Diabetic foot ulcer (DFU) is a prevalent and severely detrimental complication of diabetes affecting approximately 40 million people worldwide, leading to high mortality rate and heavy medical burden. Therefore, identifying modifiable and nonmodifiable risk factors for better prevention of DFU is in urgent need. Though risk factors of DFU have been studied extensively, current studies focused only on some specific aspects of risk factors, namely, sociodemographic factors, genetic factors, blood inflammation markers, lifestyle, etc. Thus, little is known about the comprehensive risk factors of DFU causing a limited clinical application of risk stratification system causing a lack of reliable prediction model.
The aim of this study is to use data from UK Biobank, the large population-based cohort to examine a comprehensive set of modifiable and nonmodifiable risk factors related to the incidence of diabetes foot ulcer and predict the incidence of DFU by machine learning. We would use Cox proportional hazards regression to analyze the association between risk factors and incident DFU with a manual stepwise forward manner and machine learning would be used to develop a prediction model.
We believe our prediction model of DFU would significantly propel further clinical researches on DFU prevention, the development of targeted preventive strategies, the improvement of clinical guidelines and the promotion of policies for DFU prevention.