Risk Analysis of Diabetic Leg Amputation : A Systematic Study

Abstract

Diabetic Foot Ulcer is considered a critical complication of diabetes, characterized by injuries and frequent exposure of the diabetic patient’s foot. Approximately 20 % of diabetic patients may develop foot ulcers, with around 10 % requiring hospitalization due to additional complications. Typically, these ulcers affect individuals who have had diabetes for more than ten years. Neglecting or leaving Diabetic Foot Ulcers untreated can result in severe damage, leading to worsened infections and potentially necessitating amputation, often accompanied by multiple complications that may even result in mortality. Therefore, early prediction of foot-threatening risks is crucial to prevent worsening situations. In this work visualization methods are applied for a better understanding of the dataset to draw meaningful insights and to observe the behavior of amputation risks in diabetic patients. The feature values fluctuate, so selecting the best feature from a combination of statistical and graphical data analysis is not trivial. Data visualization techniques (data-driven approach), and statistical analysis were used to select important features, that lead to leg amputation. The Machine learning models were implemented to forecast foot ulcers depending on clinical outcomes. A predicted accuracy of 85 % is observed using Ensemble Methods.