Modelling and forecasting Positive and Negative Syndrome Scale scores to achieve remission using time series analysis

Area of Science:

• Psychiatry and Mental Health
• Biostatistics
• Health Services Research

Background:

• Schizophrenia is a chronic mental illness requiring ongoing management.
• Accurate forecasting of patient visits is crucial for effective treatment planning and resource allocation in mental healthcare.
• Predicting the number of outpatient visits needed for remission can optimize treatment pathways for schizophrenia patients.

Purpose of the Study:

• To apply time series modeling to predict the number of outpatient visits required for schizophrenia remission.
• To compare the predictive performance of Autoregressive Integrated Moving Average (ARIMA) and ARIMAX models for forecasting remission timelines.

Main Methods:

• Retrospective data from schizophrenia outpatients in New Delhi were analyzed.
• Autoregressive Integrated Moving Average (ARIMA) and ARIMAX models were fitted to Positive and Negative Syndrome Scale scores.
• Model accuracy was evaluated using Mean Absolute Percentage Error and Mean Absolute Deviation.

Main Results:

• ARIMA (1, 2, 1) and ARIMAX (1, 2, 1) models were identified as suitable for predicting remission visit numbers.
• The ARIMAX model, incorporating the Clinical Global Impression Severity scale, showed better predictive accuracy than the ARIMA model.

Conclusions:

• Time series models, specifically ARIMA and ARIMAX, are effective tools for predicting the number of visits needed for schizophrenia remission.
• The ARIMAX model offers improved forecasting capabilities over the standard ARIMA model for this application.