Recurrence risk prediction of acute coronary syndrome per patient as a personalized ACS recurrence risk: a retrospective study

PeerJ. 2022 Nov 15:10:e14348. doi: 10.7717/peerj.14348. eCollection 2022.

Abstract

Acute coronary syndrome (ACS) has been one of the most important issues in global public health. The high recurrence risk of patients with coronary heart disease (CHD) has led to the importance of post-discharge care and secondary prevention of CHD. Previous studies provided binary results of ACS recurrence risk; however, studies providing the recurrence risk of an individual patient are rare. In this study, we conducted a model which provides the recurrence risk probability for each patient, along with the binary result, with two datasets from the Korea Health Insurance Review and Assessment Service and Chungbuk National University Hospital. The total data of 6,535 patients who had been diagnosed with ACS were used to build a machine learning model by using logistic regression. Data including age, gender, procedure codes, procedure reason, prescription drug codes, and condition codes were used as the model predictors. The model performance showed 0.893, 0.894, 0.851, 0.869, and 0.921 for accuracy, precision, recall, F1-score, and AUC, respectively. Our model provides the ACS recurrence probability of each patient as a personalized ACS recurrence risk, which may help motivate the patient to reduce their own ACS recurrence risk. The model also shows that acute transmural myocardial infarction of an unspecified site, and other sites and acute transmural myocardial infarction of an unspecified site contributed most significantly to ACS recurrence with an odds ratio of 97.908 as a procedure reason code and with an odds ratio of 58.215 as a condition code, respectively.

Keywords: Acute coronary syndromes; Logistic regression; Personalized recurrence risk.

Publication types

  • Research Support, Non-U.S. Gov't

MeSH terms

  • Acute Coronary Syndrome* / diagnosis
  • Aftercare
  • Coronary Disease* / complications
  • Humans
  • Myocardial Infarction* / complications
  • Patient Discharge
  • Retrospective Studies

Grants and funding

This research was supported by the Ministry of Trade, Industry and Energy (MOTIE) and the Korea Institute for Advancement of Technology (KIAT) through the International Cooperative R&D program (grant number: 1415166057) and by the Basic Science Research Program through the National Research Foundation of Korea (NRF) funded by the Ministry of Education (grant number: NRF-2020R1I1A1A01071884). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.