Exploration of a Multimodal Machine Learning Model Integrating Ultrasound and Clinical Indicators for the Diagnosis of Diabetic Peripheral Neuropathy

The authors aimed to develop and evaluate multimodal machine learning models that integrate ultrasound and clinical indicators for diagnosing diabetic peripheral neuropathy (DPN). Among the four models tested, the random forest (RF) model exhibited the best performance, achieving an AUC of 0.852, and identified key risk factors for DPN, indicating its significant potential for clinical application in risk assessment.