{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,9,16]],"date-time":"2026-09-16T19:04:30Z","timestamp":1789585470193,"version":"build-2803163510"},"reference-count":46,"publisher":"Oxford University Press (OUP)","issue":"9","license":[{"start":{"date-parts":[[2026,6,18]],"date-time":"2026-06-18T00:00:00Z","timestamp":1781740800000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/academic.oup.com\/pages\/standard-publication-reuse-rights"}],"content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":[],"published-print":{"date-parts":[[2026,9,1]]},"abstract":"<jats:title>Abstract<\/jats:title>\n                  <jats:sec>\n                    <jats:title>Background<\/jats:title>\n                    <jats:p>People with type 1 diabetes mellitus (T1DM) show glucose variability driven by insulin dosing, meals, activity, and circadian rhythms. Many deep learning approaches treat glucose forecasting and hypoglycemia detection as separate tasks and provide limited transparency.<\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec>\n                    <jats:title>Objective<\/jats:title>\n                    <jats:p>We developed an explainable, multi-task temporal graph framework that jointly predicts glucose trajectories and hypoglycemia risk at 30 and 60 minute horizons, and provides bounded, patient-specific insulin adjustment recommendations.<\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec>\n                    <jats:title>Methods<\/jats:title>\n                    <jats:p>Temporal GAT-BiGRU transforms multimodal continuous glucose monitoring (CGM) time series into a temporal k-neighborhood graph, with each time point represented as a feature-enriched node. A graph-attention encoder performs multi-head message passing over history edges, while an attention-based BiGRU captures longer dependencies. We evaluated OhioT1DM and BrisT1D using hypoglycemia and predicted Time in Range (TIR) metrics. A prediction-driven counterfactual module retrospectively generates bounded basal\/bolus adjustments using patient-specific Insulin-to-Carbohydrate Ratio (ICR) and Insulin Sensitivity Factor (ISF); interpretability is supported via GNNExplainer.<\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec>\n                    <jats:title>Results<\/jats:title>\n                    <jats:p>On OhioT1DM, Temporal GAT-BiGRU achieved hypoglycemia precision\u2013recall area under the curve (PR-AUC) 0.93, mean absolute error (MAE) 9.40 mg\/dL, root mean square error (RMSE) 15.8 mg\/dL, mean absolute relative difference (MARD) 6.01%, and predicted TIR 71.39%. On BrisT1D, performance remained strong with PR-AUC 0.98, MAE 9.36 mg\/dL, RMSE 15.3 mg\/dL, MARD 7.55%, and predicted TIR 64.78%. The insulin module generated bounded, subject-specific recommendations, typically suggesting \u223c10% basal increases with individualized meal-bolus updates.<\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec>\n                    <jats:title>Conclusions<\/jats:title>\n                    <jats:p>Temporal GAT-BiGRU provides accurate glucose prediction through temporal graph reasoning, sequence modeling, and interpretable explanations. It supports personalized decision support and closed-loop glucose management systems.<\/jats:p>\n                  <\/jats:sec>","DOI":"10.1093\/jamia\/ocag104","type":"journal-article","created":{"date-parts":[[2026,6,3]],"date-time":"2026-06-03T11:47:18Z","timestamp":1780487238000},"page":"1713-1725","source":"Crossref","is-referenced-by-count":0,"title":["GAT-BiGRU: explainable multi-task temporal graph learning for glucose forecasting, hypoglycemia risk, and counterfactual insulin adjustment"],"prefix":"10.1093","volume":"33","author":[{"ORCID":"https:\/\/orcid.org\/0000-0003-1173-0730","authenticated-orcid":false,"given":"Muhammad Abdullah","family":"Sarwar","sequence":"first","affiliation":[{"name":"Department of Software Engineering, Faculty of Informatics, Kaunas University of Technology , Kaunas, LT-51386,","place":["Lithuania"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Sarmad","family":"Maqsood","sequence":"additional","affiliation":[{"name":"Department of Software Engineering, Faculty of Informatics, Kaunas University of Technology , Kaunas, LT-51386,","place":["Lithuania"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Egle","family":"Belousovien\u0117","sequence":"additional","affiliation":[{"name":"Department of Intensive Care, Lithuanian University of Health Sciences , Kaunas, LT-50009,","place":["Lithuania"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Rytis","family":"Maskeli\u016bnas","sequence":"additional","affiliation":[{"name":"Department of Software Engineering, Faculty of Informatics, Kaunas University of Technology , Kaunas, LT-51386,","place":["Lithuania"]}],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"286","published-online":{"date-parts":[[2026,6,18]]},"reference":[{"key":"2026091614320584000_ocag104-B1","doi-asserted-by":"crossref","first-page":"1464755","DOI":"10.3389\/fendo.2024.1464755","article-title":"Flash glucose monitoring system help reduce the frequency of hypoglycemia and hypoglycemic fear behavior in type 1 diabetes patients","volume":"15","author":"Dong","year":"2024","journal-title":"Front Endocrinol (Lausanne)"},{"key":"2026091614320584000_ocag104-B2","doi-asserted-by":"crossref","first-page":"109","DOI":"10.1186\/s12902-025-01917-0","article-title":"Recurrent education: a promising strategy for enhancing diabetes management and reducing hypoglycemia in children with type 1 diabetes","volume":"25","author":"G\u00fcne\u015f Kaya","year":"2025","journal-title":"BMC Endocr Disord"},{"key":"2026091614320584000_ocag104-B3","first-page":"1792","article-title":"Hypoglycemia in diabetes: Challenges and strategies for optimal management","volume":"7","author":"Alabdulwahab","year":"2024","journal-title":"J Int Crisis Risk Commun Res"},{"key":"2026091614320584000_ocag104-B4","doi-asserted-by":"crossref","first-page":"1387537","DOI":"10.3389\/fendo.2024.1387537","article-title":"Recommendations for recognizing, risk stratifying, treating, and managing children and adolescents with hypoglycemia","volume":"15","author":"Zucchini","year":"2024","journal-title":"Front Endocrinol (Lausanne)"},{"key":"2026091614320584000_ocag104-B5","first-page":"3","article-title":"Continuous glucose monitoring versus self-monitoring for patients with type 1 diabetes: a systematic review and meta-analysis","volume":"49","author":"Poowaruttanawiwit","year":"2025","journal-title":"Thai J Pharm Sci"},{"key":"2026091614320584000_ocag104-B6","doi-asserted-by":"crossref","first-page":"112228","DOI":"10.1016\/j.diabres.2025.112228","article-title":"Continuous glucose monitoring as equinox of nocturnal and daytime hypoglycaemia in type 1 diabetes: insights from the randomized controlled hypode trial","volume":"224","author":"Hermanns","year":"2025","journal-title":"Diabetes Res Clin Pract"},{"key":"2026091614320584000_ocag104-B7","doi-asserted-by":"crossref","first-page":"770","DOI":"10.1007\/s13340-025-00846-6","article-title":"Association of impaired awareness of hypoglycemia and real-time continuous glucose monitoring with the occurrence and frequency of severe hypoglycemia in adults with type 1 diabetes mellitus: a real-world panel data analysis from the pr-iah study","volume":"16","author":"Sakane","year":"2025","journal-title":"Diabetol Int"},{"key":"2026091614320584000_ocag104-B8","doi-asserted-by":"crossref","first-page":"7","DOI":"10.1177\/19322968211042621","article-title":"Incorporating glucose variability into glucose forecasting accuracy assessment using the new glucose variability impact index and the prediction consistency index: an lstm case example","volume":"16","author":"Mosquera-Lopez","year":"2022","journal-title":"J Diabetes Sci Technol"},{"key":"2026091614320584000_ocag104-B9","doi-asserted-by":"crossref","first-page":"100253","DOI":"10.1016\/j.endmts.2025.100253","article-title":"Optimizing blood glucose predictions in type 1 