Plasma metabolite based clustering of breast cancer survivors and identification of dietary and health related characteristics: an application of unsupervised machine learning

• DOI: 10.4162/nrp.2025.19.2.273
• Publication Date: 2025-04-04T06:35:38Z
• Authors: Ga-Eun Yie; Woojin Kyeong; Sihan Song; Zisun Kim; Hyun Jo Youn; Jihyoung Cho; Jun Won Min; Yoo Seok Kim; Jung Eun Lee
• Keywords: Identification

This study aimed to use plasma metabolites identify clusters of breast cancer survivors and compare their dietary characteristics health-related factors across the using unsupervised machine learning. A total 419 were included in this cross-sectional study. We considered 30 metabolites, quantified by high-throughput nuclear magnetic resonance metabolomics. Clusters obtained based on 4 different clustering methods: k-means (KM), partitioning around medoids (PAM), self-organizing maps (SOM), hierarchical agglomerative (HAC). The t-test, χ2 test, Fisher's exact test used sociodemographic, lifestyle, clinical, clusters. P-values adjusted through a false discovery rate (FDR). Two identified methods. Participants cluster 2 had lower concentrations apolipoprotein A1 large high-density lipoprotein (HDL) particles smaller HDL particle sizes, but higher chylomicrons extremely very-low-density-lipoprotein (VLDL) glycoprotein acetyls, ratio monounsaturated fatty acids acids, larger VLDL sizes compared with 1. Body mass index was significantly 1 (FDR adjusted-P KM < 0.001; P PAM = SOM HAC 0.043). clustered basis distinct characteristics. Further prospective studies are needed investigate associations between obesity, factors, prognosis.