Javaid I. Sheikh

Gestational diabetes mellitus (GDM) significantly increases the risk of developing type 2 diabetes mellitus (T2DM) post partum, with up to half of affected women progressing within a decade. Early identification of high-risk individuals is critical for implementing preventive interventions. Artificial intelligence (AI) offers enhanced predictive capabilities that can substantially enhance the prevention of postpartum diabetes. This systematic review and meta-analysis aimed to evaluate the performance of AI models in predicting the progression from GDM to T2DM or prediabetes. A total of 7 databases (MEDLINE, Embase, Scopus, Web of Science, IEEE Xplore, ACM Digital Library, and Google Scholar) were systematically searched from inception through September 12, 2025, supplemented by backward and forward reference screening and biweekly alerts to capture newly published studies. This review included peer-reviewed English-language studies that applied AI algorithms to predict T2DM or prediabetes among women with previous GDM. Eligible studies focused on human participants; reported performance metrics (eg, accuracy, sensitivity, and specificity); and excluded non-AI models, animal studies, reviews, protocols, abstracts, and non-English publications. Moreover, 2 reviewers independently conducted study selection, data extraction, and risk of bias assessment using the PROBAST (Prediction Model Risk of Bias Assessment Tool)+AI tool. Pooled estimates were computed using random-effects meta-analysis models. In total, 10 studies met the inclusion criteria, of which 8 were eligible for meta-analysis. The reviewed studies spanned from 2011 to 2025 and were conducted across 7 countries, predominantly in the United States (3/10, 30%). Most publications were journal articles (9/10, 90%), and retrospective designs (6/10, 60%) were slightly more common than prospective designs (4/10, 40%). AI models demonstrated high predictive performance for T2DM, with pooled accuracy of 0.85 (95% CI 0.79‐0.90; prediction interval [PI] 0.64‐0.98), sensitivity of 0.89 (95% CI 0.81‐0.95; PI 0.63‐1.00), specificity of 0.88 (95% CI 0.81‐0.93; PI 0.67‐0.99), F₁-score of 0.80 (95% CI 0.75‐0.85; PI 0.68‐0.93), and area under the curve of 0.86 (95% CI 0.77‐0.91; PI 0.54‐0.97). However, AI performance for prediabetes prediction was modest (area under the curve=0.69, 95% CI 0.60‐0.77). Subgroup analyses showed that random forest, decision tree, logistic regression, and naïve Bayes models performed comparably. Fasting plasma glucose and BMI were the most identified significant predictors in the included studies. AI models show potential in predicting T2DM after GDM. However, evidence remains limited by small sample sizes, high heterogeneity, lack of external validation, and high risk of bias. Our findings have important implications for digital health, supporting the integration of AI-driven risk prediction into electronic health record systems and postpartum care pathways to enable early identification, targeted prevention, and improved long-term outcomes. Future research should use large, diverse cohorts, integrate multidimensional data, adopt standardized reporting frameworks, and encourage open-access data sharing.

