Recent Advances in the Diagnosis and Management of Gestational Diabetes Mellitus: A Narrative Review

Abstract

Gestational diabetes mellitus (GDM) is one of the most common metabolic disorders of pregnancy and is associated with adverse short- and long-term health outcomes in both mothers and offspring. Its increasing global prevalence and substantial heterogeneity in pathogenesis and clinical presentation underscore the limitations of conventional glucose-centered approaches to risk assessment and management. Although the oral glucose tolerance test (OGTT) remains the established diagnostic standard, its relatively late diagnostic window and limited ability to predict disease development have stimulated interest in novel strategies for early risk stratification and individualized care. Importantly, emerging approaches are not yet sufficiently validated to replace the OGTT in routine clinical practice. This narrative review summarizes recent advances in GDM diagnosis and management, focusing on emerging biomarkers, multi-omics technologies, artificial intelligence (AI), and precision medicine. We examine the potential roles of inflammatory biomarkers, adipokines, metabolomic signatures, gut microbiota, and genomic and transcriptomic profiles in the early prediction of GDM, as well as AI-based models integrating multidimensional clinical and molecular data. Advances in precision nutrition, continuous glucose monitoring, pharmacological optimization, digital health technologies, and microbiota-targeted interventions are also discussed. Particular attention is given to the distinction between clinically established strategies, promising approaches supported by emerging evidence, and exploratory technologies that remain investigational. Despite substantial progress, important challenges remain, including heterogeneity in diagnostic criteria, limited reproducibility and clinical translation of multi-omics findings, insufficient external validation of AI models, and barriers to the implementation of individualized management. Key evidence gaps include the need for prospective and external validation, cost-effectiveness assessments, regulatory evaluation, and real-world implementation studies. Precision medicine in GDM should therefore be viewed as an evolving framework that integrates molecular, metabolic, clinical, and environmental information to improve risk stratification and guide individualized management rather than as a replacement for established diagnostic standards. Future GDM care is likely to move toward integrated precision medicine models combining multi-omics profiling, AI-assisted decision support, continuous metabolic monitoring, and personalized interventions, with the ultimate goal of improving maternal and offspring health across the life course.

Introduction

Gestational diabetes mellitus (GDM) is defined as glucose intolerance of varying severity that is first recognized or develops during pregnancy and represents one of the most common metabolic disorders affecting pregnancy worldwide. In recent decades, the global prevalence of GDM has continued to increase, driven by rising obesity rates, delayed childbearing age, and changes in lifestyle patterns. GDM has therefore emerged as a major public health concern with substantial implications for maternal and offspring health. The estimated global prevalence of GDM ranges from approximately 5% to 20%; however, considerable heterogeneity exists among populations due to differences in ethnic characteristics, screening strategies, and diagnostic criteria across regions.¹ Beyond its immediate impact on maternal metabolic homeostasis during pregnancy, GDM is strongly associated with an increased risk of hypertensive disorders of pregnancy, cesarean delivery, postpartum type 2 diabetes mellitus (T2DM), and cardiovascular disease later in life.² Furthermore, offspring exposed to an intrauterine hyperglycemic environment exhibit an increased susceptibility to obesity, insulin resistance, and metabolic syndrome, suggesting that GDM may exert long-term intergenerational metabolic consequences.³

Currently, the diagnosis of GDM primarily relies on the oral glucose tolerance test (OGTT), which is generally performed between 24 and 28 weeks of gestation. Several international organizations, including the International Association of Diabetes and Pregnancy Study Groups (IADPSG), the World Health Organization (WHO), and the American College of Obstetricians and Gynecologists (ACOG), have proposed different diagnostic criteria for GDM.^4–6 Although OGTT remains the cornerstone of clinical diagnosis, its application is associated with several limitations, including a relatively late diagnostic window, poor patient compliance, limited reproducibility, and insufficient ability to predict individual disease risk before clinical manifestation.⁷ Therefore, identifying novel biomarkers capable of detecting high-risk individuals during early pregnancy and establishing more precise diagnostic frameworks have become critical priorities in current GDM research.

With rapid advances in molecular biology, high-throughput sequencing technologies, and artificial intelligence (AI), the diagnostic paradigm of GDM is gradually shifting from conventional glucose-based assessment toward multidimensional risk prediction approaches. Increasing evidence indicates that inflammatory mediators, adipokines, metabolomic signatures, and alterations in the gut microbiota contribute significantly to the initiation and progression of GDM and may serve as potential early predictive indicators.⁸ For instance, decreased adiponectin levels, dysregulated leptin expression, and elevated inflammatory cytokines, including interleukin-6 (IL-6) and tumor necrosis factor-α (TNF-α), have been closely associated with pregnancy-related insulin resistance and impaired pancreatic β-cell function.^9–11 In addition, metabolomic studies have demonstrated that women with GDM exhibit early alterations in amino acid metabolism, lipid metabolism, and energy metabolism during the first trimester, providing new opportunities for early disease prediction.¹² More recently, the gut microbiota–metabolism axis has emerged as an important area of investigation in GDM. Dysbiosis of the intestinal microbiome may contribute to GDM development by modulating short-chain fatty acid production, systemic inflammation, and insulin sensitivity.¹³

Parallel to advances in diagnosis, therapeutic strategies for GDM are also undergoing substantial transformation. Conventional management primarily includes medical nutritional therapy, physical activity interventions, and insulin therapy when glycemic targets cannot be achieved. However, substantial interindividual variations in genetic background, metabolic phenotype, and lifestyle factors limit the effectiveness of uniform treatment strategies and highlight the need for personalized management approaches.¹⁴ Recently, the development of continuous glucose monitoring (CGM), digital health technologies, and AI-assisted clinical decision-making systems has provided new opportunities for dynamic glucose assessment and individualized intervention. CGM offers comprehensive insights into glucose variability beyond conventional glucose measurements and enables the application of novel metrics, such as time in range (TIR), to optimize glycemic management during pregnancy.¹⁵ Furthermore, machine learning algorithms integrating clinical characteristics, circulating biomarkers, and multi-omics datasets have shown promise in developing more accurate GDM risk prediction models, facilitating early screening and personalized prevention strategies.¹⁶

