Key result
Machine learning algorithms utilizing 15-16% of physical examination indicators achieved 66-99% accuracy (average AUC 87.6%) in predicting 35 different physical statuses compared to healthy individuals.
Why the study?
A systematic investigation of correlations between physical examination indicators is lacking, leading to indicators being used independently and giving general physical examinations limited diagnostic value.
Cross-Sectional (n=811,244)
No
Machine learning algorithms utilizing physical examination indicators can accurately predict various underlying health statuses, potentially enhancing the diagnostic value of general physical examinations.
No takes yet. Share an insight, caveat, or question.
Large observational correlations among PEIs may aid integrated diagnostics; leaves open clinical utility and generalizability.
Wang et al. (2022) conducted a cross-sectional in Healthy and 34 unhealthy statuses (chronic diseases) (n=811,244). Physical examination indicators (PEIs) vs. Healthy physical status was evaluated on Prediction of health status using a Random Forest machine learning algorithm. Machine learning algorithms utilizing 15-16% of physical examination indicators achieved 66-99% accuracy (average AUC 87.6%) in predicting 35 different physical statuses compared to healthy individuals.
Synapse has enriched 4 closely related papers on similar clinical questions. Consider them for comparative context: