Continuous assessment of schizophrenia using heart rate and accelerometer data.

Area of Science:

• Biomedical Engineering
• Psychiatry
• Computational Neuroscience

Background:

• Schizophrenia is linked to altered heart rate (HR) and physical activity.
• Quantifying the impact of analysis window length on classifier accuracy using these features is crucial.

Purpose of the Study:

• To investigate the relationship between analysis window length and classification accuracy for schizophrenia detection using HR and activity data.
• To determine the optimal duration of data collection for reliable daily illness severity estimation.

Main Methods:

• Objective HR and activity data were collected from 12 medicated schizophrenia patients and 12 healthy controls.
• Features included statistical characteristics, rest-activity metrics, transfer entropy, and multiscale fuzzy entropy.
• A support vector machine classifier was trained and validated using varying analysis window lengths (2-8 days).

Main Results:

• An 8-day analysis window achieved an AUC of 0.96, while a 2-day window yielded an AUC of 0.91.
• The most predictive features varied depending on the analysis window length.
• High classification accuracy was maintained even with shorter data windows.

Conclusions:

• Continuous tracking of individuals with schizophrenia over short durations is sufficient for daily illness severity estimation.
• Objective HR and activity monitoring offer a promising approach for real-time assessment in schizophrenia.