Despite substantial advances in the diagnosis and management of gestational diabetes mellitus (GDM), several important challenges remain, including the lack of globally harmonized diagnostic criteria, insufficient large-scale validation of emerging biomarkers, limited clinical translation and reproducibility of multi-omics findings, and inadequate interpretability and external validation of artificial intelligence (AI)-based prediction models. Addressing these gaps is essential for translating emerging precision medicine approaches into clinically meaningful strategies for GDM prevention, risk stratification, and management. We searched PubMed, Web of Science, Scopus, and Embase for literature published between July 2020 and July 2026 using combinations of terms related to “gestational diabetes mellitus”, “biomarkers”, “multi-omics”, “artificial intelligence”, “precision medicine”, “continuous glucose monitoring”, “digital health”, and related concepts. We considered original research articles, systematic reviews and meta-analyses, and relevant international clinical guidelines, prioritizing recent high-quality evidence and contemporary recommendations. This narrative review critically examines advances in GDM diagnosis and management, focusing on emerging biomarkers, multi-omics technologies, AI-based prediction models, continuous glucose monitoring (CGM), digital health technologies, and personalized therapeutic strategies (Figure 1). We assess the current evidence, clinical applicability, and limitations of these approaches, with particular emphasis on distinguishing clinically established strategies from promising approaches supported by emerging evidence and exploratory technologies that remain investigational. Finally, we discuss future directions for integrating molecular, metabolic, clinical, and environmental information into precision medicine frameworks to enable more individualized risk assessment and management of GDM.

Figure 1.

Evolution of GDM diagnosis and management toward precision medicine.

Recent Advances in Diagnostic Strategies for Gestational Diabetes Mellitus

The diagnostic strategies for gestational diabetes mellitus (GDM) are undergoing a paradigm shift from conventional glucose-based assessment toward precision risk prediction approaches. Currently, the oral glucose tolerance test (OGTT) remains the primary clinical method for diagnosing GDM; however, its limitations include a relatively late diagnostic window, inability to achieve early risk prediction, and insufficient capacity for individualized assessment. With increasing understanding of the complex pathophysiological mechanisms underlying GDM, emerging approaches involving inflammatory profiling, metabolic signatures, gut microbiome characterization, and artificial intelligence (AI)-based prediction models have attracted growing attention as promising strategies for improving early diagnosis and personalized risk stratification.

Evolution and Limitations of Conventional Diagnostic Criteria

Currently, OGTT remains the cornerstone for clinical diagnosis of GDM. According to recommendations from the International Association of Diabetes and Pregnancy Study Groups (IADPSG) and the World Health Organization (WHO), pregnant women typically undergo a 75-g OGTT between 24 and 28 weeks of gestation, with fasting, 1-h, and 2-h plasma glucose concentrations used to determine the presence of abnormal glucose metabolism.¹⁷ To facilitate comparison among currently adopted international guidelines, the major diagnostic criteria for GDM proposed by different professional organizations are summarized in Table 1. Despite sharing the common objective of identifying hyperglycemia during pregnancy, these diagnostic strategies differ considerably in screening approaches, glucose thresholds, and clinical applicability, contributing to substantial variations in GDM prevalence and management across populations. In addition, organizations such as the American Diabetes Association (ADA) and the American College of Obstetricians and Gynecologists (ACOG) have proposed alternative diagnostic criteria; however, discrepancies among different guidelines remain substantial.¹⁸ The widespread implementation of OGTT has significantly improved the standardization of GDM diagnosis and clinical management. Nevertheless, several limitations have gradually become evident. First, OGTT is generally performed during the second trimester, when many women may have already developed significant insulin resistance and metabolic disturbances, thereby reducing opportunities for early preventive intervention.¹⁹ Second, OGTT requires strict fasting conditions and multiple blood sampling procedures, which may negatively affect patient compliance and test reproducibility.²⁰ Furthermore, variations in diagnostic thresholds among different guidelines contribute to considerable differences in reported GDM prevalence, complicating the establishment of universally applicable management strategies.²¹ More importantly, GDM represents a complex metabolic disorder involving multiple pathological processes beyond hyperglycemia, including chronic low-grade inflammation, lipid metabolism abnormalities, pancreatic β-cell dysfunction, and alterations in gut microbial composition. Therefore, glucose-based diagnostic approaches alone may not fully capture individual disease susceptibility and metabolic risk. Consequently, recent research efforts have increasingly shifted from identifying established disease to predicting disease development, thereby driving the emergence of novel diagnostic strategies.

Table 1.

Comparison of International Diagnostic Criteria for GDM

Guideline	Screening Strategy	OGTT	Diagnostic Thresholds	Advantages	Limitations
IADPSG	One-step	75 g	FPG≥5.1 mmol/L 1 h≥10.0 2 h≥8.5	High sensitivity	High prevalence
WHO	One-step	75 g	Same as IADPSG	International acceptance	Increased healthcare burden
ADA	One-step/Two-step	75 g or 100 g	One-step (75 g) or two-step (100 g); ADA-specific cut-offs	Widely adopted	Regional variation
ACOG	Two-step	50 g+100 g	Carpenter-Coustan	Two-step approach may reduce false-positive diagnoses	May miss some cases of mild GDM
NICE	Risk-based	75 g	Different cut-offs	Cost-effective	Lower sensitivity

