A Machine Learning-Derived Risk Score Based on Dietary Nutrient Intake for Early Detection and Prognostic Prediction...

Area of Science:

• Pulmonary Medicine
• Data Science
• Preventive Health

Background:

• Preserved Ratio Impaired Spirometry (PRISm) is a subclinical lung condition linked to increased risks of COPD, cardiovascular disease, and mortality.
• Early identification and targeted prevention of PRISm present significant clinical challenges.

Purpose of the Study:

• To develop and validate a machine learning model for predicting PRISm risk.
• To stratify individuals based on PRISm risk and associated health outcomes.
• To explore the utility of dietary and demographic data for non-invasive PRISm assessment.

Main Methods:

• A stacked machine learning model was developed and validated using US National Health and Nutrition Examination Survey (NHANES) data (2007-2012).
• The model integrated dietary intake and demographic features to generate a continuous PRISm risk score.
• Model performance was assessed using ROC curves and calibration; associations with health outcomes were analyzed via logistic regression and Kaplan-Meier analysis.

Main Results:

• The stacked ML model achieved strong predictive performance with an AUC of 0.818 in the test set.
• The PRISm risk score was significantly associated with hypertension, diabetes, cardiovascular disease, and COPD.
• High-risk individuals exhibited substantially increased mortality rates; healthy lifestyle benefits were observed only in the low-risk group.

Conclusions:

• A non-invasive, data-driven model for PRISm risk prediction and health outcome stratification has been developed.
• The PRISm risk score can potentially assist in early screening and personalized prevention strategies.
• Dietary and demographic data offer a valuable approach for identifying individuals at risk for PRISm.