joint moments are significantly increased by forefoot varus, and that the shank follows the expected sequence of abduction during the power phase and adduction through the recovery phase (Ruby, 1992). Unfortunately, simply allowing multi degree freedom at the pedal is not effective in reducing joint moments (Boyd et al, 1997). Instead the authors conclude that pedal parameters needed to be adjusted on an individual by individual basis. Across the board use of the varus wedge may actually be detrimental as varus pedal angulation has been shown to increase forces affecting the knee under experimental conditions (Gregersen et al 2006).  Studies on injury are limited to one very small clinical study that showed increased medial lateral knee motion to be associated with knee pain pain (Hannaford et al 1986) and  a larger retrospective study did show that a more medial knee position  was associated with injury (Bailey et al 2003).

What is not clear from available studies is whether the commercially available cleat wedges are effective in altering shank abduction during the power phase or in reducing injury risk. The only study to specifically test the effectiveness varus valgus cleat wedges found a position shift in lateral extreme of motion but no clear effect on medial most position or range of medial lateral motion of the knee (Sanderson et al 1994). It is difficult to draw firm conclusions as this study reported significant variability which may have been caused by the failure to control for major parameters of bike fit or the use of pedal strap systems which are not as secure as clipless pedal systems.

To date, no study has evaluated commercially available wedges or established that varus wedges decrease injury risk.

Know Your Limits

As a bicycle fitter you ARE providing an invaluable service in assessing the cyclists needs and abilities and making appropriate bike fit recommendations. At NO point are you diagnosing medical problems or prescribing interventions. For any suspected medical problem recommend the cyclist see a licensed medical provider.

Assessing the Cyclist

Goals and History

Identify Primary Goal: Start by finding out what is most important to the cyclist. This goal may be anything from improved performance to accommodating an injury. This goal will take priority when balancing the various aspects of the fit.

Identify Secondary Goals: This is typically the point when the cyclist will tell you they want it all. It is a good point to begin to temper unrealistic expectations. For example if the primary goal is to relieve neck pain then a secondary goal of getting more aero is not realistic.

Identify the type of cyclist: Recreational vs competitive, crit vs stage racer etc. Look to make sure that their Primary Goal is in line with the type of riding they actually do. If its not, reassess the primary goal.

Identify previous injuries or surgeries: Look specifically for issues that may cause asymmetries or limit the normal range of motion. Make sure that their physical limitations are compatible with their primary goal. If there is a conflict, reassess their primary goal.

Screen for any pain discomfort or nerve symptoms going from head to toe.

Discuss their previous fit and how they arrived at their current position.

Physical Assessment

Measure height, inseam, AC width, and weight.

Neck: Have the cyclist tilt their head back as far as comfortable.

• Limited range of motion will require a more upright position.

Shoulder: Check to make sure that the shoulders are an even height.

• Uneven height may be due to scoliosis or a leg length discrepancy both of which may cause the cyclist to sit twisted on the bike or one leg may be functionally shorter than the other. The fit may need to compromise between the two sides or may require shims under the cleat to accommodate the functionally shorter leg.

Back: Have the cyclist touch their toes (or as far as they can). Check to make sure the spine is straight and that one side of the rib cage is not more prominent than the other. Note the flexibility of the lower back.

• Curving of the spine or prominence of one side of the rib cage may be due to scoliosis or a length discrepancy.
• Poor flexibility of the lower back will require a more upright position to prevent the pelvis from rocking forward and closing the hip angle. Alternatively, if an aero position is desired the hip can be kept open with a combination of higher and more forward seat position.

PSIS (Posterior Superior Iliac Spine): Check to make sure that both sides of the pelvis are an even height.

• Uneven height may be due to scoliosis, a leg length discrepancy, or rotated pelvis.

Leg length: Have the cyclist lay on their back with their feet on the table and knees bent to 90 degrees. Check to make sure that the knees are the same height. Measure from the ASIS (Anterior Superior Iliac Spine) to the medial malleolus with the legs straight.

• A leg length discrepancy of the lower leg may be accommodated with shims under the cleat of the shorter leg.
• A discrepancy of the femur may cause the cyclist to sit twisted on the bike or require the fit to be compromised between the two sides.

Hip Flexion: With the cyclist lying on their back, bend one knee toward their chest until you see the opposite leg start to raise up off the table.

• Poor hip flexibility will require a more upright position to maintain a more open hip angle.

Hip Rotation: With the cyclist lying on their back, with the knee and hip flexed to 90 degrees, rotate the lower leg outward then inward.

• Poor range of motion may affect their ability to tolerate difference in stance width/qfactor.

Hamstring: With the cyclist lying on their back flex their knee and hip to 90 degrees. Then straighten the knee until you feel tension.

• Significant tightness may affect their knee extension at the bottom of the pedal stroke in the aero position decreasing performance.

Single Leg Squat: Have the cyclist stand on one leg with their hands on their hips. Instruct them to do a half squat. Check to see if the knee stays in line with the hip and foot, or if it corkscrews in. Also check to see if they let the opposite hip slouch downward.

• Either of these suggest poor neuromuscular control of the hip. Theoretically this may cause the knee to track in towards the top tube and put the cyclist at increased risk for knee injury.

Box Drop: Have the cyclist stand on a 30 cm high stool with feet shoulder width apart. Have them jump down with both feet and perform a maximal vertical jump with arms overhead towards an overhead target as soon as they hit the ground. Observe from the frontal view for the degree to which the knees collapse into the midline. From the sagittal plane observe for whether they land stiff legged with poor flexion at the knees.

• Either of these suggest poor neuromuscular control and may put them at risk for knee injury.

Arch: With the cyclist sitting with knees bent to 90 degrees feet shoulder width apart and shins vertical, measure the height at the navicular bone. Then have the cyclist stand and re-measure the height.

• Very flexible feet may benefit from arch support.
• Arch support to take up volume and give a more secure fit may also be useful.

Foot Rotation: While the cyclists is sitting, note the orientation of the feet (imagine a line from the heel to the second toe).

• Significant external rotation may require pedals with longer spindles to allow enough heel and ankle clearance.

Review the assessment with the cyclist.

• Go over any asymmetries explaining that the fit may have to compromise between the two sides.
• Go over any limitations in the range of motion and explain how they may affect the fit.
• For any suspected medical problems advise them to see a healthcare provider.

Set up the bike on the trainer making sure the bike is perfectly vertical and that the front and rear axles are level.

Equipment Baseline

Measure the bike.

Measure and mark cleat placement.

On the Bike Assessment

Have the cyclist warm up for as long as is practical including several all out efforts.

Instruct the cyclist to switch into their big ring and find a gear that would match a long time trial effort or hard tempo pace. Instruct the cyclist to no longer shift the rear cogs. Instead they will switch into the little ring for an easier “spin” resistance.

For dynamic assessments give the cyclist the instruction “Go ahead and spin. When you are ready shift to your big ring and settle into a steady pace.” Make your assessment after the cyclist has settled into their steady pace.

