{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,10,8]],"date-time":"2026-10-08T14:24:45Z","timestamp":1791469485568,"version":"4.3.4"},"reference-count":28,"publisher":"MDPI AG","issue":"1","license":[{"start":{"date-parts":[[2025,12,19]],"date-time":"2025-12-19T00:00:00Z","timestamp":1766102400000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0\/"}],"content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":["Medicina"],"abstract":"<jats:p>Background and Objectives: Type 2 diabetes mellitus (T2DM) affects over 537 million adults worldwide and disproportionately burdens low- and middle-income countries, where diagnostic resources are limited. Predictive models trained in one population often fail to generalize across regions due to shifts in feature distributions and measurement practices, hindering scalable screening efforts. Materials and Methods: We evaluated a few-shot domain adaptation framework using a simple multilayer perceptron with four shared clinical features (age, body mass index, mean arterial pressure, and plasma glucose) across three adult cohorts: Bangladesh (n = 5288), Iraq (n = 662), and the Pima Indian dataset (n = 768). For each of the six source-target pairs, we pre-trained on the source cohort and then fine-tuned on 1, 5, 10, and 20% of the labeled target examples, reserving the remaining for testing; a final 20% few-shot version was compared with threshold tuning. Discrimination and calibration performance metrics were used before and after adaptation. SHAP explainability analyses quantified shifts in feature importance and decision thresholds. Results: Several source \u2192 target transfers produced zero true positives under the strict source-only baseline at a fixed 0.5 decision threshold (e.g., Bangladesh \u2192 Pima F1 = 0.00, 0\/268 diabetics detected). Few-shot fine-tuning restored non-zero recall in all such cases, with F1 improvements up to +0.63 and precision\u2013recall gains in every zero-baseline transfer. In directions with moderate baseline performance (e.g., Bangladesh \u2192 Iraq, Iraq \u2192 Pima, Pima \u2192 Iraq), 20% few-shot adaptation with threshold tuning improved AUROC by +0.01 to +0.14 and accuracy by +4 to +17 percentage points while reducing Brier scores by up to 0.14 and ECE by approximately 30\u201380% (suggesting improved calibration). All but one transfer (Iraq \u2192 Bangladesh) demonstrated statistically significant improvement by McNemar\u2019s test (p &lt; 0.001). SHAP analyses revealed population-specific threshold shifts: glucose inflection points ranged from ~120 mg\/dL in Pima to ~150 mg\/dL in Iraq, and the importance of BMI rose in Pima-targeted adaptations. Conclusions: Leveraging as few as 5\u201320% of local labels, few-shot domain adaptation enhances cross-population T2DM risk prediction using only routinely available features. This scalable, interpretable approach can democratize preventive screening in diverse, resource-constrained settings.<\/jats:p>","DOI":"10.3390\/medicina62010007","type":"journal-article","created":{"date-parts":[[2025,12,19]],"date-time":"2025-12-19T14:27:16Z","timestamp":1766154436000},"page":"7","update-policy":"https:\/\/doi.org\/10.3390\/mdpi_crossmark_policy","source":"Crossref","is-referenced-by-count":3,"title":["Few-Shot Transfer Learning for Diabetes Risk Prediction Across Global Populations"],"prefix":"10.3390","volume":"62","author":[{"ORCID":"https:\/\/orcid.org\/0000-0002-6180-7778","authenticated-orcid":false,"given":"Shrinit","family":"Babel","sequence":"first","affiliation":[{"name":"Morsani College of Medicine, University of South Florida, Tampa, FL 33602, USA"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0009-0000-8172-9618","authenticated-orcid":false,"given":"Sunit","family":"Babel","sequence":"additional","affiliation":[{"name":"Judy Genshaft Honors College, University of South Florida, Tampa, FL 33620, USA"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"John","family":"Hodgson","sequence":"additional","affiliation":[{"name":"Department of Anesthesiology and Perioperative Medicine, University of South Florida, Tampa, FL 33620, USA"},{"name":"TEAMHealth Anesthesia, Tampa General Hospital, Tampa, FL 33606, USA"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0003-3548-1097","authenticated-orcid":false,"given":"Enrico","family":"Camporesi","sequence":"additional","affiliation":[{"name":"Department