Machine-learning based prediction model for acute kidney injury induced by multiple wasp stings

Abstract

Acute kidney injury (AKI) following multiple wasp stings is a severe complication with potentially poor outcomes. Despite extensive research on AKI's risk factors, predictive models for wasp sting-related AKI are limited. This study aims to develop and validate a machine learning-based clinical prediction model for AKI in individuals with wasp stings. In this retrospective cohort study, conducted at a tertiary teaching hospital in Yichang, China, from July 2013 to April 2023, 214 patients with wasp sting injuries were analyzed. Using least absolute shrinkage and selection operator (LASSO) regression and multivariate logistic regression, prognostic variables for AKI were identified. A nomogram incorporating these four variables was constructed. The model's performance was assessed through internal validation, leave-one-out cross-validation, net reclassification improvement (NRI), integrated discrimination improvement (IDI), and decision curve analysis (DCA). Among 214 patients affected by wasp stings, 34.6% (74/214) developed AKI. Following LASSO regression and multivariate logistic regression, the number of stings, presence of gross hematuria, systemic inflammatory response index (SIRI), and platelet count were identified as prognostic factors. A nomogram was constructed and evaluated for its predictive accuracy, showing an area under the curve (AUC) of 0.757 (95% CI 0.711 to 0.804) and a concordance index (C-index) of 0.75. Validation confirmed the model's reliability and superior discrimination ability over existing models, as demonstrated by NRI, IDI, and DCA. The developed nomogram effectively predicts AKI risk in wasp sting patients, facilitating early identification and management of those at risk.