Application of Emerging Biomarkers for Early Prediction of GDM

In recent years, numerous biomarkers associated with the development of gestational diabetes mellitus (GDM) have been investigated, offering potential opportunities for early risk prediction during pregnancy. Given the multifactorial pathophysiology of GDM, biomarker research has increasingly expanded beyond conventional glucose measurements toward multidimensional molecular profiling. Among the candidate predictors, inflammatory biomarkers have been extensively studied. Although a tightly regulated inflammatory response is an essential component of normal pregnancy and maternal–fetal immune adaptation, excessive or dysregulated inflammation may contribute to the development of insulin resistance and metabolic dysfunction. Previous studies have reported that elevated levels of C-reactive protein (CRP), tumor necrosis factor-α (TNF-α), and interleukin-6 (IL-6) in early pregnancy are associated with an increased risk of subsequent GDM development.^22,23 These inflammatory mediators may contribute to GDM pathogenesis by interfering with insulin signaling, promoting adipose tissue inflammation, and disrupting glucose homeostasis. Adipokines have also attracted considerable attention as potential predictors of GDM. Adiponectin, an adipose-derived hormone with insulin-sensitizing and anti-inflammatory effects, is consistently lower in women who subsequently develop GDM and has been associated with the severity of insulin resistance.²⁴ In contrast, elevated leptin concentrations may be associated with enhanced inflammatory activity and impaired pancreatic β-cell function, potentially increasing susceptibility to GDM.²⁵ More recently, the gut microbiome has emerged as another potential source of biomarkers for early risk stratification. Increasing evidence suggests that women with GDM exhibit alterations in gut microbial composition and diversity, including reduced abundance of potentially beneficial taxa such as Akkermansia and enrichment of inflammation-associated microbial taxa. These microbial alterations may influence host insulin sensitivity through effects on short-chain fatty acid metabolism, intestinal barrier function, and systemic inflammatory signaling.^26,27 However, the predictive performance of individual biomarkers remains insufficient for routine clinical application, partly because of heterogeneity in study populations, sampling time points, analytical platforms, and diagnostic definitions. Recent prospective diagnostic accuracy studies suggest that inflammatory and metabolic indices readily available during the first trimester may provide additional information for early risk stratification; nevertheless, their standalone predictive performance remains modest, underscoring the need for cautious interpretation of initially promising findings.²⁸ Future approaches will therefore likely involve multimarker models that integrate inflammatory markers, adipokines, microbiome-derived signatures, and established clinical characteristics to improve the accuracy, robustness, and clinical utility of GDM risk prediction.

Multi-Omics Technologies Driving Precision Diagnosis of GDM

Recent advances in multi-omics technologies have substantially accelerated the development of precision diagnostic approaches for GDM. Compared with traditional single-marker measurements, multi-omics strategies enable comprehensive characterization of molecular alterations involved in disease initiation and progression, thereby facilitating the discovery of novel predictive signatures. Metabolomics represents one of the most widely applied approaches in GDM research. Emerging evidence suggests that metabolic abnormalities may occur before clinical diagnosis, including alterations in branched-chain amino acid (BCAA) metabolism, lipid metabolism, and bile acid pathways.^29–31 These metabolic perturbations may contribute to insulin resistance development and serve as early indicators of GDM susceptibility. In addition, genomic, transcriptomic, and proteomic approaches are increasingly being incorporated into GDM risk assessment. Genetic susceptibility factors may influence pancreatic β-cell function and insulin sensitivity, whereas placental transcriptomic alterations may reflect impaired maternal–fetal metabolic adaptation. With the advancement of multi-omics integration strategies, future GDM diagnosis may transition from single-parameter evaluation toward molecular-based precision classification systems. Nevertheless, several challenges currently limit the clinical translation of multi-omics approaches, including insufficient sample sizes, lack of standardized analytical pipelines, and limited validation in large-scale prospective cohorts. Therefore, further multicenter studies are required to establish clinically applicable multi-omics-based prediction models.

Artificial Intelligence and Machine Learning-Assisted Risk Prediction for GDM

The rapid development of artificial intelligence (AI) and machine learning (ML) has created new opportunities for early GDM risk prediction and individualized risk stratification. Conventional risk assessment primarily relies on readily available clinical characteristics, such as maternal age, body mass index (BMI), parity and previous pregnancy history, family history, and baseline glucose levels. In contrast, AI and ML approaches can integrate multidimensional clinical, laboratory, biological, and multi-omics data to identify complex and potentially nonlinear patterns across high-dimensional variables. Commonly investigated approaches for GDM prediction include logistic regression as a conventional statistical benchmark, as well as random forest, support vector machine (SVM), extreme gradient boosting (XGBoost), and deep learning models.^32–34 By combining maternal characteristics, laboratory parameters, and biomarker profiles obtained during early pregnancy, these models may improve risk estimation and facilitate the identification of women at increased risk of developing GDM. Moreover, the integration of AI with metabolomic profiles, gut microbiome data, and electronic health records (EHRs) may enable more comprehensive risk modeling and support the development of precision medicine-based approaches to GDM management. Together with emerging biomarkers and multi-omics technologies, AI-based prediction models are expanding GDM risk assessment beyond conventional glucose measurements toward multidimensional and individualized prediction. Representative biomarkers, their biological functions, and potential clinical applications are summarized in Table 2, illustrating the transition from single-parameter assessment toward integrated risk prediction within a precision medicine framework. However, most published AI models have been developed using single-center cohorts, relatively small datasets, or retrospective study designs and have not undergone rigorous external validation. In addition, concerns regarding model interpretability, calibration, reproducibility, data heterogeneity, and clinical implementation remain. Therefore, despite their promising predictive performance, current AI-based models should be considered investigational tools rather than clinically validated diagnostic instruments. Prospective multicenter studies, independent external validation, standardized reporting, and evaluation of clinical utility are required before these models can be reliably integrated into routine GDM risk assessment and management.

Table 2.