Ramadan fasting substantially alters meal timing, sleep patterns, and daily activity, thereby increasing the risk of hypoglycaemia in adults with type 1 diabetes (T1D). Although continuous glucose monitoring (CGM) systems provide real-time alerts, these are largely reactive or limited to short prediction horizons, offering insufficient warning under fasting-related behavioural and circadian disruption. This study aims to evaluate whether behaviour-aware, temporally enriched recurrent deep learning models, leveraging multimodal CGM and wearable-derived signals, can forecast hypoglycaemia one hour ahead during Ramadan and the post-fasting period. In an observational, free-living cohort study conducted in Qatar, 33 adults with T1D were monitored using CGM and a wrist-worn wearable during Ramadan 2023 and the subsequent month. Multimodal data were aggregated into hourly features and organised into rolling 36 h sequences. In addition to physiological signals, explicit temporal and circadian proxy features were engineered, including cyclic time encodings, day–night indicators, and Ramadan-specific behavioural windows (e.g., pre-iftar, iftar, post-iftar, and fasting phases). Recurrent models, including LSTM and BiLSTM architectures, were trained using patient-wise, leak-free splits, with focal loss applied to address class imbalance. Model performance was evaluated on a held-out, naturally imbalanced test set using ROC AUC, precision–recall AUC, recall, and probability calibration, alongside cross-phase evaluation between Ramadan and post-fasting periods. Following quality control, 1164 participant-days were retained, with hypoglycaemia accounting for approximately 4% of hourly observations. Temporal feature enrichment and the use of a 36 h lookback window improved both discrimination and calibration, with performance stabilizing beyond this horizon. On the imbalanced test set, the best-performing multimodal model achieved an ROC AUC of 0.867 and a precision–recall AUC of 0.341, identifying 77% of next-hour hypoglycaemic events at a sensitivity-focused operating point (precision = 0.14). The selected BiLSTM model demonstrated good probability calibration (Brier score ≈ 0.03). Models trained using wearable-derived inputs alone achieved comparable discrimination and, in some configurations, higher precision–recall AUC than CGM-only baselines. Notably, models trained on the original imbalanced data outperformed resampled variants, suggesting that temporal and behavioural features provided sufficient discriminatory signal without requiring aggressive class balancing. Cross-phase evaluation indicated robust generalisation, particularly for the BiLSTM model. Overall, behaviour-aware, temporally enriched multimodal models can provide calibrated, hour-ahead hypoglycaemia risk estimates during Ramadan fasting in adults with T1D, enabling proactive intervention beyond reactive CGM alerts. Explicit modelling of circadian and behavioural dynamics enhances predictive performance under real-world class imbalance. Furthermore, integrating wearable-derived behavioural and physiological signals adds predictive value beyond CGM alone, supporting robustness across varying levels of contextual data availability. External validation and prospective clinical evaluation are required prior to deployment.

The persistent challenge posed by drug-resistant M.tb phenotypes requires new therapeutic approaches that diminish pathogenic potential without exerting a direct bactericidal action, and thus reduce the natural selective pressure driving drug resistance. PE_PGRS proteins, a multi-gene family confined to pathogenic mycobacteria, act as virulence effectors that regulate host immune responses and promote survival in the intracellular environment. Specific inhibition of these proteins presents a mechanistically distinct and novel anti-virulence therapeutic strategy whereby virulence phenotypes may be impaired by circumventing classical pathways of resistance. In the present study, PE_PGRS32 (Rv1803c) and PE_PGRS35 (Rv1983) were selected as virulence targets for an FDA-approved anti-virulence drug repurposing strategy based on their structural conservation, host-pathogen relevance, and favorable druggability. Structure-based virtual screening of ~2,600 compounds, followed by hierarchical molecular docking and 100 ns all-atom molecular dynamics (MD) simulations, allowed the identification of Fenoterol and Pentagastrin as leads among the hits, due to their docking scores (<– 5 kcal/mol), interaction, and conformational stability in both targets. In vitro assays against M.tb H37Rv showed dose-dependent inhibition with MICs in the range of 12.5-25 µg/mL, with Fenoterol displaying higher potency. In addition, complementary ex vivo infection models confirmed a marked reduction in intracellular bacteria (~78%), validating the efficacy of the compound with no detectable cytotoxicity. These findings validate the PE_PGRS proteins as virulence targets with pathogen specificity and suggest the therapeutic potential of dual-target inhibition mediated through repurposed scaffolds. Hence, the anti-virulence approach is a resistance-sparing strategy to disrupt the host-pathogen interactions pivotal for the persistence of M.tb.

