@ARTICLE{Liyanage_Himanshi_Computational_2026,
 author={Liyanage, Himanshi and Lipnicka, Marta},
 volume={vol. 36},
 number={No 1},
 pages={99-131},
 journal={Archives of Control Sciences},
 howpublished={online},
 year={2026},
 publisher={Committee of Automatic Control and Robotics PAS},
 abstract={This paper presents a novel computational framework for assessing cardiovascular disease (CVD) risk by integrating unsupervised clustering techniques with survival analysis. The proposed method enables dynamic and individualized risk prediction by organizing patient data into structured clusters based on shared cardiovascular risk factors. The framework begins with competitive learning, an unsupervised clustering method, to group patients into clusters that reflect distinct risk profiles. Each cluster is represented by its centroid, calculated as the mean of the 9-dimensional feature vectors of its members, ensuring that the clusters effectively summarize patient data while preserving critical risk characteristics. For each cluster, an independent Cox Proportional Hazards Model is applied to analyze survival data, capturing the unique relationships between cardiovascular risk factors and survival outcomes within that cluster. A key innovation of this study is the introduction of the Cumulative Prevalence Ratio (CPR), a new metric that aggregates hazard rates over time separately for each cluster. This approach provides a comprehensive view of cumulative cardiovascular risk, enabling precise categorization of the patient into risk groups based on cumulative exposure to evolving risk factors. By integrating cluster-specific hazard functions and temporal risk metrics, the proposed framework improves the precision and adaptability of CVD risk predictions, paving the way for personalized and data-driven healthcare interventions.},
 title={Computational framework for dynamic cardiovascular risk assessment with cluster-specific Cox models and cumulative risk analysis},
 type={Article},
 URL={http://www.czasopisma.pan.pl/Content/138659/PDF-MASTER/acs-art05.pdf},
 doi={10.24425/acs.2026.158423},
 keywords={cardiovascular disease, cox proportional hazards model, independent hazard functions, competitive learning, cluster-based data simplification, cumulative prevalence ratio, personalized risk assessment},
}