For static assessments coach the cyclist on gliding to a stop from the light spin at the BDC and cranks horizontal position and holding in the same position as they were when pedaling.

All measurements should be relative to anatomical position, ie the cyclist standing comfortably with arms at the side, palms facing forward.

Static vs Dynamic

When available dynamic measurements (from video while pedaling) are preferred over static measurements (cyclist at a stop). One study comparing static versus dynamic at different efforts the following differences (Peveler, 2011):

• Knee Angle: Static goniometer 25 deg, Camera static 28 deg, Camera dynamic low effort 35 deg, Camera dynamic max effort 33 deg.
• Foot Angle Plantar flexion from horizontal: Camera static 15 deg, Camera dynamic low effort 27, Camera dynamic max effort 22 deg.

The implication of this data is that the target range should be shifted by 5 – 10 degrees for knee angle and 7 – 12 degrees for ankle angle when using static measurements.

Similarly it is preferred to perform a bike fit on a very well warmed up if not fatigued cyclist, at moderate to high intensities, as cyclists demonstrate less plantar flexion at the ankle and less knee flexion as they increase intensity or fatigue.

Sagittal Plane Assessment  (side view)

Ankle Dorsiflexion at BDC

Measure ankle dorsiflexion at the bottom of the pedal stroke (neutral is zero, toe down is negative toe up is positive)

Metatarsals Over Spindle at Horizontal

With the cranks horizontal and the measured foot forward, assess the location of the 1st and 5th metatarsals relative to the spindle.

Knee Angle BDC

Measure the knee angle at the bottom of the pedal stroke.

Knee Over Spindle at Horizontal

In the cranks horizontal position measure the knee over spindle of the forward foot using your plumb bob.

Sacral Tilt

Measure the forward tilt of the sacrum from vertical.

Trunk Tilt

Measure the forward tilt of the trunk from vertical.

Neck Extension

Check to make sure the neck is not fully extended when the cyclist is looking up the road.

Shoulder Forward Flexion

Measure the forward flexion of the upper arm relative trunk (a superman position would be 180 degrees forward flexed)

Elbow Flexion

Measure the flexion at the elbow.

Wrist Extension

Measure the extension at the wrist.

Frontal Plane Assessment (front/back view)

Foot Rotation

Note the internal/external rotation of the foot.

Float

Check the float at BDC and cranks horizontal.

Stance Width

Note the stance width of the feet relative to the hips and thighs.

Ankle AD/ABduction

Note the ankle AD/ABduction during the power phase

Shank AD/ABduction

Note the Shank AD/ABduction during the power phase

Frontal Plane Knee Motion

Note the medial lateral Range of Motion of the knee during the pedal stroke.

Look for any jerky movements at the top and bottom of the pedal stroke.

Hand Position Width

Note the position of the hands relative to the shoulders.
Fit Adjustments

Fix anything that is way off. There is no point in spending time on small adjustments on any are if there is a large adjustment to be made elsewhere as the large adjustment is likely to shift kinematics throughout the rest of the fit;

Always ask the cyclist how they feel after each adjustment. If they feel worse after any adjustment reassess the fit as a whole.

Cleat Fore-Aft

Adjust the cleats so that the spindle splits the difference between the 1st and 5th metatarsals.

Cleat Center of Rotation

Adjust the cleat medial lateral position to bring the center of rotation under the middle of the ball of the foot.

Cleat Rotation/Float

Adjust the float and or rotation of the cleats to allow some internal and external float throughout the pedal stroke. Take out excess float.

Varus/Valgus Wedges

Consider wedges to bring the ankle AD/ABduction to neutral.

Stance Width

If possible, adjust the width of the pedals to bring the feet in line with the hips and knees.

Seat Height

Adjust the seat height for a target knee angle of 25-30 degrees with the ankle neutral plus minus 5 degrees.

Saddle Fore-Aft

With the cranks horizontal adjust saddle fore aft until the knee (tibial tubercle) is 0 – 1 cm behind the spindle.

Saddle Tilt

Set the saddle level.

Handlebar Drop and Reach

Adjust the drop and reach until the trunk is forward flexed 40 – 45 degs, with the shoulders at 90 degrees, and a 15 degree bend at the elbows.

Handlebar Rotation

Adjust handle bar tilt and hoods to bring the wrists to a neutral position.

Handlebar Drop

For sprinters consider and deep drop bar. For everyone else chose a drop shallow enough that they can achieve a 90 degree bend at the elbow in the drops.

Handlebar Width

Choose handlebar width to bring the hands in line with the shoulders.

Reassess the Entire Fit

Always check both sides. And recheck all measurements when finished. Make sure you accommodated for any issues noted in the physical assessment.

Problem Fits

Asymmetries: In general split the difference between the two sides.

Lower leg length difference: You may try shimming up to half of the difference in leg length.

Knee ankle discoordination: Recheck cleat position and knee over BDC. Excess plantar flexion maybe a cleat too far back or seat too far forward. Excess dorsiflexion may be a cleat too far forward or a seat too far back. Note that this is purely theoretical so assess the effect with some skepticism. If discoordination persists split the difference on the seat height.

Overlap between the elbows and knees: Instruct the cyclist on tilting the pelvis forward to lengthen the back. A high bar position may be needed to keep the hip angle open as the pelvis rolls forward. A wider bar can be a temporary fix.

Problem movements at the top and bottom of the pedal stroke: Consider shorter cranks

Excess medial lateral knee motion: Advise the cyclist that they may have poor neuromuscular control at the hip.

Hip drop or pulling to one side: Reassess for asymmetries, if none are found advise the cyclist that it may be a neuromuscular control issue.

Saddle Discomfort: If a trial of several saddles fails, reassess the fit, asymmetries, and neuromuscular control.

Reassessment and Documentation of the Standard Fit

On the Bike Assessment

Sagittal Plane Assessment  (side view)

Ankle Dorsiflexion at BDC

Measure ankle dorsiflexion at the bottom of the pedal stroke (neutral is zero, toe down is negative toe up is positive)

Metatarsals Over Spindle at Horizontal

With the cranks horizontal and the measured foot forward, assess the location of the 1st and 5th metatarsals relative to the spindle.

Knee Angle BDC

Measure the knee angle at the bottom of the pedal stroke.

Knee Over Spindle at Horizontal

In the cranks horizontal position measure the knee over spindle of the forward foot using your plumb bob.

Sacral Tilt

Measure the forward tilt of the sacrum from vertical.

Trunk Tilt

Measure the forward tilt of the trunk from vertical.

Neck Extension

Check to make sure the neck is not fully extended when the cyclist is looking up the road.

Shoulder Forward Flexion

Measure the forward flexion of the upper arm relative trunk (a superman position would be 180 degrees forward flexed)

Elbow Flexion

Measure the flexion at the elbow.

Wrist Extension

Measure the extension at the wrist.

Frontal Plane Assessment (front/back view)

Foot Rotation

Note the internal/external rotation of the foot.

Float

Check the float at BDC and cranks horizontal.

Stance Width

Note the stance width of the feet relative to the hips and thighs.

