An Artificial Intelligence and Biochemical Modelling Approach for LDL Cholesterol Prediction
Year: 2025
Type: article
Abstract: Estimating LDL cholesterol accurately remains a critical challenge in cardiovascular medicine. While ultracentrifugation delivers precise results, its high cost and complexity make it unsuitable for everyday clinical practice. Current estimation formulas like Friedewald and Martin-Hopkins are convenient but frequently miss the mark due to patient-specific factors and biological variations. Direct LDL testing, though reliable, puts a significant financial strain on healthcare systems. We developed a sophisticated machine learning approach that combines the strengths of 13 existing estimation equations with essential patient data—age, sex, total cholesterol, triglycerides, and high-density lipoprotein levels—alongside actual LDL measurements. The framework uses three different algorithms (k-Nearest Neighbors, Random Forest, and Support Vector Machine) as foundational learners, while XGBoost acts as the final decision-maker to identify intricate data relationships. What sets our method apart is its ability to classify LDL levels according to both NCEP III and European Society of Cardiology standards, offering clinicians a more complete risk assessment tool. When we tested our integrated model against simpler versions, the results were striking: our comprehensive approach achieved an R² value exceeding 0.98, significantly outperforming models that relied solely on basic clinical parameters (R² around 0.95). This advancement could transform how we assess cardiovascular risk, especially in settings where expensive laboratory methods aren't feasible. Received: 13 April 2025 | Revised: 30 June 2025 | Accepted: 18 August 2025 Conflicts of Interest The authors declare that they have no conflicts of interest to this work. Data Availability Statement The data that support this work are available upon reasonable request to the corresponding author. Author Contribution Statement Petros Paplomatas: Conceptualization, Methodology, Software, Validation, Formal analysis, Investigation, Resources, Data curation, Writing – review & editing, Visualization, Project administration. Maria Kantartzi: Validation, Writing – original draft, Visualization. Kostas Anagnostopoulos: Conceptualization, Investigation, Resources, Data curation, Writing – original draft, Visualization, Supervision. Aristidis Vrahatis: Validation, Visualization, Supervision.
Source: Medinformatics
Institutions: Ionian University, Democritus University of Thrace
Language: English
Subfield: Health Information Management
Field: Health Professions
Domain: Health Sciences
Open Access status: diamond
OpenAlex ID: W4414148757