Prediction model for gestational diabetes mellitus using the XG Boost machine learning algorithm

Development of a machine learning-based model for the prediction and progression of diabetic kidney disease: A single centred retrospective study

BACKGROUND

Diabetic kidney disease (DKD) is a diabetic microvascular complication often characterized by an unpredictable progression. Hence, early detection and recognition of patients vulnerable to progression is crucial.

OBJECTIVE

To develop a prediction model to identify the stages of DKD and the factors contributing to progression to each stage using machine learning.

METHODOLOGY

A retrospective study was conducted in a South Indian tertiary care hospital and collected the details of patients diagnosed with DKD from January 2017 to January 2022. Bayesian optimization-based machine learning techniques such as classification and regression were employed. The model was developed with the help of an optimization framework that effectively balances classification, prediction accuracy, and explainability.

RESULTS

Of the 311 patients diagnosed with DKD, 227 were selected for the study. A system for predicting DKD has been created for a patient dataset utilizing a variety of machine-learning approaches. The eXtreme gradient (XG) Boost method excelled, achieving 88.75% accuracy, 88.57% precision, 91.4% sensitivity,100% specificity, and 89.49% F1-score. An interpretable data-driven method highlights significant features for early DKD diagnosis. The best explainable prediction model uses the XG Boost classifier, revealing serum uric acid, urea, phosphorous, red blood cells, calcium, and absolute eosinophil count as the major predictors influencing the progression of DKD. In the case of regression models, the gradient boost regressor performed the best, with an R2 score of 0.97.

CONCLUSION

Machine learning algorithms can effectively predict the stages of DKD and thus help physicians in providing patients with personalized care at the right time.

Feasibility study of rehabilitation for cardiac patients aided by an artificial intelligence web-based programme: a randomised controlled trial (RECAP trial)-a study protocol

INTRODUCTION

Cardiac rehabilitation (CR) delivered by rehabilitation specialists in a healthcare setting is effective in improving functional capacity and reducing readmission rates after cardiac surgery. It is also associated with a reduction in cardiac mortality and recurrent myocardial infarction. This trial assesses the feasibility of a home-based CR programme delivered using a mobile application (app).

METHODS

The Rehabilitation through Exercise prescription for Cardiac patients using an Artificial intelligence web-based Programme (RECAP) randomised controlled feasibility trial is a single-centre prospective study, in which patients will be allocated on a 1:1 ratio to a home-based CR programme delivered using a mobile app with accelerometers or standard hospital-based rehabilitation classes. The home-based CR programme will employ artificial intelligence to prescribe exercise goals to the participants on a weekly basis. The trial will recruit 70 patients in total. The primary objectives are to evaluate participant recruitment and dropout rates, assess the feasibility of randomisation, determine acceptability to participants and staff, assess the rates of potential outcome measures and determine hospital resource allocation to inform the design of a larger randomised controlled trial for clinical efficacy and health economic evaluation. Secondary objectives include evaluation of health-related quality of life and 6 minute walk distance.

ETHICS AND DISSEMINATION

RECAP trial received a favourable outcome from the Berkshire research ethics committee in September 2022 (IRAS 315483).Trial results will be made available through publication in peer-reviewed journals and presented at relevant scientific meetings.

TRIAL REGISTRATION NUMBER

ISRCTN97352737.

Associations between a polygenic risk score and the risk of gestational diabetes mellitus in a Chinese population: a case-control study

Our objective was to construct a polygenic risk score (PRS) and assess its utility and effectiveness in predicting the risk of gestational diabetes mellitus (GDM) in a Chinese population. We performed a case-control study involving 638 patients with GDM and 1,062 healthy controls. Genotyping was conducted utilizing a genome-wide association study (GWAS), and a PRS was constructed. We identified 12 susceptibility loci that exhibited significant associations with the risk of GDM at a p-value threshold of ≤5.0 × 10-8, of which four loci were newly discovered. A higher PRS was associated with an increased risk of GDM (OR: 1.44; 95% CI: 1.03, 2.01 for the highest quartile compared to the lowest quartile). The PRS demonstrated a clear linear relationship with the fasting plasma glucose (FPG), 1-hour postprandial glucose (1hPG), and 2-hour postprandial glucose (2hPG) levels. The maximally adjusted β coefficients and their corresponding 95% CIs were 0.181 (0.041, 0.320) for FPG, 0.225 (0.103, 0.346) for 1hPG, and 0.172 (0.036, 0.307) for 2hPG. Among the genetic variants examined, TCF7L2 rs7903146 displayed the strongest association with GDM risk (logOR = 0.18, p = 2.37 × 10-19), followed by ADAMTSL1 rs10963767 (logOR = 0.14, p = 3.58 × 10-15). The areas under the curve (AUCs) was significantly increased from 0.703 (0.678, 0.728) in the traditional risk factor model to 0.765 (0.741, 0.788) by including PRS. These findings indicate that pregnant women with a higher PRS could potentially derive considerable advantages from the implementation of a feasible PRS-based GDM screening program aimed at delivering precision prevention strategies within Chinese populations.