Ankle AD/ABduction

Note the ankle AD/ABduction during the power phase

Shank AD/ABduction

Note the Shank AD/ABduction during the power phase

Frontal Plane Knee Motion

Note the medial lateral Range of Motion of the knee during the pedal stroke.

Look for any jerky movements at the top and bottom of the pedal stroke.

Hand Position Width

Note the position of the hands relative to the shoulders.

Equipment Standard Fit

Measure the bike.

Measure and mark cleat placement.

Measure the bike.

Advise the cyclist that some time may be necessary to adjust to changes but that they should stop riding and return for any pain.

Schedule a Follow Up

No fit is complete until the cyclist has successfully accommodated to the new position without any pain or significant discomfort.
Works Cited

Asplund C, St Pierre P. Knee pain and bicycling: fitting concepts for clinicians. Phys Sportsmed. 2004 Apr;32(4):23-30.

Bailey MP, Maillardet FJ, Messenger N. Kinematics of cycling in relation to anterior knee pain and patellar tendinitis. J Sports Sci. 2003 Aug;21(8):649-57.

Bini R, Hume PA, Croft JL. Effects of bicycle saddle height on knee injury risk and cycling performance. Sports Med. 2011 Jun 1;41(6):463-76

Boyd TF, Neptune RR, Hull ML. Pedal and knee loads using a multi-degree-of-freedom pedal platform in cycling. J Biomech. 1997 May;30(5):505-11

Cavanagh PR, Sanderson DJ, The Biomechanics of Cycling: Studies of the Pedaling Mechanics of Elite Pursuit Riders.In: E.R. Burke and M.M. Newsom, Editors, Medical and Scientific Aspects of Cycling, Human Kinetics Books, Champaign, IL (1988), pp. 3–16.

Dannenberg AL, Needle S, Mullady D, Kolodner KB. Predictors of injury among 1638 riders in a recreational long-distance bicycle tour: Cycle Across Maryland. Am J Sports Med. 1996 Nov-Dec;24(6):747-53.

Davis RR, Hull ML. Measurement of pedal loading in bicycling: II. Analysis and results. J Biomech. 1981;14(12):857-72.

Dingwell JB, Joubert JE, Diefenthaeler F, Trinity JD. Changes in muscle activity and kinematics of highly trained cyclists during fatigue. IEEE Trans Biomed Eng. 2008 Nov;55(11):2666-74.

Ericson MO, Nisell R. Patellofemoral joint forces during ergometric cycling. Phys Ther. 1987 Sep;67(9):1365-9.

Ericson MO, Nisell R. Tibiofemoral joint forces during ergometer cycling. Am J Sports Med. 1986 Jul-Aug;14(4):285-90.

Ericson, M.O., Nisell, R. and Ekholm, J., 1984. Varus and valgus loads on the knee joint during ergometer cycling. Scandinavian Journal of Sports Sciences 6, pp. 39–45.

Francis PR, Pathomechanics of the lower extremity in cycling. In: E.R. Burke and M.M. Newsom, Editors, Medical and Scientific Aspects of Cycling, Human Kinetics Books, Champaign, IL (1988), pp. 3–16.

Gregersen CS, Hull ML, Hakansson NA. How changing the inversion/eversion foot angle affects the nondriving intersegmental knee moments and the relative activation of the vastii muscles in cycling. J Biomech Eng. 2006 Jun;128(3):391-8.

Hamley EJ, Thomas V. Physiological and postural factors in the calibration of the bicycle ergometer. J Physiol. 1967 Jul;191(2):55P-56P.

Hannaford DR, Moran GT, Hlavac HF. Video analysis and treatment of overuse knee injury in cycling: a limited clinical study. Clin Podiatr Med Surg. 1986 Oct;3(4):671-8.

Holmes JC, Pruitt AL, Whalen NJ. Lower extremity overuse in bicycling. Clin Sports Med. 1994 Jan;13(1):187-205.

Mills OS, Hull ML. Rotational flexibility of the human knee due to varus/valgus and axial moments in vivo. J Biomech. 1991;24(8):673-90.

Nordeen-Snyder KS. The effect of bicycle seat height variation upon oxygen consumption and lower limb kinematics. Med Sci Sports. 1977 Summer;9(2):113-7.

O’Brien T. Lower extremity cycling biomechanics. A review and theoretical discussion. J Am Podiatr Med Assoc. 1991 Nov;81(11):585-92. Review.

Peveler WW, Shew B, Johnson S, Palmer TG. A Kinematic Comparison of Alterations to Knee and Ankle Angles from Resting Measures to Active Pedaling During a Graded Exercise Protocol. J Strength Cond Res. 2011 Dec 8.

Peveler WW, Green JM. Effects of saddle height on economy and anaerobic power in well-trained cyclists. J Strength Cond Res. 2011 Mar;25(3):629-33.

Peveler WW. Effects of saddle height on economy in cycling. J Strength Cond Res. 2008 Jul;22(4):1355-9.

Peveler WW, Pounders JD, Bishop PA. Effects of saddle height on anaerobic power production in cycling. J Strength Cond Res. 2007 Nov;21(4):1023-7.

Price D, Donne B. Effect of variation in seat tube angle at different seat heights on submaximal cycling performance in man. J Sports Sci. 1997 Aug;15(4):395-402.

Ruby P, Hull ML, Kirby KA, Jenkins DW. The effect of lower-limb anatomy on knee loads during seated cycling. J Biomech. 1992 Oct;25(10):1195-207.

Sanderson DJ, Black AH, Montgomery J. The effect of varus and valgus wedges on coronal plane knee motion during steady-rate cycling. Clin J Sports Med. 1994;4:(2):120-4.

Sanderson DJ, Amoroso AT. The influence of seat height on the mechanical function of the triceps surae muscles during steady-rate cycling. J Electromyogr Kinesiol. 2009 Dec;19(6):e465-71.

Sanner WH, O’Halloran WD. The biomechanics, etiology, and treatment of cycling injuries. J Am Podiatr Med Assoc. 2000 Jul-Aug;90(7):354-76.

Sauer JL, Potter JJ, Weisshaar CL, Ploeg HL, Thelen DG. Biodynamics. Influence of gender, power, and hand position on pelvic motion during seated cycling. Med Sci Sports Exerc. 2007 Dec;39(12):2204-11.

Shennum PL, deVries HA. The effect of saddle height on oxygen consumption during bicycle ergometer work. Med Sci Sports. 1976 Summer;8(2):119-21.

SICI Personalized Class Companion Manual 2007

Swift P, Schoenfeldt V, The Bicycle Fitting System Manual, www.bikefit.com

Umberger BR, Martin PE, Testing the planar assumption during ergometer cycling. Journal of Applied Biomechanics, 2001, 17, 55-62.

Wanich T, Hodgkins C, Columbier JA, Muraski E, Kennedy JG. Cycling injuries of the lower extremity. J Am Acad Orthop Surg. 2007 Dec;15(12):748-56.

Weiss BD. Nontraumatic injuries in amateur long distance bicyclists. Am J Sports Med. 1985 May-Jun;13(3):187-92.

