A team of researchers developed a model to predict the probability of lower-limb amputation for hospitalized patients with diabetic foot ulcers. Their method showed a 77.8 accuracy rate.
The researchers used the National Inpatient Sample database to select patients with diabetic foot ulcers from 2008 to 2014 who were then further divided by major amputation status. International Classification of Diseases, Ninth Edition, Clinical Modification and Agency for Healthcare Research and Quality comorbidity codes were used to compare patient characteristics.
A total of 326,853 inpatients were identified, and 5.9 percent underwent major amputation. The top five contributory variables were gangrene, peripheral vascular disease, weight loss, systemic infection, and osteomyelitis.
The model performance of the training data was 77.7 percent and of the testing data was 77.8 percent. The researchers further validated the model with boosting and random forest models, which demonstrated similar performance.
The study, “A machine learning model for prediction of amputation in diabetics,” was published in the Journal of Diabetes Science and Technology.