diabetes patients using a stacking ensemble approach","volume":"18","author":"Liu","year":"2025","journal-title":"Endocr Metab Sci"},{"key":"2026091614320584000_ocag104-B10","doi-asserted-by":"crossref","first-page":"103409","DOI":"10.1016\/j.artmed.2026.103409","article-title":"Systematic review of artificial intelligence-based methods for glycemic control and risk prediction in intensive care units","volume":"176","author":"Sarwar","year":"2026","journal-title":"Artif Intell Med"},{"key":"2026091614320584000_ocag104-B11","doi-asserted-by":"crossref","first-page":"1590","DOI":"10.1177\/19322968221092785","article-title":"Long-term prediction of blood glucose levels in type 1 diabetes using a cnn-lstm-based deep neural network","volume":"17","author":"Jaloli","year":"2023","journal-title":"J Diabetes Sci Technol"},{"key":"2026091614320584000_ocag104-B12","doi-asserted-by":"crossref","first-page":"7670","DOI":"10.3390\/s24237670","article-title":"Hybrid cnn-gru model for real-time blood glucose forecasting: Enhancing iot-based diabetes management with ai","volume":"24","author":"Alkanhel","year":"2024","journal-title":"Sensors"},{"key":"2026091614320584000_ocag104-B13","doi-asserted-by":"crossref","first-page":"1605406","DOI":"10.3389\/fphys.2025.1605406","article-title":"Fusion-driven multimodal learning for biomedical time series in surgical care","volume":"16","author":"Che","year":"2025","journal-title":"Front Physiol"},{"key":"2026091614320584000_ocag104-B14","doi-asserted-by":"crossref","first-page":"4191","DOI":"10.1007\/s00521-020-05248-0","article-title":"Long short-term memory neural network for glucose prediction","volume":"33","author":"Carrillo-Moreno","year":"2021","journal-title":"Neural Comput Appl"},{"key":"2026091614320584000_ocag104-B15","doi-asserted-by":"crossref","first-page":"e1012738","DOI":"10.1371\/journal.pcbi.1012738","article-title":"Multi-region infectious disease prediction modeling based on spatio-temporal graph neural network and the dynamic model","volume":"21","author":"Wang","year":"2025","journal-title":"PLoS Comput Biol"},{"key":"2026091614320584000_ocag104-B16","doi-asserted-by":"crossref","first-page":"bbac266","DOI":"10.1093\/bib\/bbac266","article-title":"A deep learning method for predicting metabolite\u2013disease associations via graph neural network","volume":"23","author":"Sun","year":"2022","journal-title":"Brief Bioinform"},{"key":"2026091614320584000_ocag104-B17","doi-asserted-by":"crossref","first-page":"136085","DOI":"10.1016\/j.snb.2024.136085","article-title":"Generalized predictive analysis of reactions in paper devices via graph neural networks","volume":"417","author":"Sun","year":"2024","journal-title":"Sens Actuators B Chem"},{"key":"2026091614320584000_ocag104-B18","doi-asserted-by":"crossref","first-page":"112588","DOI":"10.1016\/j.patcog.2025.112588","article-title":"Dtfpnet: temporal and frequency dynamic graph neural network for time series classification","volume":"172","author":"Yu","year":"2026","journal-title":"Pattern Recognit"},{"key":"2026091614320584000_ocag104-B19","doi-asserted-by":"crossref","first-page":"408","DOI":"10.1186\/s12889-025-21618-6","article-title":"A novel graph neural network based approach for influenza-like illness nowcasting: exploring the interplay of temporal, geographical, and functional spatial features","volume":"25","author":"Luo","year":"2025","journal-title":"BMC Public Health"},{"key":"2026091614320584000_ocag104-B20","doi-asserted-by":"crossref","first-page":"612","DOI":"10.1038\/s41746-025-01994-4","article-title":"Generalized multi task learning framework for glucose forecasting and hypoglycemia detection using simulation to reality","volume":"8","author":"Hwang","year":"2025","journal-title":"NPJ Digit Med"},{"key":"2026091614320584000_ocag104-B21","doi-asserted-by":"crossref","first-page":"7416","DOI":"10.1111\/dom.70147","article-title":"Gluconet-mm: a multimodal attention-based multi-task learning framework with decision transformer for personalised