An Application to Monitor Freezing of Gait in Parkinson's Disease

Overview

Parkinson’s disease (PD) is the leading cause of disability worldwide. Indeed, PD cases are rapidly rising, especially in underdeveloped areas that don’t have ready access to therapeutic care. Over 70% of people with PD (PWP) experience potentially fatal falls attributed to freezing of gait (FoG), a sudden muscular stiffness. To address this, an end-to-end FoG detection and prevention system was built with the success criteria of real-time feedback, accuracy, and accessibility. A machine learning model (LSTM) was trained on upper-thigh triaxial accelerometer data from Daphnet in which 14 PWP experienced FoG episodes while completing daily tasks. A mobile application that continuously collects accelerometer data was developed; this data is then analyzed by the LSTM. The completed prototype was evaluated in multiple trials: (1) algorithmic, (2) live testing, and (3) navigability. (1) Algorithmically, the model had an accuracy of 90%. (2) When used by 10 PWP, the application was able to predict FoG 2 seconds before it occurred and subsequently play an auditory cue to reduce FoG severity. (3) All subjects reported the prototype to be navigable without assistance. The upper-thigh accelerometer data alone proved to be statistically significant when comparing FoG to unaffected walking, demonstrating that a smartphone can detect and prevent FoG while being discreetly placed in a patient’s pocket. Overall, the prototype can effectively monitor FoG changes in response to prescribed medication, suggesting potential for lowering PWP’s injury risk especially in under-resourced regions.

Awards (1)

• Third Award of $1,000 $1,000

Competition history

• ISEF 2021 Translational Medical Science · Entry TMED043 Affiliated fair in CA, United States of America

Resources

• ISEF abstract page ↗

Related projects

Closest projects by meaning, across every fair and year in the corpus.