AUTHOR=Fan Yuting , Long Enwu , Cai Lulu , Cao Qiyuan , Wu Xingwei , Tong Rongsheng 

TITLE=Machine Learning Approaches to Predict Risks of Diabetic Complications and Poor Glycemic Control in Nonadherent Type 2 Diabetes

JOURNAL=Frontiers in Pharmacology

VOLUME=Volume 12 - 2021

YEAR=2021

URL=https://www.frontiersin.org/journals/pharmacology/articles/10.3389/fphar.2021.665951

DOI=10.3389/fphar.2021.665951

ISSN=1663-9812

ABSTRACT=Purpose: The objective of this study was to evaluate the efficacy of Machine Learning algorithms in predicting risks of complications and poor glycemic control among non-adherent Type 2 Diabetes (T2D). 
Materials and Methods: Data in this study were obtained through face to face investigation and the Electronic Health Medical Record System (EHRS). The T2D who had neither been monitored for Glycosylated Hemoglobin A nor had changed their hyperglycemia treatment regimens within recent 12 months was the object of this study. Seven types of Machine Learning algorithms were utilized to develop 18 prediction models. The predictive performance was mainly assessed by area under the curve (AUC) of the testing set. 
Results: Of 800 T2Ds, 165 (20.6%) met the inclusion criteria and 129 (78.2%) patients had poor glycemic control (defined as Glycosylated Hemoglobin A >7%). The highest AUCs of the testing set for diabetic nephropathy, diabetic peripheral neuropathy, diabetic angiopathy, diabetic eye disease, and Glycosylated Hemoglobin A were 0.902±0.040, 0.859±0.050, 0.889±0.059, 0.832±0.086, and 0.825±0.092, respectively. 
Conclusion: This was the first study to use Machine Learning algorithms exploring the potential adverse outcomes of non-adherent T2D. The performances of the final prediction models were acceptable, and those models have potential clinical applicability in improving T2D care.