Emerging Biomarkers and Multi-Omics Approaches for Early Prediction of Gestational Diabetes Mellitus

Category	Biomarkers	Biological Significance	Clinical Application	Level of Evidence	Validation Status	Limitations
Inflammatory biomarkers	CRP/IL-6/TNF-α	Insulin resistance	Early risk prediction	Moderate (meta-analyses/observational)	Limited external validation	Standalone predictive ability limited
Adipokines	Adiponectin Leptin	Energy metabolism	Risk stratification	Moderate	Inconsistent across cohorts	Confounded by BMI and gestational age
Metabolomics	BCAA/Acylcarnitines/Lipids	Metabolic reprogramming	Early diagnosis	Emerging	Limited prospective validation	Heterogeneous analytical platforms
Gut microbiota	Akkermansia/Bifidobacteriu/SCFA	Gut-liver axis	Precision intervention	Emerging	Limited	Strain/dose heterogeneity
Genomics	TCF7L2/MTNR1B/GCK	Genetic susceptibility	Genetic risk score	Moderate	Replicated GWAS loci	Low individual predictive value
Transcriptomics	Differential placental RNA	Molecular mechanisms	Biomarker discovery	Emerging	Limited	Tissue-specific, small samples
Proteomics	Cytokines Placental proteins	Disease progression	Candidate biomarkers	Emerging	Limited	Analytical variability
AI & Machine learning	RF/SVM/XGBoost/Deep learning	Multi-factor integration	Personalized prediction	Emerging	Mostly single-center; limited external validation	Interpretability and generalizability

Recent Advances in the Management Strategies of Gestational Diabetes Mellitus

The management of gestational diabetes mellitus (GDM) aims not only to achieve optimal glycemic control during pregnancy but also to reduce the short- and long-term metabolic risks affecting both mothers and offspring. Traditionally, GDM management has primarily relied on medical nutrition therapy (MNT), physical activity interventions, and pharmacological treatment, with insulin considered the standard therapeutic option for women requiring medication. However, with the emergence of precision medicine concepts, GDM management has gradually shifted from a glucose-centered approach toward comprehensive strategies incorporating individual metabolic characteristics, dynamic glucose profiles, and personalized risk assessment. Recent advances in precision nutrition, continuous glucose monitoring (CGM), digital health technologies, artificial intelligence (AI)-assisted decision-making systems, and gut microbiota-targeted interventions have provided new opportunities to improve therapeutic efficacy and long-term maternal–offspring outcomes.

From Conventional Lifestyle Intervention to Precision Nutrition Management

Lifestyle modification remains the cornerstone of GDM management and is recommended as the first-line therapeutic approach in most clinical guidelines. Medical nutrition therapy (MNT) plays a critical role in improving pregnancy-related glucose intolerance by regulating total energy intake, optimizing nutrient composition, and reducing postprandial glucose excursions.³⁵ Traditional dietary recommendations are generally developed according to gestational age, body mass index (BMI), and maternal energy requirements, including reducing consumption of high-glycemic-index (GI) foods, appropriately distributing carbohydrate intake, and increasing dietary fiber consumption. However, recent evidence indicates that GDM represents a highly heterogeneous metabolic disorder, with substantial interindividual differences in insulin sensitivity, lipid metabolism, inflammatory status, and genetic background.^36,37 Therefore, standardized dietary strategies may not adequately address the metabolic needs of all patients. Based on this concept, precision nutrition has emerged as a promising approach for individualized GDM management. Precision nutrition emphasizes the integration of metabolic phenotypes, nutritional status, and lifestyle characteristics to develop personalized dietary interventions. Previous studies have demonstrated that low-GI dietary patterns can reduce postprandial glucose peaks and improve insulin sensitivity, whereas dietary patterns enriched in dietary fiber, high-quality proteins, and unsaturated fatty acids may further improve metabolic homeostasis by modulating inflammatory responses and lipid metabolism.^38,39 Moreover, with increasing recognition of the role of the gut microbiome in metabolic regulation, interactions between dietary composition and intestinal microbial communities have attracted growing attention. Different nutrients may influence glucose homeostasis through alterations in microbial composition and microbial-derived metabolites, providing a theoretical basis for precision interventions targeting the “diet–microbiota–metabolism axis”.⁴⁰ In addition to dietary modification, exercise intervention represents another essential component of GDM management. Regular physical activity enhances glucose uptake in skeletal muscle, improves insulin sensitivity, and promotes overall metabolic health. Recent studies suggest that moderate-intensity aerobic exercise combined with resistance training can effectively improve glycemic control and reduce the requirement for pharmacological therapy in some patients.^41,42 Nevertheless, exercise prescriptions should be individualized according to maternal age, gestational stage, body weight status, and pregnancy-related risks. Therefore, future lifestyle-based management of GDM is expected to evolve from empirical recommendations toward personalized intervention strategies guided by metabolic characteristics and treatment responses.

Continuous Glucose Monitoring Facilitating Dynamic Management of GDM

Traditional glucose management in GDM relies primarily on fasting and postprandial glucose measurements. However, intermittent glucose monitoring provides limited information regarding the complex fluctuations in glucose metabolism throughout pregnancy. The emergence of continuous glucose monitoring (CGM) technology has transformed conventional glucose management by enabling continuous and dynamic assessment of glycemic patterns.⁴³ CGM utilizes subcutaneous sensors to continuously measure interstitial glucose concentrations, allowing detection of glucose fluctuations that may be missed by conventional measurements, including transient postprandial hyperglycemia, nocturnal glucose abnormalities, and increased glycemic variability. With the widespread application of CGM, evaluation of glycemic control in GDM has gradually expanded beyond single glucose values toward comprehensive dynamic indicators, including time in range (TIR), time above range (TAR), time below range (TBR), and coefficient of variation (CV).⁴⁴ These parameters provide a more comprehensive assessment of glucose control quality and may be associated with adverse fetal growth patterns and pregnancy outcomes. Therefore, CGM not only improves glucose monitoring but also provides quantitative data to support individualized treatment adjustments. Furthermore, integration of CGM with digital health technologies has further promoted the evolution of GDM management. Through mobile health applications and telemedicine platforms, patients can continuously upload glucose data and integrate information regarding diet, physical activity, and medication use. Healthcare providers can subsequently adjust interventions based on real-time metabolic changes. This feedback-driven management model enhances patient engagement and provides more individualized strategies for women at high risk of adverse outcomes. In the future, CGM may serve as a critical technological platform connecting patient data acquisition, AI-based analysis, and precision therapeutic decision-making.

