JMIR Publications

ABSTRACT

Background:

Gestational diabetes mellitus (GDM) is a common complication during pregnancy, with its incidence increasing year by year. It poses numerous adverse health effects on both mothers and newborns. Accurate prediction of GDM can significantly improve patient prognosis. In recent years, artificial intelligence (AI) algorithms have been increasingly used in the construction of GDM prediction models. However, there is still no consensus on the most effective algorithm or model.

Objective:

To evaluate and compare the performance of existing GDM prediction models constructed using AI algorithms and propose strategies for enhancing model generalizability and predictive accuracy, thereby providing evidence-based insights for the development of more accurate and effective GDM prediction models.

Methods:

A comprehensive search was conducted across PubMed, Web of Science, Cochrane Library, EMBASE, Scopus, and OVID, covering publications from the inception of databases to June 1, 2025, to identify studies that developed or validated GDM prediction models based on AI algorithms. A bivariate mixed-effects model (MIDAS) was employed to summarize sensitivity and specificity, and to generate a summary receiver operating characteristic (SROC) curve, calculating area under the curve (AUC). And the Hartung-Knapp-Sidik-Jonkman (HKSJ) method was further employed to adjust for the pooled sensitivity and specificity. Between-study standard deviation (τ) and variance (τ²) were extracted from the bivariate model to quantify absolute heterogeneity. Deeks’ test was used to evaluate small-study effects among the included studies. Additionally, subgroup analysis and meta-regression were conducted to compare the performance differences among algorithms and to explore the sources of heterogeneity.

Results:

Fourteen studies (involving 96,020 patients) reported on the predictive value for AI algorithms for GDM. After adjustment with the HKSJ method, the pooled sensitivity and specificity were 0.72 (95%CI 0.67-0.77; τ=0.15, τ2=0.02) and 0.81 (95%CI 0.76-0.85; τ=0.12, τ2=0.02), respectively. The SROC curve showed that the AUC for predicting GDM using AI algorithms was 0.87 (95%CI 0.84-0.90), indicating a strong predictive capability. Deeks’ test (P=0.01) and the funnel plot both showed clear asymmetry, suggesting the presence of small-study effects. Subgroup analysis showed that random forest (RF) algorithm exhibited the highest sensitivity (0.97, 95%CI 0.38-1.00), while the extreme gradient boosting (XGBoost) algorithm exhibited the highest specificity (0.96, 95%CI 0.77-0.99). And meta-regression further revealed an evaluation in predictive accuracy in prospective study designs (regression coefficient=2.289, P=0.001).

Conclusions:

This study evaluated prediction models constructed using various AI algorithms, and the results demonstrated that AI algorithms exhibited strong performance in predicting GDM. However, many existing models still have limitations. Therefore, future research should standardize model development, optimize model performance, adopt appropriate study designs, and explore how to better integrate predictive models into clinical practice. Clinical Trial: PROSPERO CRD42025645913; https://www.crd.york.ac.uk/PROSPERO/view/CRD42025645913