Establishing a clinical prediction model for diabetic foot ulcers in type 2 diabetic patients with lower extremity arteriosclerotic occlusion using machine learning

Le résumé fourni par la source

The burden of diabetic foot ulcers (DFU) is exacerbated in diabetic patients with concomitant arteriosclerotic occlusion disease (ASO) in the lower extremities, who experience more severe symptoms and poorer prognoses. The study aims to develop a predictive model grounded in machine learning (ML) algorithms, specifically tailored to forecast the occurrence of DFU in diabetic patients with lower extremity ASO. The study involves the data from diabetic patients diagnosed with lower extremity ASO from January 1, 2011 to August 31, 2023. We conducted quality control on the data. Subsequently, the dataset was divided into a training set comprising patients before 2020 and a validation set comprising patients in 2020 and onwards. Patients were stratified into the DFU group or the non-DFU group based on the occurrence of DFU. Intergroup comparisons were conducted to analyze the differences between these two groups. Logistic regression analyses, 3 kinds of machine learning algorithms, a predictive model and nomogram was formulated to estimate the risk of DFU occurrence among diabetic patients with lower extremity ASO. Internal validation of the model was undertaken using the bootstrap method, combing with external temporal validation, with the results visually presented through the Receiver Operating Characteristic (ROC) curve and the Calibration curve. To evaluate the clinical practicality of the model, Decision Curve Analysis (DCA) and Clinical Impact Curve (CIC) were employed. Body Mass Index (BMI), hypertension, coronary heart disease, diabetic nephropathy, the number of lower leg artery occlusions, controlling glucose by insulin injection, age, number of cigarettes smoked per day, diastolic blood pressure, and C-reactive protein (CRP) were utilized to construct a clinical prediction model. This model exhibited a high predictive performance (AUC = 0.962), and the results from both internal validation and external temporal validation further confirmed its high accuracy and reproducibility (AUC = 0.968 and AUC = 0.977, respectively). Additionally, DCA and CIC demonstrated the high clinical practicality of this model. The clinical prediction model exhibited excellent accuracy and reproducibility, along with broad clinical practicality. It provides a good reference for the diagnosis and treatment of DFU.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.