Disease risk analysis for schizophrenia patients by an automatic AHP framework

Area of Science:

• Psychiatry and Mental Health
• Data Science and Artificial Intelligence
• Public Health and Epidemiology

Background:

• Schizophrenia poses significant challenges for chronic disease management and clinical research.
• Large-scale patient data analysis is crucial for understanding and predicting disease trajectories.
• Existing models may not fully capture the complex, multifactorial nature of schizophrenia risk.

Purpose of the Study:

• To develop and validate a disease risk analysis and prediction model for schizophrenia patients.
• To support chronic disease management and clinical research using real-world data.
• To identify key factors influencing schizophrenia risk.

Main Methods:

• An automatic Analytic Hierarchy Process (AHP) framework, AutoAHP, was designed.
• Extensive follow-up data (demography, treatment, disease course) from over 400,000 patients were processed.
• Age-period-cohort, logistic regression, and Cox models were integrated for risk assessment and prediction.

Main Results:

• Essential risk factors identified include policy changes, public support, region, gender, compliance, and social function.
• The AutoAHP framework achieved high precision (0.923), recall (0.924), and F1-score (0.923) in validation.
• The model demonstrated superior performance compared to general prediction models.

Conclusions:

• The AutoAHP framework effectively assesses and predicts schizophrenia risk, considering temporal, regional, and complication factors.
• It supports clinical analysis of disease risk factors and aids decision-making in chronic schizophrenia management.
• This approach enhances the utility of large-scale patient data for mental health research and care.