Wilber CA, Holland GJ, Madison RE, Loy SF. An epidemiological analysis of overuse injuries among recreational cyclists. Int J Sports Med. 1995 Apr;16(3):201-6.

Weiss BD. Nontraumatic injuries in amateur long distance bicyclists. Am J Sports Med. 1985 May-Jun;13(3):187-92.

Wolchok JC, Hull ML, Howell SM. The effect of intersegmental knee moments on patellofemoral contact mechanics in cycling. J Biomech. 1998 Aug;31(8):677-83.

The post veloclinic Bike Fit Procedure appeared first on veloclinic.

Description of the derivation of the veloclinic Mean Maximal Power Duration Models.

Click here for the pdf: Power Model Derivation

Also see the Video Explainers

The post Derivation of the Candidate Mean Maximal Power Duration Models appeared first on veloclinic.

DerivationPart 1 The Equations

Part 2 Example Fits

Also see the Derivation.

Below are the equations since they are a little hard to see in the video.

#2 secret exp top linear bottom

fo <- y~ w1/x*(1-exp(-x/tau1))*((1-exp(-x/10))^alpha) + pow2/(1+x/tau2)
rpowb <- nls2(fo, start=list(w1 = rw1, tau1 = rtau1b, pow2 = rpow2, tau2 = rtau2, alpha = .1 ), data = dspow)
presidb<-resid(rpowb)/y*100

plot(log=”x”, x,y)
lines(x, predict(rpowb), col=”purple”)

#2 secret exp top regen fixed bottom
fo <- y~ w1/x*(1-exp(-x/tau1))*((1-exp(-x/10))^alpha) + pow2/(1+x/5400)^(1/beta)
rpowc <- nls2(fo, start=list(w1 = rw1, tau1 = rtau1b, pow2 = rpow2, alpha = .1, beta = 1), data = dspow)
presidc<-resid(rpowc)/y*100

plot(log=”x”, x,y)
lines(x, predict(rpowc), col=”red”)
#2 secret flock exp top and  bottom
fo <- y~ w1/x*(1-exp(-x/tau1))*((1-exp(-x/10))^alpha) + pow2*tau2/x*(1-exp(-x/tau2))
rpowe <- nls2(fo, start=list(w1 = rw1, tau1 = rtau1, pow2 = rpow2, tau2 = 15000, alpha = .1), data = dspow)
preside<-resid(rpowe)/y*100

The post Video explainers of the veloclinic Mean Maximal Power Duration Models appeared first on veloclinic.

The upcoming 2014 Giro d’Italia features 6 that are likely to play a role in shaping the final podium. As with previous Grand Tours, @ammattipyoraily is my go to source for the most consistent data.

The process is simple:
1. Measure the climbs.
2. Time the finishing climbs.
3. Estimate normalized W/kg power outputs using the Martin Model.
4. Make sense of the performances with a pVAM and Mean Maximal Power curve.

First up the climbs:

Stage 14 features the Oropa. It is the shortest climb at 6.7 km and the lowest in altitude topping out a little over 1,142 meters with a mid gradient of 8%.

Stage 15 brings the Mentecampione. It is also fairly steep at 7.8%, but nearly 3 times longer at 18.7 km. Likewise it tops out in thinner air at 1,665 meters.

Stage 16 kicks up the length and altitude with the Martello. The gradient is the shallowest of the major climbs at 6.4% but also the longest at 21.1 km and topping out at 2059 meters.

Stage 18 and 19 bring a pair of mid grade/distance climbs. The Panarota is first at 8%, 15.9 km at topping out at 1760 meters, and the Grappa measures in at 8%, 19.3 km, and tops out at 1712 meters.

Finally, Stage 20 is the Zonoclan. At 11.9% is is by far the steepest pitch. After 10.1 km riders will top out at 1730 meters of altitude.

The simplest analysis of course is simply comparing times to the historical records. Going a step further, results across different climbs and eras can be compared using pVAM which allows performance predictions based on the characteristics of the climb.

For a pVAM analysis, I use equations derived by Scott Richards, who pioneered the method in A Different Approach to Comparing Climbing Performances on Cyclismas.com. The equations uses data from the supposedly clean years of 2008-2013 (pVAM), and at my request, data from the doping years of 2002-2007 (DpVAM), to predict the climbing speeds (VAM) expected from these baselines.

Above, are the times predicted by the pVAM equation.

Plugging the times into the Martin model gives the power output in normalized W/kg expected for each climb. Not surprisingly, the highest numbers, 6.3 W/kg, will likely come on the short lower altitude Oropa. The lowest numbers, 5.7 W/kg, will likely come from the effort on the long, high altitude Martello. The mid grade length climbs should come in around 5.8 W/kg while the steepness of the Zonoclan should bring out maximal sustainable efforts just above 6 W/kg.

Lastly, in case its needed, the DpVAM numbers. The red bars indicate the speeds that would be expected from a doped peleton. As Ross Tucker likes to say, it is probably best to take a step back when looking at the DpVAM. The noise inherent in these methods means that any one performance may not always be interperatable on its own.

The post 2014 Giro d’Italia Climb Preview appeared first on veloclinic.

This post introduces a new way of plotting performance to better visualize W’ and Critical Power. I haven’t seen anyone plot performance this way, so I’m going to take the liberty to name it the Veloclinic plot. In descriptive terms, it would be accurate to call it a W’ envelope plot or a Critical Power Subtraction plot. An example of the new plot is below:

Note that the y axis has now become Joules or capacity, and the x axis has become Watts or power. The plotted line is the capacity that can be generated from W’ at any given power, ie W’ envelope plot. The W’ capacity is isolated by calculating the capacity that would be generated  CP and then subtracting it out, ie CP subtraction plot.

The motivation to develop the VC plot came out of frustration with the visually un-intuitive traditional Power Duration plot.

In the traditional PD plot. The y axis is power and the x axis is duration. The x axis is typically converted to a log scale so that details of the curve can be better brought out. Even on a semi-log plot however, the only performance measure that is visually clear is Pmax.

In contrast, on the VC plot, CP is visually obvious as the first x intercept. Just as clear, W’ is the y asymptote and Pmax the x intercept or asymptote.

Of course, CP can be plotted on the traditional PD curve. The impression though is of a straight line randomly intersecting a slant. There is no obvious visual feedback whether the CP estimate is correct or a curve fitting artifact.  Even more problematic is illustrating W’ as neither scale is in capacity units.

The only way to represent W’ is to plot it with CP in the critical power model form. This representation at times is reasonably effective. However, I am often left wondering if the CP model is correctly centered over the appropriate range of the PD curve or whether it should be shifted a bit to the left or right.

An advantage of the VC plot is that it is obvious if the CP estimate is wrong. In the plot above, I lowered CP by a small but meaningful  5 percent. Underestimating CP skews the curve left and creates an odd peak above a slanted asymptote.

The opposite is true of Inflating CP by the same 5%. This changes creates a notable rightward shift of the asymptotic section producing the windblown appearance above.

When the CP estimate is correct, the VC plot simply looks right.

