AI Can Provide Appropriate Exercise Suggestions Based On Fitness Tracking Data
Indira Datta - Apr 29, 2019
Now, scientists have developed an AI tool that can provide exercise suggestions for a user based on data on their health and fitness.
- AI Creates This Advertisement For Nike After Being Trained With Its 7-Year Worth Of Commercials
- Trash Talk Robots Are So Good They May One Day Manipulate Human Decision Making
- This Startup Uses AI To Generate Enough Heat To Melt Steel From Solar Energy
Now, scientists have developed an AI tool that can provide exercise suggestions for a user based on data on their health and fitness. FitRec, developed by scientists at the University of California San Diego, has been provided with more than 250,000 workout records of over 1,000 runners.
AI will analyze exercise data into a person's past health and thereby provide a specific and most suitable training route for their future condition by predicting heart rate and their speed in each exercise.
In addition, FitRec can also identify external factors that have an influence on exercise performance. For example, it will determine whether there are hills on its users running routes.
From there, FitRec will offer alternative routes if the runner wants their heart rate to reach a specific level. In order for users to keep their heart rate stable, this tool will notify if their heart rate exceeds their desired level and ask them to slow down.
The team from the United States was one of the first to collect and model a large number of data for scholarly research. However, building and developing FitRec is not really easy because it has a large number of training records but only a small amount of individual data.
Julian McAuley, UC San Diego professor, said personalization is a very important part of this modeling process. Each person's heartbeat when exercising is completely different so it is really complicated and requires high techniques to model them.
Scientists need a tool that can learn all the data it is provided and must be able to use each user's data. The short-term memory network (also known as LSTM) is the introduction of a deep learning model for researchers to adjust and capture individual variations of each user.
The training data is taken from endomondo.com, a web and training process for many people. Researchers must clean the data and check more than 100,000 records to provide to AI.
FitRec's prediction has been compared and validated with training records that are not on the previously provided list.
FitRec can be developed to analyze other data such as changes and fitness levels of users over time, from which it can provide more relevant and effective suggestions. In addition, this tool also helps users know the safe routes to exercise.
Researchers still need to have access to more detailed exercise tracking data and the ability to handle various quality and data issues so that this AI tool can be included in the fitness app.