Step 1: With authorization, we connect to your Strava account to access your data.

Step 2: We compute a number of load-related features from your running data for each of three time periods: (i) the current week, (ii) the previous three weeks (average), and (iii) the previous twelve weeks (average). The features include total distance, average pace, and consistency metrics, among others. Each feature is a measurable load metric, i.e. the features are not estimates or formulations.

Step 3: We input the features from Step 2 into a patented (U.S. Patent No. 11,515,045) machine learning model.

Step 4: The machine learning model outputs a risk score, and we post this risk score to the most-recent activity in your Strava account. Risk scores are LOW, MEDIUM, and HIGH risk of injury.

Importantly, RunWise AI does NOT store any of your Strava data after inputting it into the machine learning model.

Example Strava Activity With RunWise Injury Risk (Green Highlighting)

What should I do with the risk score? The risk score is another data point. For example, if the risk score is MEDIUM or HIGH, the machine learning model 'sees' something in the features from Step 2 that is similar to a feature set from the training data linked to an injury. You may therefore pause and think about your short-term training plan. Maybe you need a day off? Or do you need an easier (e.g. shorter, less intense) day? In other words, use the risk score in combination with other information to guide, and possibly adjust, your training.