Electronic health record (EHR) feature engineering is a major bottleneck in clinical research and AI, accounting for 39-45% of data scientists'workload. This is especially pronounced in heart failure, which affects an estimated 6.7 million U.S. adults and requires integrating fragmented EHR data with disease-specific, guideline-based clinical reasoning. Existing rule-based and large language model (LLM)-based approaches offer only partial automation with limited maintainability and evidence traceability. We developed the Nimblemind Multi-Agent System (nMAS), an evidence-linked, rubric-grounded pipeline for automated heart-failure feature engineering, and evaluated it on 500 dummy patient records from nine EHR source tables. nMAS generated 132 structured and 70 rubric-scored aggregated features, verified for structural integrity, rubric compliance, and provenance, and audited by a restricted LLM. Adding the aggregated features improved held-out AUROC from 0.895 to 0.963 for HFrEF and 0.870 to 0.910 for HFpEF phenotyping, and an independent LLM-based rubric assessment of evidence support and methodological soundness scored the features at 81.5% of maximum points. These results demonstrate the feasibility of automated, auditable feature engineering for complex cardiovascular EHR data, though evaluation was limited to a single-institution cohort and external validation is needed.
Cardiovascular disease (CVD) is a major source of morbidity and mortality across the globe. Effective screening and risk stratification of patients can improve long‑term outcomes by allowing physicians to offer fine-tuned advice and treatments. Advances in technology have enabled the study of various biomarkers associated with CVD, allowing researchers to study the molecular processes underlying illness. In this study, we sampled recent advances in the application of Machine Learning (ML) techniques to transcriptome datasets of CVD patients in the hopes of progressing our understanding of underlying pathophysiology, improving our ability to detect disease, and refining our methods of risk stratification of patients. These studies suggest that ML algorithms may help identify complex relationships within high-dimensional datasets to isolate relationships within datasets that were previously missed, enhancing our understanding of disease. However, there are several challenges that need to be addressed before these innovations can reach the clinic. Researchers must verify their results via means that are readily obtainable for physicians and must adequately identify the indications for a particular test so they can integrate into preexisting workflows. We offer potential solutions and suggestions to these issues so that these approaches may eventually contribute to improved patient care.