Background: Postpartum maternal mental health (MMH) symptoms, including depression, anxiety, and childbirth-related post-traumatic stress disorder, are known to influence infant sleep trajectories. While previous research has examined their individual and combined associations, the predictive utility of these MMH symptoms for the early identification of infant sleep problems through machine learning (ML) remains understudied. Objective: This study aimed to examine whether postpartum MMH measures can predict infant sleep outcomes during the first year of life. The analysis focused on 2 clinically relevant sleep indicators: (1) nocturnal sleep duration and (2) night awakening frequency. Methods: A total of 409 mother-infant dyads were included in the study. Predictor variables comprised postpartum MMH symptoms assessed between 3 and 12 months postpartum, along with sociodemographic characteristics of mothers and infants. MMH symptoms were measured using 3 validated instruments: the Edinburgh Postnatal Depression Scale, the Hospital Anxiety and Depression Scale, and the City Birth Trauma Scale. Infant sleep outcomes were assessed using the Brief Infant Sleep Questionnaire. Six supervised ML algorithms were evaluated: logistic regression, random forest, support vector classifier, extreme gradient boosting, Light Gradient Boosting Machine, and multilayer perceptron. Post hoc feature importance analyses were conducted to identify the most influential predictors associated with each infant sleep outcome. Results: All models demonstrated high predictive performance. The best model achieved a precision-recall area under the curve of 0.92, F ₁ -score of 0.84, and accuracy of 0.88 for predicting short nocturnal sleep duration. For frequent night awakenings, the top precision-recall area under the curve was 0.91, with an F ₁ -score of 0.78 and accuracy of 0.85. Key predictors included maternal age and total scores from the Edinburgh Postnatal Depression Scale, Hospital Anxiety and Depression–Anxiety subscale, and City Birth Trauma Scale, with individual symptom items offering additional discriminative value. Conclusions: ML models can accurately predict which infants are at risk for suboptimal sleep based on MMH measures, enabling personalized, responsive, and developmentally informed postpartum care that promotes long-term maternal and infant well-being.

Background: Postpartum depression (PPD) affects up to 20% of mothers globally. Early detection is vital for better outcomes, yet screening lacks scalability and predictive power. Artificial intelligence (AI)—through machine learning, deep learning, and natural language processing—enhances the early identification of mothers at risk with greater accuracy. Objective: This study aims to systematically map the existing literature on AI-based methods for detecting and predicting PPD. Methods: This scoping review was conducted in accordance with the PRISMA-ScR (Preferred Reporting Items for Systematic Reviews and Meta-Analyses Extension for Scoping Reviews) guidelines. We included empirical studies that applied AI techniques to detect or predict PPD and were published in peer-reviewed journals, conference proceedings, or dissertations. Studies were excluded if they were nonempirical (eg, reviews, editorials, and abstracts), not published in English, focused on general perinatal mental health without a specific emphasis on PPD, or used AI solely for monitoring or treatment rather than prediction or detection. We systematically searched 8 databases—MEDLINE, Embase, PsycINFO, CINAHL, Scopus, IEEE Xplore, ACM Digital Library, and Google Scholar—from inception through February 28, 2025. The search strategy was supplemented by backward and forward reference screening and biweekly alerts to capture newly published studies. Two independent (M [Alkhateeb] and A [Nayeem])reviewers (M [Alkhateeb] and A [Nayeem]) screened the retrieved studies, with disagreements resolved by a third reviewer (AA [Alrazaq]). Data were extracted by 2 independent reviewers using a standardized extraction form capturing study characteristics, AI model types, data sources, features, preprocessing, validation strategies, and performance metrics. A formal risk-of-bias assessment was not performed due to the scoping nature of the review. All extracted data were synthesized narratively. Results: Out of 503 retrieved studies, 65 met the inclusion criteria. The United States contributed the largest proportion of studies (18/65, 27.7%). The highest number of publications occurred in 2024 (17/65, 26%). Most included studies were journal articles (46/65, 71%). Short-term postpartum outcomes (≤12 weeks) were most frequently assessed (20/65, 30.8%). Most included studies (52/65, 80%) applied AI models for predicting PPD, while 14 of 65 (22%) studies used them for detection. Sociodemographic data were most frequently used (49/65, 75.4%), followed by psychological data (44/65, 68%) and obstetric data (35/65, 55%). Data preprocessing mostly relied on basic scaling (51/65, 79%) and some missing data imputation (29/65, 44.6%). Machine learning dominated (57/65, 87.7%), especially random forest, support vector machines, and logistic regression. Internal validation (k-fold, hold-out) was standard, while external validation was scarce. Ensemble-based boosting models consistently demonstrated superior performance across key metrics, highlighting their potential for accurate and scalable PPD prediction. Current studies suffer from limited sample sizes, geographic bias, lack of standardized feature sets, minimal external validation, and inconsistent reporting of comprehensive model metrics. Conclusions: This scoping review analyzes 65 studies on AI in PPD, highlighting dominant use of classical machine learning, limited deep learning adoption, underuse of advanced preprocessing, inconsistent validation, and reliance on structured, unimodal data—mainly sociodemographic, clinical, and obstetric features.

