Tuesday 4 p.m.–4:30 p.m. in PyData
ML for automated behavior analysis in schizophrenia
Talia Tron
- Audience level:
- Intermediate
Abstract
In this talk we will describe the usage of accelerometer and 3d cameras for analysis of motor behavior and facial expressions in schizophrenia patients. We combine descriptive statistical methods together with data-driven analysis techniques including time series forecasting, machine learning and natural language processing (NLP) algorithms. With these techniques, we obtain a wide range of nonverbal characteristic measures, including the intensity, dynamics, consistency and appropriateness of facial and motor behavior. These measures are used to automatically classify clinical sub-population, evaluate symptom severity and identify significant irregularities in patients behavior over time.