The fit of hyperbolic shapes on the other hand can be difficult to visually assess. The same small but meaningful differences in parameter estimates make no dramatic changes in the shape. The result is the appearance of a reasonable fit simply shifting to a slightly different region of the curve:

The VC plot can be further refined by using the Ward-Smith equation to generate a smooth x axis:

Note that the flat region of the y asymptote becomes even more distinct, and the goodness of fit of the WS model is visually confirmed by good alignment of the model and subject data.

An important detail that now also emerges is a concept that I am calling the Super Critical Power.

Going from left to right, the VC plot starts at 0 at CP and jumps up to the y asymptote at W’. The flat region at the asymptote represents the range of power where W’ can be fully or nearly fully developed. In this region, changes in power result in no significant loss in W’ availability. Following this region, the curve falls exponentially away from the asymptote. The implication is of an upper threshold were subsequent increases in power results in an exponential cost in terms of a loss of total W’. I am terming this threshold, the Super Critical Power. As a working definition, SCP is a power threshold above which less that 95% of W’ can be generated before failure occurs.

Super critical power has important implications in terms of Skiba’s intermittent W’ balance model as well as underlying physiological implications.

Once possible way that VC plot can be used to integrate some of the more interesting physiology publications is to think of the dominant limiters of each zone. I will start with the W’ zone. In my mind this zone may be best explained by complete peripheral fatigue of fast and intermediate twitch motor units. This zone may result from the sequential fatigue of these motor units in a power range that allows for complete depletion without other limiters becoming dominant prior to failure. Evidence for this mechanism comes from studies that demonstrate the development of the the slow component above CP and attainment of VO2max preceding effort failure. Similarly, once failure occurs in this zone, some studies show that outside stimulation of the muscle can not elicit a response greater than voluntary contraction. In order for this to occur, it follows that all easily fatigueable intermediate and fast twitch motor units must have been fully recruited and depleted. In contrast, fatigue generated by efforts above SCP have been shown to not necessarily produce complete peripheral fatigue. When fatigue is produced near maximal efforts, external stimulation can potentially generate muscle contraction greater than voluntary contraction. These findings suggest that a central fatigue occurs resulting in sub-maximal motor unit recruitment. Similarly, I anticipate that SCP should correspond to a threshold power above which failure occurs before complete development of the slow component and therefore VO2max can not be reached. Lastly, below CP studies show that the slow component does not fully develop and VO2max is not reached. VO2 kinetics are paralleled by findings of variability in sequential motor unit recruitment and lack of complete recruitment. Failure in this zone is potentially a multi-factorial systemic mechanism or may dominated by a specific factor during severe conditions such as fuel depletion, temperature dis-regulation, or as a protective central feedback mechanism.

The post Veloclinic Plot (W’ envelope plot) appeared first on veloclinic.

We are currently recruiting participants for a performance modelling study.

Requirements for participation are that you race and train with a power meter.

Please contact Dr Jonathan Baker; jrb07@aber.ac.uk for more information.

The post Cyclists Needed For Performance Modelling Study appeared first on veloclinic.

Previous attempts have have focused on either trying to normalize power to Critical Power/Functional Threshold Power:

nP = %CP^4

or

vessel approaches of trying to dynamically track W’/FRC balance

W’b = W’ – (P-CP)*t + (CP-P)*t*(f[reconstitution]).

The normalized power approach will typically work reasonably well for power outputs close to CP/FTP. Outside a fairly narrow range however physiological variability is going to make it fairly useless for many.

The vessel approaches will potentially work better as demonstrated by Skiba. However, any dead reckoning is going to prone to cumulative drift and drift issues will only be compounded by trying to expand the model to include limiters that come into to play above Super Critical Power and below Critical Power.

My thought is to consider a statistical approach were pacing/stress follow a simple rule of local and global intermittency.

The basic premise is that the best mean maximal power for any given duration can be achieved by a nearly constant effort when starting from a primed state. Any deviation from this mean will result in a penalty in terms of lost potential work.

The deviation from the mean, or intermittency, can be quantified as the relative standard error. The penalty associated with the intemittency can be normalized to intermittency capacity at any given power.

For example between CP and SCP, since work is debited against a fully available W’ regardless of the rate, then within in this range there is little penalty for intermittency. Note that while the intermittency envelope is going to have some structural overlap with the W’ envelope, it doesn’t have the weirdness of suddenly disappearing below CP.

Plotting the results of power versus intermittency versus time gives us a three dimensional illustration of performance capability.

 Hmm, it looks like a Power Duration Curve but now all grown up.

The post Rethinking Intermittent Modelling appeared first on veloclinic.

For longer than I’d like to admit, I have been advising on Clean Protocol based out of Australia. I say longer than I’d like to admit because it has taken so damn long to get to the point were meaningful forward progress has finally started. In anticipation of finalizing some key pieces of technology I wanted to reintroduce the reward side paradigm into the sports doping/ethics discussion. The fundamental premise is that statistically it is possible to identify a far larger subgroup of athletes who are very likely clean than very likely doped. More importantly, as I will sketch out below reward side anti-doping efforts will actually be complementary rather than competing with the current punitive side anti-doping efforts.

In the figure above, the larger blue curve represents the subgroup of about 70% of athletes who are truly clean. The red curve represents the subgroup of about 30% of athletes who are truly doping. The x axis represent the range of thresholds that can be set as a cut off between a positive and negative test. The curves are bell-shaped and overlapping to show the reality that there will never be a perfect test. Instead, any given threshold will always be a trade off between catching cheats and wrongly accusing clean riders.

The current paradigm in anti-doping is a punish and deter model. Penalties are so harsh that testing thresholds have to be set very conservatively to minimize the chance of a clean rider suffering such a harsh penalty. Dr Larry Bowers of USADA (AMSSM 2013 presentation and personal communication) argues that while only a fraction of doped athletes are actually caught the system is still a strong deterent to high level doping.

Unfortunately, under this system the majority of clean and doped athletes both end up with the ambiguous label of “not positive.”

Without any modification to the current testing whatsoever, an alternative approach would be to take a reward side anti-doping model. Since the goal here is to reward clean athletes with an endorsement and no harsh penalties are involved, the thresholds can be shifted so that the vast majority of doped athletes are excluded. Under this approach the majority of clean athletes can be given the far less ambiguous endorsement of very likely clean. Such a designation would be far more likely to be embraced by fans. It would also provide sponsors with a far more secure investment. The potential issue of course, is if nobody is banned then individuals will be given the option to trade credibility for podiums.

Fortunately, the punitive/deterrent and reward based models are not mutually exclusive. Instead, they are obviously quite complementary. By including both top down and bottom thresholds the greatest advantage can be had from the imperfect tests. As illustrated here, the very likely clean athletes can benefit from reward based endorsement, meaningful penalties can be kept in place, and the number of clean riders left in the “not positive” grey zone can be minimized.

The post The Imminent Arrival of Reward Side Anti-Doping appeared first on veloclinic.

