AUTHOR=Pan Hong , Sun Jijia , Luo Xin , Ai Heling , Zeng Jing , Shi Rong , Zhang An 

TITLE=A risk prediction model for type 2 diabetes mellitus complicated with retinopathy based on machine learning and its application in health management

JOURNAL=Frontiers in Medicine

VOLUME=Volume 10 - 2023

YEAR=2023

URL=https://www.frontiersin.org/journals/medicine/articles/10.3389/fmed.2023.1136653

DOI=10.3389/fmed.2023.1136653

ISSN=2296-858X

ABSTRACT=Objective: This study aimed to establish a risk prediction model using inspection indicators as little as possible for diabetic retinopathy (DR) in the Chinese type 2 diabetes mellitus (T2DM) population and to propose suggestions for chronic disease management.
Methods: This multi-centered retrospective cross-sectional study was conducted in 2,385 patients with T2DM. The predictors of the training set were respectively screened by extreme gradient boosting (XGBoost), a random forest recursive feature elimination (RF-RFE) algorithm, a backpropagation neural network (BPNN), and a least absolute shrinkage selection operator (LASSO) model. Based on the predictors repeated ≥3 times in the four screening methods, a prediction model called Model I was established through multivariate logistic regression analysis. To evaluate the effectiveness of the model, logistic regression model II built on the predictive factors in the previously released DR risk study are introduced to our current study. Nine evaluation indicators were used to compare the performance of the two prediction models, including the area under the receiver operating characteristic curve (AUROC), accuracy, precision, recall, F1 score, balanced accuracy, calibration curve, Hosmer-Lemeshow test, and Net Reclassification Index (NRI).
Results: Including predictors, such as glycosylated hemoglobin A1c, course of disease, postprandial blood glucose, age, systolic blood pressure, and albumin/urine creatinine ratio, multivariate logistic regression Model I demonstrated a better prediction ability than Model II. Model I revealed the highest AUROC (0.706), accuracy (0.802), precision (0.650), recall (0.038), F1 score (0.072), Hosmer-Lemeshow test (0.887), NRI (0.004), and balanced accuracy (0.514).
Conclusion: We have built a more accurate DR risk prediction model for patients with T2DM with fewer indicators, which can be used to effectively predict the individualized risk of DR patients in China. In addition, the model can provide powerful auxiliary technical support for the clinical and health management of patients with diabetes comorbidities.