Predictions of City-based Respiratory Hospital Visits: Developing and Validating a Machine Learning Model with a...

Area of Science:

• Environmental Health
• Epidemiology
• Data Science

Background:

• City-specific tools for assessing and warning about respiratory disease risks are underdeveloped.
• Effective public health response is limited by the lack of localized risk assessment tools.

Purpose of the Study:

• To develop and validate a novel city-specific prediction framework (WHA_air-LSTM) for forecasting daily respiratory outpatient visits.
• To integrate a composite air pollution health index into a prediction model for respiratory disease risk.

Main Methods:

• Constructed and validated a five-level morbidity-driven composite air pollution index (WHA_air) for each city using city-specific exposure-response relationships.
• Built an LSTM model using WHA_air, temperature, humidity, and historical visit data to predict next-day visits.
• Developed the framework with city-level data and externally validated it using datasets from other cities.

Main Results:

• Higher WHA_air levels were significantly associated with increased outpatient visits.
• The WHA_air-LSTM model demonstrated excellent predictive performance (e.g., Beijing: R² = 0.963) and captured visit surges.
• Excluding WHA_air degraded model accuracy, highlighting its importance; the framework showed robust performance in external validation.

Conclusions:

• The WHA_air-LSTM framework offers a scalable and practical tool for city-level respiratory disease early warning.
• It effectively bridges environmental monitoring with clinical practice for improved public health response.
• The framework's transferability was confirmed through external validation.