Diabetic Foot Ulcer Prediction Using Machine Learning Algorithms-Based on Multimodal Sensors Measurements

Abstract

Diabetic foot ulcers (DFUs) are a complication of diabetes that affects the feet. This occurs due to poor blood circulation and nerve damage from high blood sugar levels. DFUs are characterized by ulcers that heal slowly and increase the risk of infection. Delayed detection may lead to amputation. Therefore, early detection of DFU is critical. Most previous studies have relied on single-sensor data or small datasets, which limits their applicability. This study addresses this gap by integrating multimodal sensor data with ML algorithms for DFU prediction. In this paper, we propose a DFU prediction system that incorporates multiple sensors in a specific wireless sensor network (WSN). The sensors measure foot temperature, pressure, and humidity to assess diabetic foot health conditions. ESP32 microcontrollers instantly process sensor data for remote communication with smartphones. The measurements collected from physiological sensors were integrated with machine learning (ML) algorithms to predict DFU. Four ML algorithms were used: decision tree (DT), random forest (RF), k-nearest neighbors (KNN), and support vector machine (SVM). These algorithms classify and predict whether feet are healthy or unhealthy using multimodal sensor data. The data was divided into 80% training and 20% testing sets, and results were validated using 10-fold cross-validation. The accuracy, recall, specificity, AUC, F1 scores, and precision were evaluated for the implemented algorithms. The DT algorithm showed the highest accuracy, achieving 99.0% compared to other algorithms in this research and earlier studies. The integration of WSN and ML has proven valuable in classifying and forecasting DFU, confirming that the proposed method can enhance current diagnostic methods.