Digital Biomarkers

Actigraphy-based meal and toilet events detection

Background

Reduced bowel activity is a persistent symptom in Parkinson’s, appearing years before motor symptoms, and affecting QOL. Constipation and food intake are often responsible for the slow absorption of levodopa.

Approach

Circadic uses wearable-based activity recognition algorithms to establish a GI function index that logs meals and toilet use to help doctors prescribe medication in the right dosage and timing to help patients better manage their schedule.  

Under the hood

We use consumer-grade wearables to capture movement profile (IMU-based) that are fed (in 10-second periods) into a previously trained feature-based ML algorithm whose score expresses the probability of a subject having a meal or a using the toilet.