One of the fun parts about modelling climbing performances in Grand Tours is the ability to make historical and virtual head to head comparisons. In the figure above the grey dots are the average performance of the top 3 finishers from each Grand Tour for each finishing climb from 2008 – 2013. Overlayed on this historical context, is a head to head comparison of 2013 Froome in sky blue and 2014 Nibali orange-yellow for each of their Tour de France wins. Although, we never got to actually see them battle it out directly, the data suggests it would have been a hell of a battle.

Something to keep in mind here though is that the estimated W/kg is normalized both to a standard 67kg 0.35 CdA rider as well as to a typical 1256 m Grand Tour mid climb altitude. By making this normalization a bit of trade off is made in terms of giving up some understanding of the individuals physiology and morphology for the sake of having a better understanding of who would finish better out on the road.

The post Comparing Performances Across Grand Tours appeared first on veloclinic.

One of the weaknesses of the current Athlete Biological Passport software seems to be an over-emphasis on outlier detection versus level change detection. To illustrate this point consider the Lance Armstrong case:

According to the UCI, Armstrongs blood data never triggered a flag on the software. The software screen was non-positive despite the obvious reticulocyte  suppression starting after the Giro and persisting through the end of the Tour de France. The suppression was later calculated to be completely statistically implausible as part of the expert evidence in the USADA decision.

The question of course is why did the software fail?

The answer seems to be in the interaction between level changes, outliers, and the variance (spread in the data). Outliers are basically one off values that are unusually far from the average. Level changes on the other hand, are an unusual number of consecutive values that are above or below the mean but not necessarily unusually far from the average for any given point. Since ouliers are one off points, they have little effect on the overall variance. Level changes on the other hand, given enough points in the cluster, will potentially increase the variance significantly.

What does that mean for the biopassport?

Consider this mock up example I made using some modified code from the MARSS package for the Nile river flow data. The data is useful as an example because it contains both outlier points as well as a level shift (there was a damn built just before 1900). After dividing the flow data by 1000, it looks a lot like a reticulocyte profile with an average just below 1. The code is useful because it uses a Kalman filter which is essentially a special class of Bayesian statistics and so we can get the basic gist of what the ABP software is doing. I further modified the MARSS example code by adding the 99.5% confidence intervals in the later figures as would be used by the ABP software

The figure above shows the fitting of a flat (top) and stochastic (bottom) model to the data. In both cases, the hidden state being modeled is the true reticulocyte count. In the flat model the true value is a flat line that falls on the average and any deviation away from this true state is model and observation error. In the stochastic model, the true value can move randomly in time and the observed value is this movement plus the model and observation error.

The implication for interpreting the observed values is illustrated next when the standard residuals are plotted with the 95% (black) and 99.5% (red) confidence intervals below.

Comparing the two models it can be seen that the level shift inflates the variance of the data when attempting to use a flat model. Essentially, since the hidden state can’t move then the error is by necessity larger to account for error. In the stochastic model, since the hidden state can move the error is much smaller. In terms of picking up outliers, notice that the stochastic model actually does better than the flat model.

So far so good, but the pink elephant (the level change) is still in the room. The solution of course is to take advantage of the fact that the hidden state moves in the stochastic model and run a test to detect level change.

And there it is, in the bottom panel, the level change just before 1900. So in the case of this mock up data it appears that a level change potentially could work in some scenarios where outlier detection would not. It also would potentially be more specific for modern doping which uses EPO micro and masking doses as well more frequent small volume withdrawals and transfusions.

The post Athlete Biological Passport For Level Change Detection appeared first on veloclinic.

Invitation to Clean Athletes from Clean Protocol.

The post Invitation to Clean Athletes competing in the 2014 Ironman World Championship appeared first on veloclinic.

Match fixing may not be the first topic that springs to mind when thinking about the problem of doping and what to do about it. But it is a useful one. It forces the discussion to step well outside of the usual box. The reason of course is that nobody expects match fixing to be detectable in blood or urine samples. Instead, any test for match fixing will have to look for markers of the memory of match fixing or the cognitive manipulation necessary to conceal this information. This cognitive manipulation, commonly known as deception, is thought to be the most detectable component of these illegal/unacceptable behaviors. Without other alternatives, it is an obvious choice to develop and test for match fixing based on the detection of deception.

Early work in deception testing focused on monitoring simple physiological parameters that change in response to deception. The most well known test of course is the polygraph. Like the bio-passport, the polygraph monitored simple physiological markers that do respond to the target of the testing. Also like the bio-passport the polygraph parameters respond to various other factors as well. The end result is a test that often “works,” but is a far from ideal balance between sensitivity and specificity. Further, because of the simplicity and non-specific nature of the parameters, both tests are generally regarded as beatable with simple interventions that manipulate the markers.

Research in deception testing has moved forward with the more recent development of cognition-based testing. The basic premise of this class of tests is that the process of telling a lie is more mentally complex or challenging than telling the truth. One of the new methods from this class that has done particularly well in mock-crime experiments is Occular-motor Deception Testing. In OMDT a special optical scanner is used to track differences in pupil diameter, fixations, and reading/rereading time in response to true false questions. The OMDT technology is now available commercially as EyeDetect through Converus. Converus puts the accuracy of their product at 85% from controlled mock-crime experiments. Converus expects better results as the consequences increase in real world application.

For purposes of Clean Protocol, deception testing such as OMDT is a particularly attractive option for several reasons. The first is the flexibility. Since the target of the test is deception, the technology could potentially be adapted to address nearly any banned method or behavior. This flexibility opens up the possibility of going after previously un-testable issues such as match fixing or the facilitation of doping. Possibly even more important however is the broad detection window. Instead of the 6-8 hour glow times common with EPO micro-dosing, the window does not close for deception until the truth comes out. As a result, OMDT would potentially be the first anti-banned methods test that could be expected to be effective on an annual or pre-event only testing schedule. Lastly, the EyeDetect system produced by Converus is commercially available now.

Like any test, deception such as OMDT is not perfect and has its limitations. OMDT falls into the expanding omics branch of indirect detection which includes the bio-passport. This class of tests will always bring out creative explanations of how the test got it wrong. Similarly, these tests will never have a perfect ground truth to validate against. Elite athletes in competition can not be put under perfectly controlled conditions, nor can real-world conditions be perfectly simulated in the lab setting. Instead, these tests will always have to generalize out from laboratory or un-controlled field conditions. Fortunately the bio-passport cases have gone a long way to demonstrate that omics methods will stand up to both legal and public perception challenges despite these limitations.

In summary, match fixing is a problem in sport that forces a rethink of the traditional approaches to enforcing rules and ethical adherence. Using match fixing as an example, it is easier to see why OMDT is a leading candidate test for a project like Clean Protocol. OMDT could open up the flexibility to detect any banned method or behavior,  create a theoretically unlimited detection window, and is commercial availability through Converus. The limitations of indirect testing  including the need to generalize out from a related ground truth are acknowledged and accepted following the precedent set by the bio-passport. The ultimate success of such technologies will be determined by their results in upcoming real-world applications such as anti-doping.

The post Developing a test for match fixing, an illustrative example appeared first on veloclinic.

