Forecasting the incidence frequencies of schizophrenia using deep learning

Area of Science:

• Neuroscience
• Psychiatry
• Data Science

Background:

• Rising global prevalence of mental disorders necessitates accurate incidence forecasting.
• Effective mental health strategies depend on reliable prediction of mental disorder incidence.

Purpose of the Study:

• To develop and evaluate a Long Short-Term Memory (LSTM) recurrent neural network model for predicting schizophrenia incidence in inpatients in Taiwan.
• To compare the predictive performance of the LSTM model against other statistical and machine learning models.

Main Methods:

• Utilized a Long Short-Term Memory (LSTM) recurrent neural network model.
• Data from Taiwan's National Health Insurance Research Database (NHIRD) for individuals diagnosed with schizophrenia (1998-2015).
• Compared LSTM with exponential smoothing, ARIMA, PSO, PSO-SVR, and DNN models.

Main Results:

• The LSTM model achieved the highest accuracy in predicting schizophrenia incidence.
• LSTM demonstrated the lowest Mean Absolute Percentage Error (2.34), Root Mean Square Error (157.42), and Mean Average Error (154,831.70).

Conclusions:

• The LSTM model is a reliable tool for forecasting mental disorder incidence, specifically schizophrenia.
• Findings offer valuable insights for government administrators and policymakers in developing clinical strategies, healthcare education, and financial planning for mental health support.