New Study Redefines Machine Learning Detection of Familial Hypercholesterolemia

Machine Learning


UK: The quest for early detection of genetic diseases in medical diagnostics is increasingly turning to innovative technologies, with one focus on familial hypercholesterolaemia (FH), an inherited condition that causes high blood cholesterol levels that, if left untreated, significantly increases the risk of heart disease from an early age.

Emerging findings suggest that machine learning may have a role in enhancing the detection of potential cases of familial hypercholesterolemia: Machine learning derived models increased pre-test probability in identifying individuals with a molecular diagnosis of FH compared with current approaches.

“This provides a cost-effective, promising and scalable tool to implement in electronic health records to prioritize possible cases of FH for genetic confirmation,” the researchers wrote. Journal of the American Heart Association.

Familial hypercholesterolemia is a monogenic disease that is highly prevalent yet severely underdiagnosed. Recently, medical researchers and technology experts have been exploring the potential of machine learning algorithms to improve the identification and management of familial hypercholesterolemia cases. Familial hypercholesterolemia is often diagnosed late, in part because it remains asymptomatic until complications such as premature heart disease occur. Early identification of patients with familial hypercholesterolemia could allow for timely intervention and potentially prevent serious cardiovascular events.

Against this background, Christophe A. T. Stevens and his colleagues from Imperial College London, UK, aimed to evaluate whether machine learning algorithms are superior to clinical diagnostic criteria (history, symptoms, biomarkers) and UK-recommended screening criteria for identifying individuals with mutations that cause FH and to propose screening criteria that could be scaled for the general population.

The analysis included UK Biobank participants who had their whole exome sequenced and were classified as FH if a (probably) pathogenic variant was detected in the LDLR, APOB, or PCSK9 genes.Data were stratified into three datasets: (1) derivation of state-of-the-art statistical and machine learning models, (2) feature importance analysis, and (3) evaluation of the predictive performance of the models against clinical diagnostic and screening criteria (Simon Broome, Dutch Lipid Clinic Network, Family Case Ascertainment Tool, and Make Early Diagnosis to Prevent Early Death).

The main findings of this study are:

  • Of 454,710 participants, 1,03 were classified as having FH.
  • The stacking ensemble model showed the best predictive performance (precision 0.61%, sensitivity 74.93%, accuracy 72.80%, area under the receiver operating characteristic curve 79.12%) and outperformed clinical diagnostic criteria and recommended screening criteria in identifying FH mutation carriers in the validation data set (figures for the best baseline model, the Familial Case Ascertainment Tool, were 69.55%, 0.44%, 65.43%, and 71.12%, respectively).
  • Our model reduced the number needed to screen compared with the family case ascertainment tool (164 vs. 227).

The researchers noted that incorporating machine learning derived models into electronic health records may improve prioritization for genetic confirmation of FH and provide a more effective and efficient approach to identify FH across different clinical settings and adult populations. The researchers suggested that implementing machine learning based screening criteria may enhance early identification and management, potentially reducing the incidence of acute myocardial infarction, revascularization, and cardiovascular deaths associated with undetected cases.

reference:

Stevens CAT, Vallejo-Vaz AJ, Chora JR, et al. “Improving detection of potential cases of familial hypercholesterolemia: is machine learning part of the solution?” J Am Heart Assoc. 2024;13(12):e034434. doi:10.1161/JAHA.123.034434



Source link

Leave a Reply

Your email address will not be published. Required fields are marked *