A repeating theme heard in performance analysis discussions is that “Performance doesn’t prove doping.” Of course this is true because proof, as an absolute, doesn’t really exist. Instead, most real-world evidence driven judgments are based on probability and not “proof.” From this perspective, this post will illustrate the concept of using a mathematical model to estimate the probability of doping as a function of performance.

The first step is to create a probability distribution for clean riders at the Tour de France:

To make the distribution some assumptions need to be made based on available information. In this case, the assumption is that on a prototypical Tour de France climb a couple of clean riders should be able to sustain about 6 W/kg. A second assumption is that riders need to be able to sustain at least 5 W/kg to make the team and survive the race. Choosing a Gaussian model (a bell shaped curve) a mean of 5.5 W/kg and standard deviation of 0.25 W/kg will generate a distribution that meets these assumptions/observations. (Don’t worry if you don’t like my assumptions, at the end is a link to a Google spreadsheet that you can manipulate them for yourself)

Next, the overall prevalence of doping can be estimated from the published literature:  http://www.ncbi.nlm.nih.gov/pubmed/25169441

In this review, the prevalence is estimated from 14-39%. The mean would be 31%, but for the sake of giving cycling the benefit of the doubt I used 25%. Using, the distribution function above and number of clean riders we can generate the distribution of clean riders as a function of power.

As you can see this model predicts 1-2 riders above 6 W/kg and 1-2 riders below 5 W/kg.

Now the performance effect of doping needs to be considered. Ashenden has previously shown that EPO micro-dosing producing an increase of 10% in hemoglobin mass (the equivalent of 2 blood bags) is not flagged by the bio passport.  http://www.ncbi.nlm.nih.gov/pubmed/21336951

Since O2 metabolism is the primary determinant of endurance performance it can be extrapolated from this study that a 10% increase in performance from O2 vector doping alone may still be possible. However to stay reasonably conservative I used a 5% benefit from doping.

So now a second probability function is made with the mean in creased by 5%. The standard deviation is unchanged. From the probability distribution and the estimated prevalence of 25% doping we can now generate the distribution of doped riders as a function of power and overlay that with the clean rider distribution.

This figure should start giving you some clue as to why it is misleading to say that “performance doesn’t prove” anything.

Finally, from the distribution models we can calculate the probability that a level of performance is likely to be produced by a clean versus doped rider.

As you can see based on this model and the assumptions above, the probability of doping at the 6 W/kg performance level on a prototypical climb is about 60%. The probability of doping increases to 80% at 6.2 W/kg, and approaches 100% at 7 W/kg.

This model/example is meant to be illustrative of the concept. With better input and more sophisticated modelling, performance could be realistically used as an indirect measure of doping probability. Triangulating across multiple indirect measures including biological, performance, and psychometric measures would likely improve our understanding of the true doping burden on sport.

If you would like to experiment with the model it is available at the link below. The fields to manipulate are highlighted in yellow.

https://docs.google.com/spreadsheets/d/1-YgGROA3ZPtZ_kycr4UIwdqOL1S85NRTpb3zDTQ3Tuw/edit?usp=sharing

The post Estimating the Probability of Doping as a Function of Power appeared first on veloclinic.

The Golden Cheetah crew has now integrated R into Golden Cheetah in the latest development build. This update instantly gave Golden Cheetah statistical and modelling super-powers to the point where I gave it the nickname GoldenR Cheetah. As an example, this script uses the change point package to auto discover intervals within a power file.

As you can see, the red horizontal lines highlight the intervals discovered within the ride. You can adjust a penalty for breaking up intervals to fine tune what you are looking for.

In a follow up post I will show how to use your mean maximal power curve to then normalize your intervals as a percentage of the max duration that you can sustain that power fresh.

The script is as follows:

## R script will run on selection.

##

## GC.activity()

## GC.metrics(all=FALSE)

##

## Get the current ride or metrics

##

## This script requires the change point package

require(changepoint)

## Get your activity

act <- GC.activity()

## Extract your ride power data in to a time series

y <- act$power

myts <- ts(y, start=c(1), end=c(length(y)), frequency= 1)

## Discover your intervals, change the “pen.value” to adjust the penalty separating intervals

disc = cpt.mean(myts, penalty=’Manual’,pen.value=’60000*log(n)’, method=’PELT’);

## Do a happy dance that you can now do this right in GC!

plot(disc, xlab=”seconds”, ylab=”watts”, main=”Interval Discovery”, ylim=c(0, 1500));

The post GoldenR Cheetah Script For Interval Discovery appeared first on veloclinic.

In the white is Skiba’s W’balance that you all know and love. I’ve combined this now with the interval discovery algorithm so that we can now visualize the change in W’bal by interval in green.

## R script will run on selection.

##

## GC.activity()

## GC.metrics(all=FALSE)

##

## Get the current ride or metrics

##

#start here

require(changepoint)

#read power file from csv. file should be one column of power with header Power

act <- GC.activity()

y <- act$power

## Set Your CP W’ K and Pen

## K is a gain term for recovery introduced by Mike Patton

## Pen is the a term that penalizes for finding smaller intervals with the changepoint algorythm

CP <- 250

W <- 20000

k <- 1

Pen <- 7000

## Calculate W’bal with differential model

Wbal <- vector()

Wexp <- 0

u = 0

n = 1

while (n < length(y)){

if (y[n] > CP){

Wexp <- Wexp + y[n]- CP

Wbal[n] <- W – Wexp

n <- n + 1

u <- 0

}

else{

Wexp <- Wexp – (Wexp/W)*k*(CP-y[n])

Wbal[n] <- W – Wexp

n <- n + 1

u <- u + 1

print(Wbal[n])

}

}

## Make a time series with the power data

myts <- ts(y, start=c(1), end=c(length(y)), frequency= 1)

## Run the changepoint PELT algo, adjust pen.value to optimize penalty

mvalue1 = cpt.mean(myts, penalty=’Manual’,pen.value=’Pen*log(n)’, method=’PELT’)

## Calculate the duration for each interval found

time <- cpts(mvalue1)

time <- append(time, length(y))

## Calculate the delta Wbal for each interval

## pWbal is the Wbal at the start of the interval

pWbal <- vector()

dWbal <- vector()

i <- 1

pWbal[1] <- Wbal[1]

dWbal[1] <- Wbal[time[1]] – Wbal[1]

while(i < length(time)){

pWbal[i+1] <- Wbal[time[i]]

dWbal[i+1] <- Wbal[time[i+1]] – Wbal[time[i]]

i <- i + 1

}

## Recreate the full set of data points

i <- 1

ii <- 1

tpWbal <- vector()

tdWbal <- vector()

while (i < length(y)){

if (i < time[ii]){

tpWbal[i] <- pWbal[ii] – W

tdWbal[i] <- dWbal[ii]

i <- i+1

}

else{

ii <- ii + 1

}

}

##Visualize

plot(tdWbal, ylim = c(-W, W), col=’green’ )

lines(tdWbal,col=’green’)

grid()

lines(Wbal, col=’grey’)

The post GoldenR Cheetah Script for Visualizing Delta W’balance By Interval appeared first on veloclinic.

