New machine learning model helps predict heart disease risk in women – Critical Care

Image: New machine learning model helps solve the problem of underdiagnosed heart disease in women (Photo credit: 123RF)

In the field of heart health, cardiovascular disease is significantly underdiagnosed in women compared to men. The commonly used Framingham Risk Score predicts the likelihood of developing cardiovascular disease within the next 10 years and is based on certain criteria such as age, gender, cholesterol levels, and blood pressure. However, this does not take into account the anatomical differences between the sexes. For example, women's hearts are generally smaller and have thinner walls. As a result, using the same diagnostic criteria for men and women means that women's hearts need to grow disproportionately more than men's to meet the same risk criteria. A team of researchers has now used a large dataset to build a cardiovascular risk model that is more accurate than the Framingham Risk Score and also quantified underdiagnosis in women compared to men.

Researchers at Stanford University (Stanford, California, USA) quantified the underdiagnosis of women compared to men and found that using gender-neutral criteria resulted in significant underdiagnosis of female patients. To achieve a more accurate prediction of men and women, they examined her four additional indicators not included in the Framingham risk score (cardiac magnetic resonance imaging, pulse wave analysis, electrocardiogram, and carotid ultrasound). ) has been incorporated. They applied machine learning techniques, drawing on data from her more than 20,000 individuals in the UK Biobank, a comprehensive biomedical database of around 500,000 UK residents aged over 40. They found that electrocardiograms were particularly effective at increasing the detection of cardiovascular disease in both men and women. Nevertheless, researchers say traditional risk factors remain valuable in assessing risk.

This study is a first step in reevaluating risk factors for heart disease by incorporating advanced technology to improve risk prediction. Nevertheless, this study faces limitations that future research should address. One such limitation is the binary treatment of sex in UK Biobank, which includes hormones, chromosomes, and physical characteristics that may not fit neatly into the categories of “male” or “female.” ignores the complex nature of Additionally, this study focused on middle-aged and older UK residents, which may limit the applicability of the findings to other demographic groups and geographic locations.

“We found that gender-neutral criteria do not adequately diagnose women. If gender-specific criteria were used, this underdiagnosis would be less severe,” said researchers at Stanford University's Living Matter Institute. Schuyler St. Pierre said. “We also found that the best test to improve the detection of cardiovascular disease in both men and women is the electrocardiogram (EKG).”

“Traditional clinical models are easy to use, but we use machine learning to examine thousands of other possible factors to find new and meaningful features that have the potential to significantly improve early detection of disease. ” added Saint-Pierre.

Related Links:
stanford university