of Anesthesiology and Perioperative Medicine, University of South Florida, Tampa, FL 33620, USA"},{"name":"TEAMHealth Anesthesia, Tampa General Hospital, Tampa, FL 33606, USA"}],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"1968","published-online":{"date-parts":[[2025,12,19]]},"reference":[{"key":"ref_1","doi-asserted-by":"crossref","unstructured":"Rodriguez-Saldana, J. (2019). The Dynamics of Diabetes Prevalence, Morbidity, and Mortality. The Diabetes Textbook: Clinical Principles, Patient Management and Public Health Issues, Springer International Publishing.","DOI":"10.1007\/978-3-030-11815-0"},{"key":"ref_2","doi-asserted-by":"crossref","first-page":"203","DOI":"10.1016\/S0140-6736(23)01301-6","article-title":"Global, regional, and national burden of diabetes from 1990 to 2021, with projections of prevalence to 2050: A systematic analysis for the Global Burden of Disease Study 2021","volume":"402","author":"Ong","year":"2023","journal-title":"Lancet"},{"key":"ref_3","first-page":"30","article-title":"Navigating the global challenge of diabetes mellitus: Insights from Kazakhstan\u2019s healthcare landscape and strategies for improved management","volume":"5","author":"Otegenova","year":"2024","journal-title":"Interdiscip. Approaches Med."},{"key":"ref_4","doi-asserted-by":"crossref","first-page":"154","DOI":"10.1016\/S2213-8587(21)00328-4","article-title":"Maximising use of population data on cardiometabolic disease","volume":"10","author":"Flood","year":"2022","journal-title":"Lancet Diabetes Endocrinol."},{"key":"ref_5","doi-asserted-by":"crossref","first-page":"103132","DOI":"10.1016\/j.artmed.2025.103132","article-title":"Advances in artificial intelligence for diabetes prediction: Insights from a systematic literature review","volume":"164","author":"Khokhar","year":"2025","journal-title":"Artif. Intell. Med."},{"key":"ref_6","doi-asserted-by":"crossref","first-page":"197","DOI":"10.1038\/s41746-023-00933-5","article-title":"A scoping review of artificial intelligence-based methods for diabetes risk prediction","volume":"6","author":"Mohsen","year":"2023","journal-title":"npj Digit. Med."},{"key":"ref_7","doi-asserted-by":"crossref","first-page":"109720","DOI":"10.1016\/j.compbiomed.2025.109720","article-title":"A robust and generalized framework in diabetes classification across heterogeneous environments","volume":"186","author":"Zhou","year":"2025","journal-title":"Comput. Biol. Med."},{"key":"ref_8","doi-asserted-by":"crossref","first-page":"19","DOI":"10.1016\/S2213-8587(13)70103-7","article-title":"Non-invasive risk scores for prediction of type 2 diabetes (EPIC-InterAct): A validation of existing models","volume":"2","author":"Kengne","year":"2014","journal-title":"Lancet Diabetes Endocrinol."},{"key":"ref_9","doi-asserted-by":"crossref","first-page":"1075","DOI":"10.3945\/ajcn.2009.28009","article-title":"Generalizability of dietary patterns associated with incidence of type 2 diabetes mellitus","volume":"90","author":"Imamura","year":"2009","journal-title":"Am. J. Clin. Nutr."},{"key":"ref_10","doi-asserted-by":"crossref","unstructured":"Vettoretti, M., Longato, E., Camillo, B.D., and Facchinetti, A. (2018, January 18\u201321). Importance of Recalibrating Models for Type 2 Diabetes Onset Prediction: Application of the Diabetes Population Risk Tool on the Health and Retirement Study. Proceedings of the 2018 40th Annual International Conference of the IEEE Engineering in Medicine and Biology Society (EMBC), Honolulu, HI, USA.","DOI":"10.1109\/EMBC.2018.8513554"},{"key":"ref_11","doi-asserted-by":"crossref","first-page":"e42615","DOI":"10.2196\/42615","article-title":"Digital Health Data Quality Issues: Systematic Review","volume":"25","author":"Syed","year":"2023","journal-title":"J. Med. Internet Res."},{"key":"ref_12","doi-asserted-by":"crossref","first-page":"26","DOI":"10.1038\/s41746-023-00773-3","article-title":"The impact of inconsistent human annotations on AI driven clinical decision making","volume":"6","author":"Sylolypavan","year":"2023","journal-title":"npj Digit. Med."},{"key":"ref_13","unstructured":"Kouw, W.M., and Loog, M. (2019). An introduction to domain adaptation and transfer learning. arXiv."},{"key":"ref_14","doi-asserted-by":"crossref","unstructured":"Ebbehoj, A., Thunbo, M.