Search

We wanted to assess if "Explainable AI" in the form of extreme gradient boosting (XGBoost) could outperform traditional logistic regression in predicting myocardial infarction (MI) in a large cohort. Two machine learning methods, XGBoost and logistic regression, were compared in predicting risk of MI. The UK Biobank is a population-based prospective cohort including 502 506 volunteers with active consent, aged 40 to 69 years at recruitment from 2006 to 2010. These subjects were followed until end of 2019 and the primary outcome was myocardial infarction. Both models were trained using 90% of the cohort. The remaining 10% was used as a test set. Both models were equally precise, but the regression model classified more of the healthy class correctly. XGBoost was more accurate in identifying individuals who later suffered a myocardial infarction. Receiver operator characteristic (ROC) scores are class size invariant. In this metric XGBoost outperformed the logistic regression model, with ROC scores of 0.86 (accuracy 0.75 (CI ±0.00379) and 0.77 (accuracy 0.77 (CI ± 0.00369) respectively. Secondly, we demonstrate how SHAPley values can be used to visualize and interpret the predictions made by XGBoost models, both for the cohort test set and for individuals. The XGBoost machine learning model shows very promising results in evaluating risk of MI in a large and diverse population. This model can be used, and visualized, both for individual assessments and in larger cohorts. The predictions made by the XGBoost models, points toward a future where "Explainable AI" may help to bridge the gap between medicine and data science. [ABSTRACT FROM AUTHOR]

Background: It is commonly assumed that high oestradiol levels in women are cardioprotective. We assessed the association between oestradiol and the risk of incident myocardial infarction (MI) in women.Methods: We used data from 263 295 female UK Biobank participants [mean age 56.2; standard deviation (SD) 8.0 years] without previous cardiovascular disease (CVD). Associations of oestradiol with age and other cardiovascular risk factors were assessed. Cox proportional hazards models estimated crude, ag- and multiple-adjusted hazard ratios (HR) for MI associated with oestradiol levels.Results: After a mean follow-up of 9 years, 2206 incident cases of MI had been recorded, including 230 events among the 57 204 women (mean age 48) with detectable oestradiol levels. In the unadjusted analyses, a unit higher in log-transformed oestradiol was associated with an HR [95% confidence interval (CI) for MI of 0.73 (0.58; 0.92)]. After adjusting for age, this HR became 0.94 (0.75; 1.17), and after further adjusting for classical CVD risk factors, it was 1.05 (0.83; 1.31. Results were similar in subgroup analyses defined by age, menopausal status, socioeconomic status, contraceptive pill use and the use of hormone replacement therapy. The multivariable-adjusted HR for the 171 431 women (mean age 59) with undetectable levels of oestradiol, compared with those with detectable levels, was 0.97 (0.92; 1.02).Conclusions: Higher levels of oestradiol were not associated with a decreased risk of MI. The presumed cardioprotective effects of oestradiol seem to be largely confounded by age and further confounded by other cardiovascular risk factors. [ABSTRACT FROM AUTHOR]

Objective: Diabetes has shown to be a stronger risk factor for myocardial infarction (MI) in women than men. Whether sex differences exist across the glycemic spectrum is unknown. We investigated sex differences in the associations of diabetes status and glycated hemoglobin (HbA1c) with the risk of MI.Research Design and Methods: Data were used from 471,399 (56% women) individuals without cardiovascular disease (CVD) included in the UK Biobank. Sex-specific incidence rates were calculated by diabetes status and across levels of HbA1c using Poisson regression. Cox proportional hazards analyses estimated sex-specific hazard ratios (HRs) and women-to-men ratios by diabetes status and HbA1c for MI during a mean follow-up of 9 years.Results: Women had lower incidence rates of MI than men, regardless of diabetes status or HbA1c level. Compared with individuals without diabetes, prediabetes, undiagnosed diabetes, and previously diagnosed diabetes were associated with an increased risk of MI in both sexes. Previously diagnosed diabetes was more strongly associated with MI in women (HR 2.33 [95% CI 1.96; 2.78]) than men (1.81 [1.63; 2.02]), with a women-to-men ratio of HRs of 1.29 (1.05; 1.58). Each 1% higher HbA1c, independent of diabetes status, was associated with an 18% greater risk of MI in both women and men.Conclusions: Although the incidence of MI was higher in men than women, the presence of diabetes is associated with a greater excess relative risk of MI in women. However, each 1% higher HbA1c was associated with an 18% greater risk of MI in both women and men. [ABSTRACT FROM AUTHOR]