Prediction of Diabetic Sensorimotor Polyneuropathy Using Machine Learning Techniques

Abstract

Diabetic sensorimotor polyneuropathy (DSPN) is a major complication in patients with diabetes mellitus (DM), and early detection or prediction of DSPN is important for preventing or managing neuropathic pain and foot ulcer. Our aim is to delineate whether machine learning techniques are more useful than traditional statistical methods for predicting DSPN in DM patients. Four hundred seventy DM patients were classified into four groups (normal, possible, probable, and confirmed) based on clinical and electrophysiological findings of suspected DSPN. Three ML methods, XGBoost (XGB), support vector machine (SVM), and random forest (RF), and their combinations were used for analysis. RF showed the best area under the receiver operator characteristic curve (AUC, 0.8250) for differentiating between two categories—criteria by clinical findings (normal, possible, and probable groups) and those by electrophysiological findings (confirmed group)—and the result was superior to that of linear regression analysis (AUC = 0.6620). Average values of serum glucose, International Federation of Clinical Chemistry (IFCC), HbA1c, and albumin levels were identified as the four most important predictors of DSPN. In conclusion, machine learning techniques, especially RF, can predict DSPN in DM patients effectively, and electrophysiological analysis is important for identifying DSPN.

Prediction of the risk for diabetic neuropathy among diabetes mellitus patients at selected referral hospitals, in Amhara regional state Northwest Ethiopia, January, 2005- December 2021; Development and validation of risk prediction model

“…The performance of our model was found to be consistent with other prediction models developed to predict diabetic neuropathy among DM patients, using hypertension, age, heart rate and BMI as predictors, AUC; 71% in china(69) using hypertension, comorbidities, gender, age, obesity, abnormal triglycerides as predictors, AUC; 75% in china (70),but better than a study using glomerular filtration rate, glibenclamide and creatinine as predictors AUC=66.05% in Mexico (71), using age, FBG, PBG, HbA1c, LDL, HDL and BMI as predictors AUC; 55.6% in china(72), using FBG, BMI, age as predictors AUC; 63.50 in Korea (73). This might be due to difference in study participants involved through socio-demographic characteristics, difference number of predictors used in the model development (74).…”

Addressing Chemotherapy-Induced Peripheral Neuropathy Using Multi-Frequency Vibrometry and Patient-Reported Outcomes

“…Here, several baseline risk factors such as baseline blood samples (hemoglobin, micronutrients, neutrophil-to-lymphocyte ratio, etc.) and patient risk factors [ 42 ] could be analyzed together with semiquantitative measures of high correlation with CIPN outcomes (such as MF-V VPTs) to produce a more accurate prediction for each patient by following prior examples where machine learning has been utilized to predict the risk of vincrinstine-induced peripheral neuropathy [ 43 ] and diabetic polyneuropathy [ 44 ].…”

Recent applications of machine learning and deep learning models in the prediction, diagnosis, and management of diabetes: a comprehensive review

“…Diabetic Sensorimotor Polyneuropathy (DSPN) is a remarkable consequence of diabetes mellitus, so early diagnosis or prediction of DSPN is essential for preventing foot ulcers and neuropathic pain [ 62 ]. Three machine learning methods including SVM, XGBoost, RF, and their combinations were considered to predict four classes containing normal, possible, probable, and confirmed based on the electrophysiological and clinical characteristics of the doubtful DSPN.…”

“…DT and LR [65] 2970 youth aged 12-19 years (NHANES dataset) LR, LogitBoost, and decision tree [66] 746 subjects SVM, XGBoost, RF, and their combinations [62] GDM 22,242 singleton pregnancies (3182 women developed GDM) RF, logistic, decision tree, XGB, GDBT, LGB, AdaBoost, Vote, logistic regression with RCS and stepwise logistic regression [75] 490 pregnant women, 215 with GDM and 275 controls SVM and light gradient boosting machine (lightGBM) [76] 588,622 pregnancies from 368,351 women Gradient-boosting machine model constructed by decision-tree base-learners [74] 4378 cases CSHM, BN, LR, CHAID tree, SVM, and NN [77] 152 women AIRS [80] 4771 pregnant women in early gestation Multivariate Bayesian logistic regression using Markov Chain Monte Carlo simulation algorithm [81] All types of Diabetes 2001 cases with diabetes (Kaggle dataset)…”