Artificial intelligence (AI) has the capacity to transform health care by improving clinical decision-making, optimizing workflows, and enhancing patient outcomes. However, this potential remains limited by a complex set of technological, human, and ethical barriers that constrain its safe and equitable implementation. This paper argues for a holistic, systems-based approach to AI integration that addresses these challenges as interconnected rather than isolated. It identifies key technological barriers, including limited explainability, algorithmic bias, integration and interoperability issues, lack of generalizability, and difficulties in validation. Human factors such as resistance to change, insufficient stakeholder engagement, and education and resource constraints further impede adoption, whereas ethical and legal challenges related to liability, privacy, informed consent, and inequity compound these obstacles. Addressing these issues requires transparent model design, diverse datasets, participatory development, and adaptive governance. Recommendations emerging from this synthesis are as follows: (1) establish standardized international regulatory and governance frameworks; (2) promote multidisciplinary co-design involving clinicians, developers, and patients; (3) invest in clinician education, AI literacy, and continuous training; (4) ensure equitable resource allocation through dedicated funding and public-private partnerships; (5) prioritize multimodal, explainable, and ethically aligned AI development; and (6) focus on long-term evaluation of AI in real-world settings to ensure adaptive, transparent, and inclusive deployment. Adopting these measures can align innovation with accountability, enabling health care systems to harness AI’s transformative potential responsibly and sustainably to advance patient care and health equity.

Background: Epilepsy affects approximately 50 million people globally and imposes a substantial clinical and societal burden, requiring continuous and personalized monitoring for effective management. Wearable artificial intelligence (AI) technologies offer a promising solution by leveraging physiological signals and machine learning for seizure detection and prediction. While various approaches have been proposed, a comprehensive overview summarizing these advances and challenges is still needed. Objective: This review aims to comprehensively explore and map the existing literature on AI-driven wearable technologies for epilepsy, identifying device characteristics, AI methodologies, biosignal measurements, validation approaches, and research gaps. Methods: A scoping review was conducted following the PRISMA-ScR (Preferred Reporting Items for Systematic Reviews and Meta-Analyses Extension for Scoping Reviews) guidelines. A systematic search was performed across six electronic databases (Scopus, MEDLINE, Embase, ACM Digital Library, IEEE Xplore, and Google Scholar) to identify relevant studies published up to December 2023. We included studies that developed AI algorithms for epilepsy using noninvasive wearable devices (eg, smartwatches, smart clothing) and excluded those using nonwearables or in-body devices. Eligible publication types included journal articles, conference papers, and dissertations. Study selection and data extraction were performed independently by six reviewers. The extracted data were synthesized narratively. Results: A total of 67 studies met the inclusion criteria. Research in this domain has increased significantly since 2021, with India, the United States, and China leading contributions. The studies examined both commercial (n=31, 46.3%) and noncommercial (n=31, 46.3%) wearable devices, with Empatica smart bands being the most frequently used. The primary biosignals monitored included activity measures (n=36, 53.7%), cardiovascular metrics (n=33, 49.3%), brain activity (n=24, 35.8%), and skin conductance (n=23, 34.3%). The most common AI models were support vector machines (n=28, 41.8%), random forests (n=14, 20.9%), and convolutional neural networks (n=10, 14.9%). Most models focused on seizure detection (n=54, 80.6%) compared to seizure prediction (n=14, 20.9%), reflecting a research imbalance that suggests the need for further development in predictive analytics. Sensitivity (n=54, 80.6%) was the most frequently reported performance metric, indicating a focus on identifying seizures; however, comprehensive clinical validation remains limited. Closed-source data predominated (n=44, 65.7%), limiting the generalizability of findings. The most used validation methods were leave-one-out cross-validation (n=21, 31.3%) and k-fold cross-validation (n=20, 29.9%), while video electroencephalography served as the primary reference standard (n=35, 52.2%). Conclusions: Wearable AI technologies show significant promise in epilepsy management, offering real-time, continuous monitoring and early seizure detection. To realize clinical impact, future research should prioritize the standardization of validation methods, promote open data exchange for reproducibility, and develop energy-efficient algorithms that support real-world deployment in wearable devices.