\u00d8., Andersen, O.E., Glindtvad, M.V., and Hulman, A. (2022). Transfer learning for non-image data in clinical research: A scoping review. PLoS Digit. Health, 1.","DOI":"10.1371\/journal.pdig.0000014"},{"key":"ref_15","unstructured":"Gardner, J.P., Perdomo, J.C., and Schmidt, L. (2025, June 19). Large Scale Transfer Learning for Tabular Data via Language Modeling. Available online: https:\/\/openreview.net\/forum?id=WH5blx5tZ1&referrer=%5Bthe+profile+of+Ludwig+Schmidt%5D%28%2Fprofile%3Fid%3D~Ludwig_Schmidt1%29&utm_source=chatgpt.com."},{"key":"ref_16","unstructured":"Lee, S., Seo, S., Kim, J., Lee, Y., and Hwang, S. (2023). Few-shot Fine-tuning is All You Need for Source-free Domain Adaptation. arXiv."},{"key":"ref_17","unstructured":"Wornow, M., Thapa, R., Steinberg, E., Fries, J.A., and Shah, N.H. (2023). EHRSHOT: An EHR Benchmark for Few-Shot Evaluation of Foundation Models. arXiv."},{"key":"ref_18","doi-asserted-by":"crossref","unstructured":"Nalmpatian, A., Heumann, C., Alkaya, L., and Jackson, W. (2024). Transfer learning for mortality risk: A case study on the United Kingdom. PLoS ONE, 20.","DOI":"10.1101\/2024.10.25.24316112"},{"key":"ref_19","unstructured":"Raheem, S.A., Taha, A., and Hamarash, I.I. (2024). Erbil Diabetes Dataset. Mendeley Data."},{"key":"ref_20","doi-asserted-by":"crossref","first-page":"111746","DOI":"10.1016\/j.dib.2025.111746","article-title":"DiaBD: A Diabetes Dataset for Enhanced Risk Analysis and Research in Bangladesh","volume":"61","author":"Prama","year":"2025","journal-title":"Data Brief."},{"key":"ref_21","unstructured":"Joseph, L.P., Joseph, E.A., and Prasad, R. (2022). Diabetes Datasets. Mendeley Data."},{"key":"ref_22","doi-asserted-by":"crossref","first-page":"1317","DOI":"10.1001\/jama.2017.18391","article-title":"Big Data and Machine Learning in Health Care","volume":"319","author":"Beam","year":"2018","journal-title":"JAMA"},{"key":"ref_23","first-page":"345","article-title":"From development to deployment: Dataset shift, causality, and shift-stable models in health AI","volume":"21","author":"Subbaswamy","year":"2020","journal-title":"Biostatistics"},{"key":"ref_24","doi-asserted-by":"crossref","first-page":"2516414","DOI":"10.1109\/TIM.2021.3122172","article-title":"Incremental Cross-Domain Adaptation for Robust Retinopathy Screening via Bayesian Deep Learning","volume":"70","author":"Hassan","year":"2021","journal-title":"IEEE Trans. Instrum. Meas."},{"key":"ref_25","doi-asserted-by":"crossref","first-page":"822","DOI":"10.1007\/s00125-006-0189-4","article-title":"The threshold for diagnosing impaired fasting glucose: A position statement by the European Diabetes Epidemiology Group","volume":"49","author":"Edeg","year":"2006","journal-title":"Diabetologia"},{"key":"ref_26","doi-asserted-by":"crossref","first-page":"45","DOI":"10.1186\/1475-2840-13-45","article-title":"Thresholds of various glycemic measures for diagnosing diabetes based on prevalence of retinopathy in community-dwelling Japanese subjects: The Hisayama Study","volume":"13","author":"Mukai","year":"2014","journal-title":"Cardiovasc. Diabetol."},{"key":"ref_27","doi-asserted-by":"crossref","first-page":"109","DOI":"10.1038\/s41746-021-00480-x","article-title":"Deep transfer learning and data augmentation improve glucose levels prediction in type 2 diabetes patients","volume":"4","author":"Deng","year":"2021","journal-title":"npj Digit. Med."},{"key":"ref_28","doi-asserted-by":"crossref","first-page":"105874","DOI":"10.1016\/j.cmpb.2020.105874","article-title":"Adversarial multi-source transfer learning in healthcare: Application to glucose prediction for diabetic people","volume":"199","author":"Ammi","year":"2021","journal-title":"Comput. Methods Programs Biomed."}],"container-title":["Medicina"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/www.mdpi.com\/1648-9144\/62\/1\/7\/pdf","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2025,12,19]],"date-time":"2025-12-19T14:39:56Z","timestamp":1766155196000},"score":1,"resource":{"primary":{"URL":"https:\/\/www.mdpi.com\/1648-9144\/62\/1\/7"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2025,12,19]]},"references-count":28,"journal-issue":{"issue":"1","published-online":{"date-parts":[[2026,1]]}},"alternative-id":["medicina62010007"],"URL":"https:\/\/doi.org\/10.3390\/medicina62010007","relation":{},"ISSN":["1648-9144"],"issn-type":[{"value":"1648-9144","type":"electronic"}],"subject":[],"published":{"date-parts":[[2025,12,19]]}}}