Development and Validation of Risk Scores for All-Cause Mortality for a Smartphone-Based "General Health Score" App:...

Area of Science:

• Digital Health
• Biostatistics
• Epidemiology

Background:

• Existing health metrics often fail to capture broad determinants of health.
• There is a need for accessible, multidimensional health assessment tools.

Purpose of the Study:

• To develop and validate a novel, points-based health score (C-Score) using smartphone-measurable data.
• To assess the predictive accuracy of the C-Score for all-cause mortality.
• To compare statistical and machine learning (ML) models for health risk prediction.

Main Methods:

• Literature review to identify predictor variables.
• Prospective cohort study (UK Biobank, n=420,560).
• Validation of C-Score and comparison of statistical/ML models for predicting 10-year all-cause mortality.

Main Results:

• The C-Score demonstrated good discrimination for all-cause mortality (c-statistic=0.66).
• Each decile increase in C-Score correlated with a 31% relative risk reduction in mortality.
• A Cox model with C-Score and age improved discrimination (c-statistic=0.74); ML models did not offer superior prediction.

Conclusions:

• The C-Score is a valid and predictive metric for 10-year all-cause mortality.
• Integrating the C-Score into a smartphone app could democratize health risk prediction.
• A simple Cox model using C-Score and age provides a balance of accuracy and interpretability for risk estimation.