A “One‐Stop Shop” Decision Tree for Diagnosing and Phenotyping Polycystic Ovarian Syndrome on Serum Metabolic Fingerprints

Abstract

Abstract Polycystic ovary syndrome (PCOS) is a common endocrine disease regulated by metabolic disorders, the effective intervention of which depends on diverse phenotypes (e.g., insulin resistance). Serum metabolic fingerprint (SMF) holds promise in characterizing the pathogenesis stress related to diseases; yet, PCOS diagnosis and phenotyping are time‐consuming and challenging due to the lack of an integrated metabolic tool. Here, a nanoparticle‐enhanced laser desorption/ionization mass spectrometry platform is introduced for one‐time serum metabolic fingerprinting and to identify the metabolic heterogeneity associated with obesity in PCOS patients. A decision tree based on the acquired SMFs is constructed, and real‐world simulations on independent internal and external cohorts are performed. The decision tree yields the area under the receiver operating characteristic curves (AUC) of 0.967 for PCOS diagnosis and AUC of 0.898 for phenotyping, respectively. The technical robustness of the “one‐stop shop” decision tree across laboratories is validated for clinical utility. The decision tree aims to improve PCOS management in comparison to clinical assessment, leading to a potential reduction in multiple blood tests and physician workload.