Optimization and Emerging Advances in Pharmacological Treatment Strategies

When lifestyle interventions are insufficient to achieve recommended glycemic targets, pharmacological therapy is indicated. Insulin remains the preferred pharmacological treatment in most clinical guidelines because it does not cross the placenta and has well-established efficacy and safety for glycemic control in pregnancy.⁴⁵ Various insulin formulations, including rapid-acting insulin analogs, short-acting insulin, and basal insulin, can be selected according to individual glycemic patterns, treatment requirements, and pregnancy-related metabolic changes. The increasing availability of continuous glucose monitoring (CGM) has further facilitated a shift from empiric, fixed-dose insulin adjustment toward individualized dose optimization based on dynamic glucose profiles, with the potential to improve glycemic control while minimizing hypoglycemic exposure.^46,47 Nevertheless, insulin therapy may impose a substantial injection burden and can be associated with treatment inconvenience, adherence challenges, and gestational weight gain.⁴⁸ These limitations have contributed to growing interest in oral glucose-lowering agents, particularly metformin, as potential alternatives or adjunctive therapies in selected patients with GDM. Metformin improves insulin sensitivity and reduces hepatic glucose production and is associated with a relatively low risk of hypoglycemia. Clinical studies have shown that metformin can improve glycemic control and reduce the need for supplemental insulin in women with GDM.^49,50 However, its use during pregnancy remains an area of ongoing debate because metformin crosses the placenta and reaches the fetal circulation, while the long-term implications of in utero exposure for offspring metabolic health remain incompletely established. Although studies have examined potential associations with childhood adiposity, insulin sensitivity, and later metabolic risk, the available evidence remains heterogeneous and warrants continued long-term follow-up. Other oral glucose-lowering agents also require careful consideration of pregnancy-specific efficacy and safety. Accordingly, future pharmacological management of GDM is likely to move beyond a simple drug-selection paradigm toward individualized treatment strategies guided by metabolic phenotype, disease severity, glycemic patterns, treatment response, maternal preferences, and fetal safety considerations. At present, insulin remains the preferred treatment when lifestyle measures fail to achieve adequate glycemic control, whereas the use of oral agents should be considered selectively in the context of their efficacy, maternal benefits, fetal exposure, and available long-term offspring safety evidence.

Digital Health and Artificial Intelligence for Precision GDM Management

The rapid development of digital health technologies and artificial intelligence (AI) has provided new opportunities for improving GDM management. Conventional GDM care typically depends on periodic outpatient follow-up, during which daily fluctuations in glucose levels, dietary behaviors, and physical activity patterns are difficult to capture. In contrast, mobile health (mHealth) platforms enable continuous health data collection and remote patient management. Through wearable devices and smartphone-based applications, patients can record glucose levels, dietary intake, physical activity, and medication use, while healthcare providers can deliver timely feedback and adjust management strategies based on real-world data.^51,52 This approach improves treatment adherence and facilitates more responsive disease management. AI technologies further accelerate the transition toward precision GDM management. Machine learning algorithms can integrate clinical characteristics, CGM-derived parameters, biomarker profiles, and lifestyle-related factors to identify complex patterns associated with disease progression and treatment response. For example, AI models may predict the likelihood of severe hyperglycemia or future insulin requirement, enabling early intervention and individualized treatment planning.^53,54 In addition, AI can assist in optimizing dietary recommendations, forecasting glucose fluctuations, and supporting clinical decision-making. Despite its considerable potential, several challenges remain, including insufficient data standardization, limited external validation, and inadequate model interpretability. Future studies should focus on establishing large-scale, multicenter, high-quality databases and developing explainable AI approaches to improve model reliability and clinical applicability.

Gut Microbiota-Targeted Interventions and Emerging Therapeutic Strategies

Increasing evidence highlights the important role of the gut microbiota in the development and progression of gestational diabetes mellitus (GDM), providing new insights into potential therapeutic strategies.^55,56 Gut microbial dysbiosis in women with GDM may contribute to disease pathogenesis through alterations in short-chain fatty acid production, intestinal barrier integrity, systemic inflammatory signaling, and host insulin sensitivity. Accordingly, microbiota-targeted interventions have emerged as a promising area of investigation in GDM management. Current approaches include probiotics, prebiotics, dietary fiber supplementation, and other strategies aimed at modulating the gut microbial ecosystem. Several studies have suggested that specific probiotic strains may improve glucose metabolism by enhancing intestinal barrier function, attenuating inflammatory responses, and modulating lipid metabolism.^57,58 However, substantial heterogeneity in bacterial strains, dosages, intervention durations, study populations, and outcome measures has limited the comparability of existing studies and precluded the establishment of standardized clinical protocols. Moreover, the clinical efficacy and safety of microbiota-targeted interventions during pregnancy require further confirmation in adequately powered, well-designed randomized controlled trials with reproducible findings and appropriate maternal and offspring follow-up. Integration of microbiome profiles with metabolomic signatures, clinical characteristics, and AI-based prediction models may facilitate the identification of patients most likely to benefit from microbiota-targeted interventions and support a more individualized therapeutic approach. As understanding of the gut microbiota–metabolite–host metabolic regulation axis continues to evolve, microbiome-based interventions may ultimately contribute to precision-oriented GDM management; however, most currently available microbiota-targeted therapies should be regarded as exploratory rather than clinically established strategies until their efficacy, safety, reproducibility, and long-term clinical utility have been adequately demonstrated.

Gestational Diabetes Mellitus Management in the Era of Multi-Omics and Precision Medicine

Gestational diabetes mellitus (GDM) is a complex metabolic disorder driven by the interaction of genetic susceptibility, environmental factors, maternal metabolic status, and placental dysfunction. Current diagnostic and therapeutic strategies for GDM primarily rely on glucose-based assessment and subsequent treatment adjustment; however, these approaches cannot fully explain the substantial heterogeneity in disease pathogenesis and therapeutic responses among individuals. With rapid advances in high-throughput sequencing technologies, multi-omics approaches, and artificial intelligence (AI), GDM research has entered an era of precision medicine. By integrating multidimensional biological information, including genomics, transcriptomics, proteomics, metabolomics, and microbiomics, multi-omics strategies provide a comprehensive understanding of the molecular mechanisms underlying GDM development and offer new opportunities for early risk prediction, patient stratification, and individualized intervention. To avoid repetition, the individual biomarkers, metabolomic signatures, the microbiome, genomics, and AI-based prediction models are described in detail in Recent Advances in Diagnostic Strategies for Gestational Diabetes Mellitus and are discussed in this section only with respect to their stage of clinical translation and validation.