and explainable blood glucose forecasting","volume":"27","author":"Maqsood","year":"2025","journal-title":"Diabetes Obes Metab"},{"key":"2026091614320584000_ocag104-B22","first-page":"14","author":"Sarwar","year":"2025"},{"key":"2026091614320584000_ocag104-B23","first-page":"333","author":"Maqsood","year":"2025"},{"key":"2026091614320584000_ocag104-B24","doi-asserted-by":"crossref","first-page":"33633","DOI":"10.1109\/JSEN.2024.3441715","article-title":"Hgmla: a multi-task learning model for assessment of hba1c and ga levels using short-term cgm sensor data","volume":"24","author":"Han","year":"2024","journal-title":"IEEE Sens J"},{"key":"2026091614320584000_ocag104-B25","doi-asserted-by":"crossref","first-page":"e0325337","DOI":"10.1371\/journal.pone.0325337","article-title":"Multi-task deep learning for predicting metabolic syndrome from retinal fundus images in a Japanese health checkup dataset","volume":"20","author":"Itoh","year":"2025","journal-title":"PLoS One"},{"key":"2026091614320584000_ocag104-B26","doi-asserted-by":"crossref","first-page":"38650","DOI":"10.1109\/ACCESS.2026.3672553","article-title":"Trustworthy glucose forecasting with physiology-constrained neural dynamics, conformal risk control, and risk-sensitive reinforcement learning","volume":"14","author":"Maqsood","year":"2026","journal-title":"IEEE Access"},{"key":"2026091614320584000_ocag104-B27","doi-asserted-by":"crossref","first-page":"105758","DOI":"10.1016\/j.ijmedinf.2024.105758","article-title":"Modeling the fasting blood glucose response to basal insulin adjustment in type 2 diabetes: an explainable machine learning approach on real-world data","volume":"195","author":"Thomsen","year":"2025","journal-title":"Int J Med Inform"},{"key":"2026091614320584000_ocag104-B28","first-page":"1","author":"Maqsood","year":"2025"},{"key":"2026091614320584000_ocag104-B29","doi-asserted-by":"crossref","first-page":"3989","DOI":"10.3390\/app12083989","article-title":"An optimization-based diabetes prediction model using cnn and bi-directional lstm in real-time environment","volume":"12","author":"Madan","year":"2022","journal-title":"Appl Sci"},{"key":"2026091614320584000_ocag104-B30","first-page":"605","author":"Saxena","year":"2024"},{"key":"2026091614320584000_ocag104-B31","doi-asserted-by":"crossref","first-page":"603","DOI":"10.1109\/JBHI.2019.2908488","article-title":"Convolutional recurrent neural networks for glucose prediction","volume":"24","author":"Li","year":"2020","journal-title":"IEEE J Biomed Health Inform"},{"key":"2026091614320584000_ocag104-B32","doi-asserted-by":"crossref","first-page":"102896","DOI":"10.1016\/j.bspc.2021.102896","article-title":"Graph convolutional network enabled two-stream learning architecture for diabetes classification based on flash glucose monitoring data","volume":"69","author":"Liu","year":"2021","journal-title":"Biomed Signal Process Control"},{"key":"2026091614320584000_ocag104-B33","first-page":"3727","article-title":"Dynamic partitioning of graphs based on multivariate blood glucose data\u2014a graph neural network model for diabetes prediction","volume":"13","author":"Li","year":"2024","journal-title":"Electronics (Basel)"},{"key":"2026091614320584000_ocag104-B34","doi-asserted-by":"crossref","first-page":"14916","DOI":"10.1109\/TIE.2024.3376798","article-title":"Towards noninvasive glucose monitoring based on bioimpedance grid sampling topology","volume":"71","author":"Liu","year":"2024","journal-title":"IEEE Trans Ind Electron"},{"key":"2026091614320584000_ocag104-B35","first-page":"72","author":"Minello","year":"2024"},{"key":"2026091614320584000_ocag104-B36","first-page":"1","article-title":"Self-attention-based graph transformation learning for anomaly detection in multivariate time series","volume":"11","author":"Wang","year":"2025","journal-title":"Complex Intell Syst"},{"key":"2026091614320584000_ocag104-B37","first-page":"71","article-title":"The ohiot1dm dataset for blood glucose level prediction: update 