Machine learning for assessing the role of immunoglobulin free light chains in the identification of circulatory diseases and survival prediction

Recent studies have highlighted the predictive value of combined Immunoglobulin Free Light Chain (FLC) levels in circulatory disease (CD), as well as other life-threatening conditions. Elevated FLC levels have been associated with poor outcomes, including increased mortality and hospital readmissions. In acute Heart Failure (HF), for example, patients had significantly higher FLC levels compared to those with stable HF or healthy controls. Moreover, FLC levels remained elevated long after clinical stabilization, suggesting potential ongoing immune disturbances, or impaired renal clearance. The present study investigated the use of Machine Learning (ML) techniques to predict the survival of patients with elevated FLC and to improve the understanding of circulatory mortality risks by identifying individuals whose deaths were attributable to circulatory system conditions. It was found that binary classification models could successfully determine patients with negative outcome (AUC > 0.80), and survival analysis models could achieve dynamic AUC higher than 0.75 in estimating the survival probability as a function of time. In the context of CD, elevated levels of FLCs have been shown to be associated with an increased risk of circulatory events. This association suggests that FLCs might play a role in the development of circulatory conditions.