Clinical ink reposted this
We are pleased to report another publication from the WATCH-PD study, in which Clinical ink's Movement Disorder platform was used to remotely monitor Parkinsonian symptoms using a multi-domain collection of sensor and wearable capabilities. In this work, we developed a high-dimensional feature engineering library to construct classification models of PD status. Highlights from this study include: High classification accuracy --> our model predicted PD status with 92% accuracy, 90% sensitivity, and 100% specificity, providing a level of accuracy higher than standard clinical methods in early PD Environment agnostic --> our model was able to predict PD status in both clinic and home environments, allowing for remote home administration with the same level of performance as when completed in clinic Platform agnostic --> our features and models were able to perform well when applied to data from the mPower study, demonstrating applicability across studies and platforms Enormous thanks for our collaborators on this work: Ray Dorsey, Jamie Adams, Melissa Kostrzebski, MS, MBA, Josh Cosman, Tairmae Kangarloo, Joan Severson, Steve Polyak, Ph.D., Shane Johnson, PhD, Michalis Kantartjis, Allen Best, Daniel Jackson Amato, Brian Severson, Anna Revelez, Michael Merickel, Sean Jezewski
