GYMTRAINER: DEVELOPMENT OF A MOBILE VIRTUAL PERSONAL TRAINER APPLICATION WITH ARTIFICIAL INTELLIGENCE FOR STRENGTH-TRAINING GENERATION AND TRACKING
DOI:
https://doi.org/10.51891/rease.v12i11.33141Keywords:
Artificial intelligence, Mobile application, React Native, Strength training, Virtual personal trainerAbstract
This work presents GymTrainer, a mobile strength-training application that acts as a virtual personal trainer. It uses artificial intelligence (AI) to build workouts and tracks the user while they train. The system has two parts: the app, built with React Native (Expo), and a server with a Node.js/Express REST API, which stores data in MongoDB and sends the user's profile to Google Gemini to generate the workouts. The app was evaluated through functional tests, an automated generation test with five profiles and a preliminary test with two real users. All 15 generations tested were completed in the correct format, after two failures caused by the usage limit of the AI service were repeated, and all weekly plans had the right number of workouts. The average response time was 2.1 seconds for single workouts and 4.9 seconds for weekly plans. According to American College of Sports Medicine criteria, 80.4% of the exercises had repetitions suited to the goal and 99.4% had sets suited to the level, and the profile with a knee injury received no exercises to avoid. The results show that it is possible to combine AI with workout tracking in a low-cost app, but the workouts still need to be evaluated by physical education professionals.
Downloads
Downloads
Published
Issue
Section
Categories
License
Atribuição CC BY