Endocrine Abstracts

Universal reach of diabetes care and assurance of minimum standard of care is a milestone yet to be achieved. Application of artificial intelligence/machine learning (AI/ML) for diabetes care can help solve both these aspects of diabetes care.

Objective

Creating a clinical decision support system using AI/ML approach for predicting best anti-diabetes drug class to be introduced to help optimally manage glycaemic control in people living with diabetes mellitus type 2.

Methodology

Study was conducted at an Endocrinology clinic and data collected from electronic clinic management system. 15485 diabetes prescriptions of 4974 patients were accessed. A data subset of 1671 diabetes prescriptions with information on diabetes drugs, demographics (age, gender, body mass index), biochemical parameters (HbA1c, fasting blood glucose, creatinine) and patient clinical parameters (diabetes duration, compliance to diet/exercise/medications, hypoglycemia, contraindication to any drug, summary of patient self monitoring of blood glucose data, diabetes complications) was used in analysis. The patients in this data set had mean HbA1c of 7.3% ±1.3% (median 7.05%, 25th percentile 6.5%, 75th percentile 7.89%). For analysis, 67% of the dataset was used as a training set and 33% as a testing set. An input of patient variables (current diabetes medications, demographics, biochemical parameters and patient clinical parameters) were used to predict all diabetes drug classes to be prescribed. Random forest algorithms were used to create decision trees for all diabetes drugs. Accuracy for predicting use of each individual drug class is depicted in Table 1 and varied from 85% to 99.4%. Multi-drug accuracy considering that all drug predictions in a prescription need to be correct stands at 72%. Multi drug class accuracy in clinical application may be higher than this result, as in a lot of clinical scenarios, two more diabetes drugs may be used interchangeably.

Clinical Impact

This report presents a first positive step in developing a robust clinical decision support system to transform access and quality of diabetes care. Multi-drug accuracy is likely to improve further with time as the depth of the dataset increases over time.