Genomics Revealing Genetic Susceptibility Mechanisms of GDM

Genetic factors play a critical role in the development of GDM. Previous studies have demonstrated that pregnant women with a family history of type 2 diabetes mellitus (T2DM) exhibit a substantially increased risk of developing GDM, suggesting that inherited genetic backgrounds may influence maternal metabolic adaptation during pregnancy.^59,60 In recent years, genome-wide association studies (GWAS) have identified multiple genetic loci associated with GDM susceptibility, including genes involved in pancreatic β-cell function, insulin secretion, and glucose metabolism regulation. For example, genetic variants in TCF7L2, KCNJ11, MTNR1B, and GCK have been associated with increased GDM risk.^61–63 These susceptibility genes may contribute to disease development by affecting insulin secretion capacity, glucose sensing, and insulin sensitivity. Furthermore, epigenetic studies have revealed that DNA methylation, histone modifications, and non-coding RNA regulation may participate in GDM-associated metabolic abnormalities and influence placental function and maternal–fetal nutrient exchange.⁶⁴ However, GDM is not determined solely by genetic factors but results from complex interactions between genetic predisposition and environmental exposures. Therefore, genetic information alone remains insufficient for accurate disease prediction. Future approaches integrating genetic risk scores (GRS) with maternal clinical characteristics, metabolic profiles, and lifestyle factors may facilitate the development of more comprehensive and individualized GDM risk prediction models.

Transcriptomics and Proteomics Deciphering Molecular Alterations in GDM

Transcriptomic and proteomic technologies provide powerful tools for characterizing molecular alterations at the gene expression and protein functional levels, thereby improving our understanding of GDM pathogenesis. Recent RNA sequencing (RNA-seq) studies have identified extensive differential gene expression profiles in placental tissues, peripheral blood, and adipose tissues from women with GDM.^65,66 These altered genes are primarily involved in inflammatory responses, insulin signaling pathways, oxidative stress regulation, and lipid metabolism. The placenta, as a critical interface between maternal and fetal systems, plays a central role in GDM development. Studies have demonstrated enhanced expression of inflammation-related genes and dysregulation of genes involved in glucose transport, fatty acid metabolism, and mitochondrial function in placental tissues from women with GDM.⁶⁷ These transcriptional alterations may impair placental nutrient transport and contribute to fetal overgrowth and metabolic abnormalities. Proteomic analyses have further revealed distinct circulating protein profiles associated with GDM. Several inflammatory proteins, lipid metabolism-related proteins, and insulin signaling regulators have been identified as potential biomarkers of disease risk.^68,69 Increased inflammatory mediators, adipokine imbalance, and abnormalities in insulin signaling-associated proteins may collectively contribute to the development of insulin resistance. Despite these advances, most transcriptomic and proteomic studies remain focused on biomarker discovery. Further large-scale validation is required to determine whether these molecular signatures can be translated into clinically applicable predictive tools.

Metabolomics Revealing Metabolic Reprogramming Associated with GDM

Metabolic abnormalities represent a central feature of GDM pathophysiology, making metabolomics an important component of precision medicine research. Metabolomics enables systematic profiling of small-molecule metabolites in biological samples, including blood, urine, and tissues, thereby providing a comprehensive assessment of metabolic alterations. Accumulating evidence indicates that women who subsequently develop GDM exhibit metabolic disturbances before clinical diagnosis, including abnormalities in amino acid metabolism, lipid metabolism, and energy metabolism.^70,71 Among these alterations, elevated branched-chain amino acids (BCAAs) have been strongly associated with insulin resistance. Increased BCAA levels may promote metabolic dysfunction through modulation of the mammalian target of rapamycin (mTOR) signaling pathway and impairment of insulin signaling.⁷² Additionally, increased free fatty acids, elevated triglycerides, and altered bile acid metabolism have also been implicated in GDM development.⁷³ Beyond biomarker discovery, metabolomics may facilitate molecular classification of GDM subtypes.^74,75 For instance, some patients may primarily exhibit insulin resistance, whereas others may present with predominant pancreatic β-cell dysfunction. Stratification based on metabolic phenotypes may therefore enable more precise selection of therapeutic strategies. In the future, integration of metabolomic profiles with clinical characteristics may facilitate the development of metabolism-based prediction models, promoting the transition from disease diagnosis toward early prediction and personalized intervention.

Gut Microbiome and the Maternal–Fetal Metabolic Regulatory Axis

The gut microbiome is increasingly recognized as an important regulatory interface linking environmental exposures, host metabolism, and immune homeostasis. Growing evidence indicates that women with gestational diabetes mellitus (GDM) exhibit distinct alterations in gut microbial composition and function, including reduced microbial diversity, changes in the abundance of specific bacterial taxa, and perturbations in microbial metabolic activity.⁷⁶ These alterations may contribute to GDM pathogenesis through several interconnected mechanisms. First, gut microbial dysbiosis may compromise intestinal barrier integrity, increasing the translocation of bacterial lipopolysaccharide (LPS) and other microbial products into the circulation and thereby promoting chronic low-grade inflammation and insulin resistance. Second, microbiota-derived metabolites, including short-chain fatty acids (SCFAs), bile acid derivatives, and tryptophan metabolites, can influence host energy metabolism, immune regulation, and glucose homeostasis through interactions with metabolic and signaling pathways.^77,78 In addition, maternal gut microbial alterations may affect offspring metabolic health by influencing placental function and the intrauterine metabolic environment, although the underlying mechanisms and causal relationships remain incompletely defined.⁷⁹ These findings have led to increasing interest in the concept of a “maternal microbiota–metabolites–placenta–fetus axis” as a potential framework for understanding the intergenerational metabolic consequences of GDM. Advances in metagenomic sequencing, metabolomics, and AI-assisted integrative analysis may facilitate the identification of reproducible microbial signatures and metabolite pathways associated with disease development and treatment response, thereby supporting the development of microbiota-informed precision interventions. While the gut microbiome has also been investigated as a potential biomarker for early GDM risk stratification, its therapeutic relevance represents a distinct area of investigation, requiring validation of causal mechanisms, reproducibility, clinical efficacy, and maternal and offspring safety before microbiota-based interventions can be incorporated into routine GDM management.

