new subscribers after first year
Reduction in measurement time
From 0 to production
The mySASY application is intended for everyone who does sports, whether on a professional level or in their free time. With this application you can improve the way you schedule your training. It helps you understand when it is a good time to continue or increase training intensity and when to rest, which helps you prevent injuries or other complications. Simply put, the application evaluates the current state of the autonomic nervous system (ANS) and gives you recommendations based on this data.
To determine the state of the autonomic nervous system, mySASY uses a unique measurement methodology. The algorithm allows to control the athlete's training based on the analysis of ANS activation using the methods of spectral analysis of heart rate variability (HRV). At this stage, we got involved in the development process to help reduce measurement time with AI methods.
We processed hundreds of thousands of measurements, more precisely 1,041 days of data, and with the help of machine learning methods we managed to shorten the whole process while maintaining the accuracy of a long measurement. It took users 15 minutes to get measured and now it can be done in 4 minutes.
We created a model from the collected data, which can extract the same information value from measurements that are three times shorter.
"The model we designed using advanced artificial intelligence methods is comparable to the experience of a someone who has been constantly analyzing thousands of people for three years," adds Jaroslav Vážný, who participated in the development of the model.
As with all machine learning applications, the most important part was the processing of raw data and the extraction of useful properties. A key element of this project is the automatic recognition of measurement phases (to activate ANS it is necessary to change the position of the body during the measurement). The correct division into phases that are unique to each individual and various loads on the body is a non-trivial task. We used a combination of methods led by wild binary segmentation (WBS). The data processed in this way was used to train the model. We tested various approaches, including recurrent neural networks, and finally used the Gradient Boosting Decision Tree (GBRT).
mySASY training application analyses heart rate HRV and activity data to help both amateur and professional athletes plan their training according to the current state of the autonomic nervous system (ANS).
We created a functional, cheaper and clearer data structure based on a new logic.
We prepared for Vodafone a three-day practical training course about big data work on the Cloudera platform.