Advancing Patient Care: Synthetic Data Generation  And Human Action Recognition for Early Screening and Patient Monitoring

Asma Bensalah will defended her dissertation on April 28, 2025, on CVC's Conference Room.

What is the thesis about?

"While health has no price, it certainly has a cost. The scope of this thesis is to reduce costs related to health issues using artificial intelligence and computer vision. In particular, optimizing costs related to stroke patient rehabilitation by providing solutions to efficiently assess patients’ progress and optimize time and effort throughout their rehabilitation journey. Second, we focus on early screening for neurodegenerative diseases through handwriting to ensure that patients have a good quality of life as long as possible.

The first part of this thesis involves the implementation of frameworks for the continuous monitoring of patients using cost-efficient wearables and for detecting the purposeful movements of both healthy subjects and those with impairments in a controlled experimental environment. The goal is to perform Human Activity Recognition using different deep learning architectures in a contraindication/unconstrained setting. Furthermore, we analyze the kinematics of recognized movements to evaluate improvements in the human neuromotor system (post-rehabilitation), define progress biomarkers, and clinically validate them.

In the context of early screening for neurodegenerative disorders, the second part of this thesis addresses the lack of data for early screening applications. It explores guided synthetic data generation for in-air movements in handwriting samples related to neurodegenerative disorders, specifically Alzheimer’s disease, using GANs. In addition, we introduce HS-GEN a Controlled Human-Like Handwriting Synthetic Generation method that is individual-centered and does not require training. "

Keynotes

Human Activity Recognition, Kinematic Computational Model, Deep Learning, Machine learning, Synthetic Handwriting Generation, Kinematic controlled Generation, Movement Smoothness, Stroke, Neurodegenerative Diseases.