Unsupervised Contrastive Representation Learning for Multimodal Clinical Stratification: Integrating Hemodynamic, Metabolic, and Proteomic Biomarkers Without Labels

Abstract

Background: The reliance on large-scale, annotated datasets remains a primary bottleneck in deploying artificial intelligence within healthcare. Traditional supervised learning requires expensive expert labeling, which is often unavailable for complex, multimodal clinical data. Objective: This study evaluates the efficacy of a self-supervised, contrastive learning framework designed to learn rich feature representations from unlabelled multimodal medical datasets. We specifically aim to stratify risk in two distinct pathological domains: cardiovascular hemodynamics and neuropsychiatric metabolic profiles. Methods: We aggregated retrospective data comprising hemodynamic parameters, serum anion gap trajectories, and lipidomic/proteomic profiles. A dual-stream contrastive neural network was trained using an InfoNCE loss function to maximize the similarity between augmented views of the same patient’s data while minimizing similarity to others. The model was evaluated on its ability to cluster latent representations associated with ICU mortality in aortic aneurysm patients and metabolic dysregulation in schizophrenia, without prior access to diagnostic labels. Results: The unsupervised model successfully identified high-risk clusters characterized by elevated serum anion gaps and hemodynamic instability, aligning with mortality outcomes in critically ill patients. Furthermore, in the neuropsychiatric cohort, the model distinguished schizophrenia phenotypes based on distinct membrane lipid homeostasis and proteomic signatures, differentiating them from neurodegenerative controls like Alzheimer’s disease. Conclusion: Contrastive approaches can effectively extract clinically relevant biomarkers from heterogeneous data without manual labels. This suggests that self-supervised learning is a viable pathway for discovering novel pathophysiological associations across hemodynamics, metabolism, and proteomics.

Keywords:

Hepatocellular Carcinoma, Inflammatory Bowel Disease, Geriatric Gastroenterology

References

Golledge J, Singh TP. Effect of blood pressure lowering drugs and antibiotics on abdominal aortic aneurysm growth: a systematic review and meta-analysis. Heart. 2021;107(18):1465-1471. doi:10.1136/heartjnl-2020-318192

Yang DE, Jo S, Lee DH, An WS, Jeong MJ, Son M. Dynamics of serum anion gaps with in-hospital mortality: Analysis of the multi-open databases. Ceruti S, ed. PLoS ONE. 2024;19(4):e0302206. doi:10.1371/journal.pone.0302206

Xu C, Sun L, Dong M, et al. Serum Anion Gap is Associated with Risk of All-Cause Mortality in Critically Ill Patients with Acute Myocardial Infarction. IJGM. 2022;Volume 15:223-231. doi:10.2147/IJGM.S336701

Ji X, Peng S. The association between serum anion gap and all‐cause mortality of unselected adult patients: A retrospective cohort study of >20,000 patients. Clinical Laboratory Analysis. 2023;37(1):e24818. doi:10.1002/jcla.24818

Chen Q, Chen Q, Li L, et al. Serum anion gap on admission predicts intensive care unit mortality in patients with aortic aneurysm. Exp Ther Med. Published online July 2, 2018. doi:10.3892/etm.2018.6391

Tan BKJ, Teo YH, Tan NKW, et al. Association of obstructive sleep apnea and nocturnal hypoxemia with all-cancer incidence and mortality: a systematic review and meta-analysis. Journal of Clinical Sleep Medicine. 2022;18(5):1427-1440. doi:10.5664/jcsm.9772

Roozeman JP, Mazzinari G, Serpa Neto A, et al. Prognostication using SpO2/FiO2 in invasively ventilated ICU patients with ARDS due to COVID-19 – Insights from the PRoVENT-COVID study. Journal of Critical Care. 2022;68:31-37. doi:10.1016/j.jcrc.2021.11.009

Giordano S, Biancari F, Loponen P, Wistbacka JO, Luther M. Preoperative haemodynamic parameters and the immediate outcome after open repair of ruptured abdominal aortic aneurysms. Interactive CardioVascular and Thoracic Surgery. 2009;9(3):491-493. doi:10.1510/icvts.2009.205930

Brunker LB, Boncyk CS, Rengel KF, Hughes CG. Elderly Patients and Management in Intensive Care Units (ICU): Clinical Challenges. CIA. 2023;Volume 18:93-112. doi:10.2147/CIA.S365968

Li M, Gao Y, Wang D, Hu X, Jiang J, Qing Y, et al. Impaired Membrane Lipid Homeostasis in Schizophrenia. Schizophr Bull. 2022; 48: 1125-35.

Maas DA, Martens MB, Priovoulos N, Zuure WA, Homberg JR, Nait-Oumesmar B, et al. Key role for lipids in cognitive symptoms of schizophrenia. Transl Psychiatry. 2020; 10: 399.

Pillinger T, McCutcheon RA, Vano L, Mizuno Y, Arumuham A, Hindley G, et al. Comparative effects of 18 antipsychotics on metabolic function in patients with schizophrenia, predictors of metabolic dysregulation, and association with psychopathology: a systematic review and network meta-analysis. Lancet Psychiatry. 2020; 7: 64-77.

Oyagi A, Hara H. Essential roles of heparin-binding epidermal growth factor-like growth factor in the brain. CNS Neurosci Ther. 2012; 18: 803-10.

Rodrigues-Amorim D, Rivera-Baltanas T, Vallejo-Curto MDC, Rodriguez-Jamardo C, de Las Heras E, Barreiro-Villar C, et al. Proteomics in Schizophrenia: A Gateway to Discover Potential Biomarkers of Psychoneuroimmune Pathways. Front Psychiatry. 2019; 10: 885.

Bastrup J, Hansen KH, Poulsen TBG, Kastaniegaard K, Asuni AA, Christensen S, et al. Anti-Abeta Antibody Aducanumab Regulates the Proteome of Senile Plaques and Closely Surrounding Tissue in a Transgenic Mouse Model of Alzheimer’s Disease. J Alzheimers Dis. 2021; 79: 249-65.

Coimbra JRM, Moreira PI, Santos AE, Salvador JAR. Therapeutic potential of glutaminyl cyclases: Current status and emerging trends. Drug Discov Today. 2023; 28: 103644.

Haque A, Alenezi KM, Abdul Rasheed MSM. Identification of imidazole-based small molecules to combat cognitive disability caused by Alzheimer’s disease: A molecular docking and MD simulations based approach. Comput Biol Chem. 2024; 112: 108152.

Ly M, Yu GZ, Mian A, Cramer A, Meysami S, Merrill DA, et al. Neuroinflammation: A Modifiable Pathway Linking Obesity, Alzheimer’s disease, and Depression. Am J Geriatr Psychiatry. 2023; 31: 853-66.

Campeau A, Mills RH, Stevens T, Rossitto LA, Meehan M, Dorrestein P, et al. Multi-omics of human plasma reveals molecular features of dysregulated inflammation and accelerated aging in schizophrenia. Mol Psychiatry. 2022; 27: 1217-25.