Artificial Intelligence-Driven Multi-Omics Integration and Precision Medicine

With the rapid accumulation of multi-omics datasets, integrating heterogeneous biological information into clinically actionable models has become a major challenge in precision medicine. Artificial intelligence (AI) and machine learning (ML) offer powerful computational tools for addressing this complexity by enabling the integration of genomic, transcriptomic, proteomic, metabolomic, microbiome, and clinical data and the identification of complex, nonlinear interactions that may be difficult to capture using conventional statistical methods. In GDM, multi-omics-based machine learning models have increasingly been explored for early risk prediction, with approaches such as random forest, support vector machine (SVM), extreme gradient boosting (XGBoost), and deep learning being used to integrate maternal characteristics, glucose-related parameters, biomarkers, and multi-omics features for individualized risk stratification.⁸⁰ Beyond risk prediction, AI-based integration of multi-omics and clinical data may also support treatment selection by identifying patients with distinct metabolic or molecular profiles who are more likely to require insulin therapy or respond favorably to specific lifestyle or pharmacological interventions.⁸¹ Such approaches could facilitate a shift from population-based treatment algorithms toward response-guided and biologically informed therapeutic strategies. However, several barriers currently limit the clinical implementation of multi-omics-based AI models, including inadequate data standardization, relatively small and heterogeneous datasets, incomplete representation of diverse populations, limited model interpretability, and uncertain generalizability across clinical settings. In addition, differences in sample collection, analytical platforms, feature selection, and outcome definitions may compromise reproducibility and hinder meaningful comparison between studies. Future research should therefore prioritize large-scale, multicenter, prospectively collected and population-diverse datasets, standardized analytical workflows, rigorous external validation, and explainable AI approaches to improve model robustness, reproducibility, and clinical utility.

Future Directions of GDM Management in the Precision Medicine Era

With continued advances in multi-omics technologies, continuous glucose monitoring (CGM), digital health, and artificial intelligence (AI), the management of gestational diabetes mellitus (GDM) is increasingly moving toward a precision medicine-oriented framework. Rather than replacing established diagnostic and therapeutic strategies, these emerging approaches may complement conventional care by enabling earlier risk stratification, more comprehensive phenotyping, and individualized treatment optimization. Integrating genetic susceptibility, metabolic phenotypes, microbiome characteristics, and dynamic glucose profiles may facilitate earlier identification of women at increased risk of GDM and support individualized intervention throughout pregnancy. An ideal precision management framework may incorporate early-pregnancy risk stratification based on validated genetic, metabolic, and inflammatory biomarkers; continuous metabolic monitoring through CGM and digital health platforms; personalized nutritional and pharmacological interventions guided by individual metabolic and molecular characteristics; selected microbiota-targeted interventions; and dynamic treatment optimization through AI-driven clinical decision-support systems. This transition from predominantly reactive treatment toward more proactive risk prediction, prevention, and response-guided management represents an important future direction in GDM research and clinical practice, with the potential to improve maternal metabolic health and long-term offspring outcomes. However, many of these approaches, including multi-omics biomarkers, microbiota-based interventions, and AI-based prediction models, remain at the research or validation stage and require prospective studies, external validation, standardized methodologies, assessment of clinical utility and cost-effectiveness, and evaluation of maternal and offspring safety before routine clinical implementation. In contrast, OGTT-based diagnosis and lifestyle modification with insulin-based pharmacological management remain established components of current GDM care. The current evidence level and clinical readiness of each emerging approach are summarized in Table 3. Future research should therefore focus not on prematurely replacing established standards, but on determining which precision medicine tools can be safely and effectively integrated into existing clinical practice.

Table 3.

Evidence-to-Translation Summary of Emerging Approaches in GDM

Technology/Approach	Current Evidence Level	Clinical Readiness
Inflammatory biomarkers (CRP, IL-6, TNF-α)	Moderate	Research stage for standalone prediction; may support risk stratification
Adipokines (adiponectin, leptin)	Moderate	Research stage
Metabolomics (BCAA, lipids)	Emerging	Research stage
Gut microbiota/microbiome	Emerging	Research stage
Genomics/genetic risk scores	Moderate	Emerging; not yet standalone
Multi-omics integration	Emerging	Research stage
AI/machine learning prediction	Emerging	Research stage; requires external validation
Microbiota-targeted interventions (probiotics/prebiotics)	Emerging	Research stage; requires RCTs
CGM and digital health	Established/emerging	Clinical use increasing
OGTT-based diagnosis and lifestyle/insulin management	Established	Established clinical practice

Current Challenges and Future Perspectives

Although substantial advances have been achieved in the diagnosis and management of gestational diabetes mellitus (GDM), several challenges continue to limit its clinical implementation and the transition toward precision medicine-based care. With increasing understanding of the complex mechanisms underlying GDM, conventional glucose-centered screening and treatment strategies are gradually evolving toward multidimensional risk assessment and individualized management approaches. However, major limitations remain, including heterogeneity in diagnostic criteria, insufficient early prediction capability, limited personalization of therapeutic strategies, and inadequate clinical translation of multi-omics discoveries. Therefore, future progress in GDM management will require the integration of clinical medicine, multi-omics technologies, artificial intelligence (AI), and digital health platforms to facilitate the transition from empirical management toward precision medicine.

Heterogeneity of Diagnostic Criteria and Insufficient Early Prediction Capacity

Currently, GDM diagnosis relies primarily on the oral glucose tolerance test (OGTT); however, diagnostic thresholds vary among international organizations, including the International Association of Diabetes and Pregnancy Study Groups (IADPSG), World Health Organization (WHO), American Diabetes Association (ADA), and American College of Obstetricians and Gynecologists (ACOG). This lack of consensus contributes to substantial variations in reported GDM prevalence across populations and complicates the development of standardized clinical management strategies. Moreover, current diagnostic approaches mainly focus on identifying abnormal glucose metabolism during the second trimester, whereas metabolic disturbances associated with GDM may occur much earlier, even before pregnancy. Early alterations in insulin resistance, inflammatory activation, lipid metabolism, and placental function may precede detectable hyperglycemia. Therefore, reliance solely on OGTT may fail to identify individuals at high risk and limits opportunities for early preventive intervention. Future diagnostic strategies should shift from “disease confirmation” toward “risk prediction”. By integrating clinical risk factors, circulating biomarkers, multi-omics signatures, and AI-based algorithms, more accurate early pregnancy prediction models may be developed to identify high-risk individuals and enable timely personalized interventions.