Background: Worldwide, 30%-45% of adults have sleep disorders, which are linked to major health issues such as diabetes and cardiovascular disease. Long-term monitoring with traditional in-lab testing is impractical due to high costs. Wearable artificial intelligence (AI)–powered solutions offer accessible, scalable, and continuous monitoring, improving the identification and treatment of sleep problems. Objective: This scoping review aims to provide an overview of AI-powered wearable devices used for sleep disorders, focusing on study characteristics, wearable technology features, and AI methodologies for detection and analysis. Methods: Seven electronic databases (MEDLINE, PsycINFO, Embase, IEEE Xplore, ACM Digital Library, Google Scholar, and Scopus) were searched for peer-reviewed literature published before March 2024. Keywords were selected based on 3 domains: sleep disorders, AI, and wearable devices. The primary selection criterion was the inclusion of studies that utilized AI algorithms to detect or predict various sleep disorders using data from wearable devices. Study selection was conducted in 2 steps: first, by reviewing titles and abstracts, followed by full-text screening. Two reviewers independently conducted study selection and data extraction, resolving discrepancies by consensus. The extracted data were synthesized using a narrative approach. Results: The initial search yielded 615 articles, of which 46 met the eligibility criteria and were included in the final analysis. The majority of studies focused on sleep apnea. Wearable AI was widely deployed for diagnosing and screening disorders; however, none of the studies used it for treatment. Commercial devices were the most commonly used type of wearable technology, appearing in 30 out of 46 (65%) studies. Among these, various brands were utilized rather than a single large, well-known brand; 19 (41%) studies used wrist-worn devices. Respiratory data were used by 25 of 46 (54%) studies as the primary data for model development, followed by heart rate (22/46, 48%) and body movement (17/46, 37%). The most popular algorithm was the convolutional neural network, adopted by 17 of 46 (37%) studies, followed by random forest (14/46, 30%) and support vector machines (12/46, 26%). Conclusions: Wearable AI technology offers promising solutions for sleep disorders. These devices can be used for screening and diagnosis; however, research on wearable technology for sleep disorders other than sleep apnea remains limited. To statistically synthesize performance and efficacy results, more reviews are needed. Technology companies should prioritize advancements such as deep learning algorithms and invest in wearable AI for treating sleep disorders, given its potential. Further research is necessary to validate machine learning techniques using clinical data from wearable devices and to develop useful analytics for data collection, monitoring, prediction, classification, and recommendation in the context of sleep disorders.

In the complex and multidimensional field of medicine, multimodal data are prevalent and crucial for informed clinical decisions. Multimodal data span a broad spectrum of data types, including medical images (eg, MRI and CT scans), time-series data (eg, sensor data from wearable devices and electronic health records), audio recordings (eg, heart and respiratory sounds and patient interviews), text (eg, clinical notes and research articles), videos (eg, surgical procedures), and omics data (eg, genomics and proteomics). While advancements in large language models (LLMs) have enabled new applications for knowledge retrieval and processing in the medical field, most LLMs remain limited to processing unimodal data, typically text-based content, and often overlook the importance of integrating the diverse data modalities encountered in clinical practice. This paper aims to present a detailed, practical, and solution-oriented perspective on the use of multimodal LLMs (M-LLMs) in the medical field. Our investigation spanned M-LLM foundational principles, current and potential applications, technical and ethical challenges, and future research directions. By connecting these elements, we aimed to provide a comprehensive framework that links diverse aspects of M-LLMs, offering a unified vision for their future in health care. This approach aims to guide both future research and practical implementations of M-LLMs in health care, positioning them as a paradigm shift toward integrated, multimodal data–driven medical practice. We anticipate that this work will spark further discussion and inspire the development of innovative approaches in the next generation of medical M-LLM systems.