Evaluation of coronary heart disease risk prediction based on simple physical examination parameters by machine learning model: a retrospective cohort model development and validation study - Takeaways - MDSpire

Evaluation of coronary heart disease risk prediction based on simple physical examination parameters by machine learning model: a retrospective cohort model development and validation study

  • By

  • Hui Xiong

  • Xiang Cao

  • Xiao Han

  • Jia-Xing Zhang

  • Jia-Rui Zhuang

  • Shuai He

  • Min Zhu

  • Ji Li

  • Wei Qin

  • April 30, 2026

  • 0 min

Share

  • 1

    A machine learning model was developed using the Framingham Heart Study cohort to predict coronary heart disease (CHD) risk from routine clinical indicators.

  • 2

    The model achieved an internal validation AUC of 0.977 and accuracy of 0.942, demonstrating strong predictive performance.

  • 3

    External validation with a separate cohort yielded an AUC of 0.929 and accuracy of 0.885, confirming the model's generalizability.

  • 4

    Key predictors identified included systolic blood pressure, age, total cholesterol, and fasting glucose, with nonlinear interactions observed.

  • 5

    The study highlights the potential of machine learning to enhance CHD risk assessment, offering a clinically interpretable tool for healthcare providers.

Original Source(s)

Evaluation of coronary heart disease risk prediction based on simple physical examination parameters by machine learning model: a retrospective cohort model development and validation study

  • by Hui Xiong, Xiang Cao, Xiao Han, Jia-Xing Zhang, Jia-Rui Zhuang, Shuai He, Min Zhu, Ji Li, Wei Qin

  • April 30, 2026

Related Content

NPs Top US Jobs List Again

The nurse practitioner profession claims the No. 1 spot across three categories in the U.S. News & World Report 2026 Best Jobs rankings for the third consecutive year.

Gov. Patrick Morrisey walking Mountaineer Miles to encourage West Virginians to lose weight

AI Scribes Lag Clinicians on Note Quality

A VHA study across 11 vendors finds AI-generated primary care notes score lower than clinician-written notes, with the largest deficits in thoroughness, organization, and usefulness