Our team, working in collaboration with a US-based startup, is developing a diagnostic system powered by Artificial Intelligence to augment physical therapy programs for patients. The application identifies patient posture- and movement disorders, and recommends personalized exercises to speed up the recovery process.
By using 3D body scans, image processing technology, and machine learning classification algorithms, the system can identify the patient's needs and use this to recommend the most suitable treatment.
Project Category: AI, data analytics, image processing, software development
Lead Member: Sparkbit
Project Description & User Benefits
The application's end is to develop a personalized movement plan for individual patients, allowing them to monitor their progress continuously and constantly upgrade their exercises as their recovery progresses.
This solution is powered by an ML-based diagnostic tool that can recognize about 110 landmarks on a 3D body scan, and then matches these landmark positions with the silhouette analysis algorithms, enabling the system to identify posture and movement defects. After identifying the defects, the application recommends predefined corrective exercises for the patients based on their defects, which can be incorporated into their rehabilitation program.
With the adaptive algorithms built into the application, the efficiency of the system increases over time, and as the algorithm collects more data about a patient, it is able to identify more nuanced patient’s posture abnormalities. Similarly, the algorithm becomes better at generalizing the acquired knowledge over time to provide better results across various patient segments with different body posture and movement disorders, and hence benefiting them with faster diagnoses, and personalized PT program recommendations.
Primary technologies used: Python, Tensorflow