Introducing myTrainingForecast.run: a science-based injury prevention web app

From garden ornament to running injury prediction algorithms

That massive otter statue had been at the back of our garden for years, forgotten by the previous owner, but that day in May 2017 was the day I was going to relocate it to its new home: the local tip. Skilfully manoeuvring it through the patio, down the steps, and while swinging it to enter the garage I firmly planted my right knee in the metal door frame. This was bad, I could barely walk. My first thought: “will I be able to run the Worthing 10K race in 4 days’ time?!“. I followed all the usual advice (rest, ice, compress, elevate), then warily stood at the start line but happily completed my very first 10 km race in 44:01, just a minute slower than my original target time. I knew racing straight after this injury was playing with fire, so I took 2 weeks almost completely off running to let my knee recover. What happened next completely changed the way I plan my weekly runs to stay injury-free.

otters garden statue

I’d only been running for a few months and I was at my peak fitness, so after a week of tapering down and two weeks off running I couldn’t wait to get straight back into hard training. And so I did. But soon the wheels started to fall off: both knees started hurting, my hamstring became too sore to race my local parkrun. Every time I tried to get back into regular training something new started hurting, and so my mileage spiralled down to a mere 5 km/week. In less than 2 months I had gone from peak fitness to being unable to run for a week without some part of me breaking down.

Too much too soon!

My predicament was swiftly diagnosed: “toomuch-toosoonitis“. After my knee injury I should have not aimed for my usual mileage, I should have ramped up more slowly. Identifying my problem was a good start but it wasn’t much help, what I needed to know was how much I could run safely at that point in time. My mileage was pretty much zero, the 10% rule was useless (and I since learnt that there’s no evidence for it whatsoever, more on that in a future blog post).

How much is too much? How soon is too soon?

I remembered reading an article by Alex Hutchinson in Runner’s World, about a new way to manage running volume to avoid injury. I did some more research and found many recent sports science articles, often hidden behind paywalls, which describe evidence-based methods to manage training load and minimise the risk of overuse injury. The “acute:chronic ratio” (ACR) method (1 week mileage divided by the average weekly mileage for the past 4 weeks) has particular evidence of working well to predict injuries. In fact it’s used by coaches in many sports at professional level including football, rugby and cycling. For some reason this hasn’t become a standard tool for runners like us, even though we clearly need a reliable way of knowing how much is too much, and how soon is too soon.

Can some running injuries be predicted?

So I set up an Excel spreadsheet to calculate my ACR and found that the 2-week recovery from my gardening incident had sent me deep into the blue ‘de-training’ zone, with my fitness plummeting. With my fitness now reduced, my attempt to quickly get back into running made my mileage spike into the red zone. As you can see in the chart below my mileage then dropped down to 20 km/week, before going into the orange zone and quickly crashing down into the blue zone. I tried to get running again but mid-July both of my knees were throwing the towel, my mileage took a final nosedive to about 5 km/week. Looking at this chart I had a satisfying realisation that this vicious training cycle of doing-too-much/crashing-down made perfect sense. But this feeling rapidly turned bitter: my aches, pains and 2 months of lost training could have been predicted (and possibly prevented) by an Excel spreadsheet!

I was curious to see if my previous injuries could have been predicted by an elevated ACR value, so I plotted my training load for January 2017 when I suffered a torn trapezius muscle during an interval run. Sure enough I had gone through another red/blue/red cycle between the end of December and the day I got injured (17 January). Although my ACR was ‘only’ 150% (down from 170% the day before) notice how my relative risk of injury had remained very high at x6. This was due to the accumulation of risk caused by dipping into the blue zone and peaking into the red within a week. If I’d had this information back in January 2017, I would have maintained my mileage at around 30 km/week (middle of green zone) rather than pushed to 45 km/week, maybe avoiding this injury and a dip in training volume over the following week.

Training forecast

It then struck me that if my ACR calculations could identify training errors in the past, I could compute all the statistics forward and forecast training errors in the future. This could tell runners how much will be too much and how soon will be too soon, based on their previous weeks of training. The chart below shows how this works, the solid black line represents the weekly distance I’ve run over the past weeks, and the dotted black line shows the plan for the next 7 days, based on the numbers I’ve typed in the ‘plan’ input fields.

As of today I’ve run 51 km over the past 7 days (red dot), and I plan to run 5 km after this post. This takes me to the edge of the orange zone, but I know it will be fine because: 1) my previous 4 weeks were almost entirely in the green zone, and 2) I’m going for an easy run, pushing out of the green zone through speedwork or racing would be much riskier. Although I still occasionally dip into the high risk zones, I make sure I never do so through a hard workout or a race, as this would mean going into the red with both distance and effort: a sure way to get hurt! In fact I now use myTrainingForecast’s planner when I need to taper for an important race, this way I can prevent my hard effort from turning into an injury. And this is the basic idea of myTrainingForecast: to turn sports science research into knowledge and actions accessible to every runner, for free.

Personalised predictions

The ACR model helped me escape a cycle of running injuries and get back onto a healthier training, but just like any other mathematical model it’s a very simplified approximation of many factors that lead us to getting injured. The ACR only accounts for mileage we have run in the past 4 weeks; our running pace, course elevation gain, running experience, our age, our cross-training and previous injuries do not go into the equation at all, even though we all know these are important factors. So the next step was to implement some of these parameters on top of the ACR model, this is the personalised injury prediction model included in the Premium version of myTrainingForecast.run

What are your thoughts? How do you manage your training volume to minimise aches, pains and full-blown injuries? Has a garden ornament ever changed the way you train?

Sign up for a free account to see if the ACR model can help you avoid training errors. Already using myTrainingForecast? Why not try our Premium personalised injury prediction model with a free 2-week trial? You’ll get more personalised predictions and it helps me keep the website going and free of adverts.

Thanks for reading, I would love your feedback!

One Reply to “Introducing myTrainingForecast.run: a science-based injury prevention web app”

  1. Great blog post! I’ve been using mytrainingforecast.run since January 2018 when I heard about it on Strava. I was interested in both preventing future injuries and understanding how I got into a cycle of knee injuries leading up to my first half marathon.

    Just as indicated in this blog post, there was a clear correlation between training above 150% ACR. Since using this tool to manage my mileage and plan my runs I’ve remained injury-free while increasing weekly mileage from 12-15 miles to a current average of greater than 35 miles per week!

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.