Challenges in Clinical Translation of Multi-Omics Research

Advances in genomics, transcriptomics, proteomics, metabolomics, and microbiomics have substantially expanded our understanding of the pathophysiology of gestational diabetes mellitus (GDM) and facilitated the identification of numerous potential biomarkers. However, most multi-omics findings remain at the exploratory stage, and their translation into clinically applicable tools continues to face substantial challenges. First, multi-omics results are highly sensitive to differences in sample size, population characteristics, biospecimen collection, analytical platforms, and computational pipelines, which may contribute to substantial variability and limit reproducibility across studies. Second, although many candidate biomarkers have demonstrated promising predictive performance in individual studies, relatively few have been validated in large-scale, multicenter, prospective, and longitudinal cohorts, limiting confidence in their generalizability, clinical reliability, and practical utility. Third, the high cost, technical complexity, and specialized infrastructure required for multi-omics profiling may pose additional barriers to its implementation in routine prenatal screening. These limitations are therefore not simply future considerations but are central to the interpretation of the current evidence supporting multi-omics biomarkers in GDM. Future research should shift from isolated biomarker discovery toward the development and validation of clinically applicable, integrated prediction systems. For example, combining genetic risk scores, metabolic signatures, inflammatory profiles, gut microbiome characteristics, and established clinical risk factors may provide a more comprehensive framework for individualized GDM risk assessment. In parallel, standardized protocols for sample collection and processing, harmonized analytical pipelines, cross-platform data integration, and independent external validation will be essential to improve reproducibility, generalizability, and clinical translation. Importantly, such integrated models should ultimately be evaluated not only for statistical discrimination but also for calibration, clinical utility, cost-effectiveness, and feasibility in real-world prenatal care before routine clinical implementation.

Further Development of Individualized Therapeutic Strategies

Current GDM management primarily relies on lifestyle modification and pharmacological intervention; however, clinical practice still largely follows relatively standardized treatment approaches. Nevertheless, substantial heterogeneity exists among patients regarding disease mechanisms, metabolic phenotypes, and therapeutic responses. For example, some individuals primarily exhibit insulin resistance, whereas others may present with impaired pancreatic β-cell function. Similarly, some patients achieve adequate glycemic control through dietary and exercise interventions alone, while others require pharmacological treatment. Therefore, future GDM management should focus on developing patient-specific therapeutic strategies. Integration of metabolic profiling may help identify distinct metabolic phenotypes, while CGM-derived glucose patterns may facilitate real-time treatment adjustment. Furthermore, AI-based models may predict individual responses to different interventions, enabling a shift toward a strategy of “selecting the optimal treatment for each patient”. In addition, several emerging therapeutic approaches still lack sufficient evidence regarding long-term safety. Although metformin demonstrates effective glucose-lowering effects in GDM, its potential impact on offspring metabolic health remains under investigation. Similarly, microbiota-based interventions, including probiotics, show promising therapeutic potential, but optimal bacterial strains, dosages, and intervention durations have not yet been established. Future large-scale randomized controlled trials are therefore required to provide stronger evidence supporting personalized therapeutic approaches.

Opportunities and Limitations of Artificial Intelligence and Digital Health Applications

Artificial intelligence and digital health technologies are increasingly transforming GDM management. By integrating electronic health records, continuous glucose monitoring data, lifestyle information, and multi-omics profiles, AI-based systems have the potential to improve risk prediction, disease stratification, and therapeutic decision-making. However, the clinical implementation of AI in GDM remains at an early stage. First, many AI models are developed using single-center or relatively small datasets and lack external validation, raising concerns regarding their generalizability across different geographic regions, ethnic populations, and healthcare settings. Second, many complex machine learning algorithms suffer from limited interpretability, often referred to as the “black-box” problem. Although these models may achieve high predictive accuracy, the inability to explain decision-making processes may reduce clinician confidence and hinder clinical adoption. Additionally, issues related to patient privacy, data security, and medical ethics represent important considerations for the implementation of digital health technologies. Future AI development should move beyond improving predictive accuracy alone and focus on enhancing model interpretability and clinical usability. Integration of explainable artificial intelligence (XAI), standardized healthcare databases, and multicenter validation studies will be essential for establishing AI as a reliable tool for precision GDM management.

From Maternal Disease Management to Long-Term Maternal and Offspring Health Protection

Historically, GDM management has primarily focused on achieving glycemic control during pregnancy and preventing short-term obstetric complications. However, accumulating evidence indicates that GDM represents not only a pregnancy-specific metabolic disorder but also an important predictor of future health risks. Women with a history of GDM have an increased likelihood of developing type 2 diabetes mellitus, cardiovascular disease, and metabolic syndrome later in life. Meanwhile, offspring exposed to an intrauterine hyperglycemic environment may have increased risks of childhood obesity, insulin resistance, and metabolic disorders during adulthood. Therefore, future GDM management should extend beyond pregnancy-specific interventions toward lifelong health management. Postpartum follow-up, metabolic risk assessment, and sustained lifestyle modification should become integral components of comprehensive GDM care. Furthermore, investigating the interactions among maternal metabolic status, placental function, and offspring epigenetic regulation may provide deeper insights into the mechanisms underlying intergenerational metabolic transmission. An ideal future GDM management framework should establish a life-course approach covering the preconception period, pregnancy, and postpartum stages, integrating risk prediction, early diagnosis, precision treatment, and long-term health maintenance.

Ethical Statement

This was deemed not applicable.

Informed Consent/Patient Consent

This was deemed not applicable.

Disclosure

The authors declare that they have no conflicts of interest.