2020","volume":"2675","author":"Marling","year":"2020","journal-title":"CEUR Workshop Proceedings"},{"key":"2026091614320584000_ocag104-B38","article-title":"BrisT1D dataset: young adults with type 1 diabetes in the UK using Smartwatches","author":"Jam"},{"key":"2026091614320584000_ocag104-B39","doi-asserted-by":"crossref","first-page":"21856","DOI":"10.1038\/s41598-025-07706-y","article-title":"Deep learning for network security: an attention-cnn-lstm model for accurate intrusion detection","volume":"15","author":"Alashjaee","year":"2025","journal-title":"Sci Rep"},{"key":"2026091614320584000_ocag104-B40","first-page":"5998","article-title":"Attention is all you need","volume":"30","author":"Vaswani","year":"2017","journal-title":"Adv Neural Inf Process Syst"},{"key":"2026091614320584000_ocag104-B41","doi-asserted-by":"crossref","first-page":"042050","DOI":"10.1088\/1742-6596\/1213\/4\/042050","article-title":"Temporal convolutional networks for anomaly detection in time series","volume":"1213","author":"He","year":"2019","journal-title":"J Phys Conf Ser"},{"key":"2026091614320584000_ocag104-B42","doi-asserted-by":"crossref","first-page":"16865","DOI":"10.1038\/s41598-023-44155-x","article-title":"The importance of interpreting machine learning models for blood glucose prediction in diabetes: an analysis using shap","volume":"13","author":"Prendin","year":"2023","journal-title":"Sci Rep"},{"key":"2026091614320584000_ocag104-B43","doi-asserted-by":"crossref","first-page":"2515","DOI":"10.3934\/mbe.2024111","article-title":"A personalized multitasking framework for real-time prediction of blood glucose levels in type 1 diabetes patients","volume":"21","author":"Yang","year":"2024","journal-title":"Math Biosci Eng"},{"key":"2026091614320584000_ocag104-B44","doi-asserted-by":"crossref","first-page":"16","DOI":"10.1007\/s12046-024-02657-y","article-title":"Forecasting blood glucose level using convolutional recurrent connection","volume":"50","author":"Arora","year":"2025","journal-title":"S\u0101dhan\u0101"},{"key":"2026091614320584000_ocag104-B45","doi-asserted-by":"crossref","first-page":"30636","DOI":"10.1038\/s41598-025-13491-5","article-title":"Personalized blood glucose prediction in type 1 diabetes using meta-learning with bidirectional long short term memory-transformer hybrid model","volume":"15","author":"Moon","year":"2025","journal-title":"Sci Rep"},{"key":"2026091614320584000_ocag104-B46","doi-asserted-by":"crossref","first-page":"623","DOI":"10.3390\/a18100623","article-title":"Hybrid stochastic\u2013machine learning framework for postprandial glucose prediction in type 1 diabetes","volume":"18","author":"Naskinova","year":"2025","journal-title":"Algorithms"}],"container-title":["Journal of the American Medical Informatics Association"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/academic.oup.com\/jamia\/advance-article-pdf\/doi\/10.1093\/jamia\/ocag104\/68554678\/ocag104.pdf","content-type":"application\/pdf","content-version":"am","intended-application":"syndication"},{"URL":"https:\/\/academic.oup.com\/jamia\/article-pdf\/33\/9\/1713\/68554678\/ocag104.pdf","content-type":"application\/pdf","content-version":"vor","intended-application":"syndication"},{"URL":"https:\/\/academic.oup.com\/jamia\/article-pdf\/33\/9\/1713\/68554678\/ocag104.pdf","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,9,16]],"date-time":"2026-09-16T18:32:15Z","timestamp":1789583535000},"score":1,"resource":{"primary":{"URL":"https:\/\/academic.oup.com\/jamia\/article\/33\/9\/1713\/8711207"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026,6,18]]},"references-count":46,"journal-issue":{"issue":"9","published-online":{"date-parts":[[2026,6,18]]},"published-print":{"date-parts":[[2026,9,1]]}},"URL":"https:\/\/doi.org\/10.1093\/jamia\/ocag104","relation":{},"ISSN":["1067-5027","1527-974X"],"issn-type":[{"value":"1067-5027","type":"print"},{"value":"1527-974X","type":"electronic"}],"subject":[],"published-other":{"date-parts":[[2026,9]]},"published":{"date-parts":[[2026,6,18]]}}}