Athlete doping rarely happens in isolation. So why do ADAs try to tackle the problem that way?

Consider this toy model of the social structure of a doping network. We may have individuals who may be Athletes, Trainers, Coaches, and Doctors. The individuals may instigate doping behavior, be reactive to doping behavior (dope if pressured, not dope if not pressured), be tolerant of doping, or not dope. They will also have varying numbers of connections to those who may be around them. In this example, it becomes visually obvious that targeting the instigators at the center of the doping networks is most likely to have the largest impact on the overall doping prevalence in the group.

Now consider who is actually in the WADA testing pool?

Athletes.

Consider who is missing from the testing pool?

Everyone else.

How does the doping network look from the perspective of current WADA testing?

Invisible.

In fairness to WADA, there is no illusion that majority of the dopers will be caught this way. Instead, the strategy is to take the dopers that are caught, and then punish them so harshly that it deters doping behavior within the group at large.

The issue that arises though is that severity of punishment must be supported by a proportional specificity of the doping test. In plain language, if you are going to end an athlete’s career, you better damn well be sure the test got it right. This system is inherently self-defeating cycle. The reason is that as you push the threshold toward better specificity you lose sensitivity. As you lose sensitivity you catch less dopers. As you catch less dopers you need to punish them even more severely to deter the group. As you punish them more severely, you better be even more damn well sure the test got it right. So you push the specificity even further and catch even less dopers and have to punish them even more severely (at some point I will code up a simple dynamical systems model to animate this example).

Does this deterrent model make any sense relative to how a doping network is most likely to work?

If we abandon the idea that testing must be biological, we can consider an alternative approach. In the case of occular motor deception detection http://converus.com/eyedetect-lie-detection/ , Converus offers a commercially available inexpensive product that is highly scalable with lab validation suggesting sensitivity and specificity over 80% https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3763937/ .

Now that testing is no longer limited by biological samples, how does that change the testing pool ?

Of course, the specificity here is not high enough to sanction any individual. But it is enough start to categories individuals into high or low credibility.

Next, we can take the now established Bayesian approach established by the bio-passport to use multiple indirect measures of individually low specificity to estimate the probability of doping to a high degree of specificity. However, unlike the current bio-passport which only considers longitudinal intra-individual data (results only for that athlete) we can expand the input data transversely across groups and testing methods.

Even if this system retained the high specificity harsh punishment approach, how likely would it be to fall into the self-defeating WADA negative feedback cycle?

I suggest that at least 2 main advantages would keep this approach from becoming just more of the same:

1. The amount of data going in to athlete’s passports would be an order of magnitude higher improving sensitivity at the same level of specificity.
2. Positive tests would more likely cluster within the center of the doping networks.

I suggest that the potential outcome would look far different to current testing.

Transverse Doping Probability Passport.

Current testing.

The post A Transverse Doping Probability Passport: A Conceptual Illustration appeared first on veloclinic.

Purpose: To demonstrate a Bayesian approach utilizing the athlete network to boost the positive predictive value of individual doping classifications.

Study Design and Methods:  Five data-sets of 10,000 individuals and their networks were simulated and statistically analyzed in R. Background prevalence (BP) of 10%, 20%, 30%, 40%, and 50% was chosen to cover the range of reported estimates (de Hon 2015). Individuals were assigned doping (D) or not doping (ND) classification randomly according to the BP. Each individual’s 10 member network was randomly assigned a co-prevalence (CP) from 0 to 1. Members were randomly assigned a D or ND classification according to a function of the individual’s classification, the CP, and the BP.

The standard test (ST) results of an emerging technology, ocular motor deception testing (Cook 2012),  were simulated by adding random error according to an 85% sensitivity and an 85% specificity. Positive ST results were classified as D. Negative ST results were classified as ND. The Bayesian approach utilized a 1:1 matching model relating network posterior probability (NPP) to individual prior probability. First the BP was recalculated from the ST results as a function of the number of positives (Pos) and the given sensitivity and sensitivity:

    GP = (Pos – 10,000*.15)/(10,000*.85 – 10000*.15).

Each NPP was then calculated as a function of the GP and the cumulative binomial probability (CBP) of k ND results within the network given the probability of an ND (PND) result:

    NPP = GP*(CBP(k, 10, PND)) + 1 – (CBP(k, 10, PND)); k < PND * 10,

    NPP = GP*(CBP(k, 10, PND)); k > PND * 10,

    NPP = GP; k = PND * 10.

Each individual’s posterior probability of doping given their ST result was calculated as a function of the test result, the NPP, and the given sensitivity and specificity of the ST:

    P(D | ST) = (NPP.85)/(NPP.85^Pos*.15^Neg + (1-NPP)*.85^Neg*.15^Pos)).

The cut-point for a D classification was set at .999 per standard WADA practice.

Differences between groups were tested by student’s t-test.

Results:

The boosted Bayesian classification improved test positive predictive values versus the standard test implementation (99.8 +/-  .2%  vs 66.1 +/- 19.3%, p = .02). The negative predictive values decreased non-significantly (73.5 +/- 15.4% vs 92.3 +/- 5.2%, p > .05). Co-prevalence was higher in the boosted Bayesian D classifications (77.6 +/- 2.8% vs 50 +/- .5%, p < .05).

The stacked bar chart of the standard implementation illustrates the usual worsening of the positive predictive value as the background prevalence decreases. Also, it is obvious that even at a background prevalence of 50%, a test with 85% specificity will still have unacceptably high numbers of false positives for sanctioning purposes. In the standard implementation this test would be good only from a surveillance perspective, or to guide targeted testing.

Compared to the above, this chart shows that the utilization of the doping co-prevalence within a network to boost the positive predictive value of an individual’s result allows the near elimination of false positives. This test with only 85% specificity in standard form, would now be useful for sanctioning purposes even at very low rates of background prevalence.

Most importantly, the box-plot shows that the boosted test now targets individuals within networks where there is a high co-prevalence of doping. This targeting means that any given sanction has a far greater probability of disrupting a doping network. Similarly, clean athletes who are pro-active in seeking out clean networks and promoting ethical behavior are at far lower risk of a random false positive.

Conclusion: A Bayesian approach utilizing transverse monitoring of the athlete network can significantly improve individual doping classification when individual probability correlates within their network.

Significance: This method has broad implications for future anti-doping strategies to more accurately target the central nodes of doping behavior.

Works Cited:

Cook AE, Hacker DJ, Webb AK, Osher D, Kristjansson SD, Woltz DJ, Kircher JC. Lyin’ eyes: ocular-motor measures of reading reveal deception. J Exp Psychol Appl. 2012 Sep;18(3):301-13.

de Hon O, Kuipers H, van Bottenburg M. Prevalence of doping use in elite sports: a review of numbers and methods. Sports Med. 2015 Jan;45(1):57-69.

The post A Bayesian approach to boost individual anti-doping classification accuracy by transverse monitoring of the athlete network appeared first on veloclinic.