Bajic, Dragana (56186463400)Dragana (56186463400)BajicPajovic, Vladislav (57219029106)Vladislav (57219029106)PajovicMatic, Marija (24491941700)Marija (24491941700)MaticVasic, Marko (56277862600)Marko (56277862600)VasicSarenac, Olivera (23971098200)Olivera (23971098200)SarenacJapundzic-Zigon, Nina (6506302556)Nina (6506302556)Japundzic-Zigon2025-07-022025-07-022020https://doi.org/10.1109/ESGCO49734.2020.9158036https://www.scopus.com/inward/record.uri?eid=2-s2.0-85091116586&doi=10.1109%2fESGCO49734.2020.9158036&partnerID=40&md5=a2d58a5b0e87ddfdca2b1585d1ab26ddhttps://remedy.med.bg.ac.rs/handle/123456789/12514Doxorubicin (DOX) is a chemotherapy medication used to treat cancer, but inducing deleterious cardiomyopathy resistant to treatment as an adverse effect. The present work aims to define phenotype/s of doxorubicin-induced cardiomyopathy in rats, firs manually, then using the automatic procedures. Male Wistar rats equipped with radio telemetry devices, were randomized in the DOX group (n=18) and control group (n=6). The variables collected included hemodynamic, echocardiography, blood analysis, patohistology, and histomorphometry. The visual arrangement of variables (a heat-map) was performed by hierarchical agglomerative clustering. A robust and reliable subject clustering was performed using the modified evidence accumulation algorithm that combines K-means++, affinity propagation, and Gaussian mixture model. The results point to at least two distinct phenotypes of DOX-induced cardiomyopathy: one with preserved and mid-range left ventricular ejection fraction (LVEF) and another with reduced LVEF. The automatic procedures a) point out the subject that should be manually rechecked and b) assist in finding the solutions for potential ambiguities. © 2020 IEEE.