A Novel, Low-Cost Device for Diagnosing Neurological Disorders

Neurological disorders, including Parkinson’s disease, Alzheimer’s disease, and the many others, are leading contributors to global disability and mortality. Effective management of these conditions depends on early and accurate diagnosis, yet access to advanced diagnostic tools like MRI is limited in low-resource settings due to high costs and infrastructure constraints. This disparity delays diagnosis and treatment, worsening patient outcomes and overburdening healthcare systems. No integrated system currently exists to diagnose a large spectrum of neurological disorders in a single shot.

In this project, I found that Deep Convolutional Neural Networks can be trained to have a high understanding of neural time series and can use techniques like power-spectral density analysis and discrete wavelet transforms to compress input data, allowing for low-cost computation and inference of neurological disorders.