PuSH - Publikationsserver des Helmholtz Zentrums München

Niu, J. ; Adam, J. ; Skurk, T.* ; Seissler, J. ; Dong, Q. ; Efiong, E.E. ; Gieger, C. ; Peters, A. ; Sharma, S. ; Grallert, H.

Machine learning approach on plasma proteomics identifies signatures associated with obesity in the KORA FF4 cohort.

Diabetes Obes. Metab. 27, 2626-2636 (2025)
Verlagsversion Forschungsdaten DOI PMC
Open Access Gold (Paid Option)
Creative Commons Lizenzvertrag
AIMS: This study investigated the role of plasma proteins in obesity to identify predictive biomarkers and explore underlying biological mechanisms. METHODS: In the Cooperative Health Research in the Region of Augsburg (KORA) FF4 study, 809 proteins were measured in 2045 individuals (564 obese and 1481 non-obese). Multivariate logistic regression adjusted for confounders (basic and full models) was used to identify obesity-associated proteins. Priority-Lasso was applied for feature selection, followed by machine learning models (support vector machine [SVM], random forest [RF], k-nearest neighbour [KNN] and adaptive boosting [Adaboost]) for prediction. Correlation and enrichment analyses were performed to elucidate relationships between protein biomarkers, obesity risk factors and perturbed pathways. Mendelian randomisation (MR) assessed causal links between proteins and obesity. RESULTS: A total of 16 proteins were identified as significantly associated with obesity through multivariable logistic regression in the basic model and subsequent Priority-Lasso analysis. Enrichment analyses highlighted immune response, lipid metabolism and inflammation regulation were linked to obesity. Machine learning models demonstrated robust predictive performance with area under the curves (AUC) of 0.820 (SVM), 0.805 (RF), 0.791 (KNN) and 0.819 (Adaboost). All 16 proteins correlated with obesity-related risk factors such as blood pressure and lipid levels. MR analysis identified AFM, CRP and CFH as causal and potentially modifiable proteins. CONCLUSIONS: The protein signatures identified in our study showed promising predictive potential for obesity. These findings warrant further investigation to evaluate their clinical applicability, offering insights into obesity prevention and treatment strategies.
Altmetric
Weitere Metriken?
[➜Einloggen]
Zusatzinfos bearbeiten [➜Einloggen]
Publikationstyp Artikel: Journalartikel
Dokumenttyp Wissenschaftlicher Artikel
Korrespondenzautor
Schlagwörter Cohort Study ; Machine Learning ; Obesity ; Proteomics; Body-mass Index; Mendelian Randomization
ISSN (print) / ISBN 1462-8902
e-ISSN 1463-1326
Zeitschrift Diabetes, Obesity and Metabolism
Quellenangaben Band: 27, Heft: 5, Seiten: 2626-2636 Artikelnummer: , Supplement: ,
Verlag Wiley
Verlagsort 111 River St, Hoboken 07030-5774, Nj Usa
Nichtpatentliteratur Publikationen
Begutachtungsstatus Peer reviewed
Institut(e) Institute of Epidemiology (EPI)
Institute of Experimental Genetics (IEG)
Förderungen German Federal Ministry of Education and Research (BMBF)
Bavarian State Ministry of Health, Care and Prevention
University Hospital of Augsburg
Helmholtz Zentrum Munchen - German Research Center for Environmental Health