BioCoach: the AI that prevents injuries in weightlifting

An AI for Weightlifting to Prevent Injuries

Artificial intelligence is now making its way into the world of weightlifting, aiming to reduce the risk of injuries. A group of researchers has developed a prototype capable of analyzing athletes' movements in real-time. This system, named BioCoach, aims to correct posture errors before they cause bodily harm.

Practicing weightlifting alone can be risky, especially without an expert eye to correct movements. An incorrect posture or a poorly executed gesture can turn a workout session into a source of pain. This is precisely the problem that BioCoach seeks to address. Thanks to AI, this system analyzes the user's biomechanics to prevent mistakes before they turn into injuries.

BioCoach: Intelligent Movement Monitoring

The pandemic saw a significant increase in exercise-related injuries at home in the United States. Contrary to popular belief, it was not the equipment that was to blame, but often the users' posture.

BioCoach, developed by researchers from Drexel University and Michigan State University, offers an innovative approach. This AI observes the user through their smartphone camera, analyzing their body mechanics and identifying the joints most stressed during exercises.

For example, when performing push-ups, BioCoach focuses on the shoulders, elbows, and wrists. Unlike other applications that merely remind users to "keep their back straight," BioCoach provides more detailed advice.

The prototype can offer precise recommendations based on the user's anatomy. It may suggest adjusting the angle of elbow flexion or modifying the range of motion. These tips are particularly valuable for beginners who may struggle to distinguish between proper execution and risky posture.

A Virtual Coach that Educates and Corrects

One of the most innovative aspects of BioCoach is its ability to explain the recommendations it provides. To achieve this, the system employs several technologies. On one hand, a model analyzes the visual appearance and movements of the user. On the other hand, a system reconstructs a 3D skeleton to examine the joints, angles, and phases of movement.

Thus, the advice provided goes beyond simple corrections. The system also strives to explain the mechanical consequences of a poor gesture. This educational approach is crucial, as understanding why a movement is incorrect is often more helpful than receiving a simple auditory alert.

The researchers compared BioCoach to other solutions developed by major tech companies. According to their studies, BioCoach outperformed its competitors in terms of the quality of explanations and the accuracy of anatomical feedback.

However, it is important to note that BioCoach remains a research prototype for now. No public application is yet available.