Digital Parkinson’s Tremor Quantification
Doctoral Researcher
During my PhD in Biomedical Engineering at Purdue University, I developed a non-invasive, wearable sensor-based diagnostic system for Parkinson’s disease tremor assessment. I designed signal processing and machine learning pipelines to extract and quantify tremor dynamics from wrist-worn inertial sensors, engineering a data-driven severity metric that demonstrated greater sensitivity than the current clinical gold standard. Working closely with neurologists at a Parkinson’s Foundation Center of Excellence, I aligned the system with established clinical measures and validated it through structured clinician feedback and human subjects testing. This work advanced objective, telehealth-enabled tremor quantification and has been published and presented in leading scientific venues.
What I built
Wearable tremor quantification system
Wrist-worn IMU sensing with a severity metric aligned to clinical scales
Core methods
Signal processing + ML pipeline
Feature extraction, tremor isolation, model-based severity estimation
Clinical validation
Neurologist aligned evaluation
Structured feedback and human-subject testing at a Center of Excellence
The problem
Clinical tremor ratings can be subjective and vary by rater and setting
Tremor fluctuates over time so snapshot clinic assessments miss dynamics
Telehealth needs objective measures that work outside the clinic
What I did
Designed a wrist-worn IMU sensing protocol and analysis pipeline
Engineered tremor-isolation and severity features from kinematics
Built ML-driven severity estimation aligned to clinical measures
Validation
Partnered with neurologists at a Parkinson’s Foundation Center of Excellence
Iterated via structured clinician feedback on interpretability and utility
Evaluated with human-subject testing and benchmark clinical comparisons
Why it matters
Objective, scalable tremor quantification supports longitudinal tracking
Enables remote monitoring and telehealth workflows
Reduces reliance on subjective ratings for symptom quantification