Skip to content
DraconiticResearch
Search Sources API
226,742 studies31 sourcesUpdated 48 min ago
Peer-reviewedCC BYFrontiers in artificial intelligence

From task-specific networks to foundation models in diabetes care: a critical review of opportunities, evaluation gaps, and safety frontiers

  1. 1School of Computer Science Engineering and Information Systems, Vellore Institute of Technology , , ,

Abstract

Artificial intelligence in diabetes care has historically relied on task-specific supervised learning, such as bespoke networks for retinopathy grading, kidney disease prediction, and glucose forecasting. The emergence of foundation models (FMs: large models pretrained on broad data, typically with self-supervision, and adapted to many downstream tasks) is displacing this paradigm across diabetes-relevant modalities: ophthalmic models such as RETFound, pretrained on 1.6 million retinal photographs, have been reported to outperform task-specific networks on retinopathy grading and systemic disease prediction, especially when labeled data are scarce; transformer-based time-series models are replacing long short-term memory (LSTM) pipelines for continuous glucose monitoring (CGM) forecasting; clinical large language models (LLMs) such as Med-PaLM and GPT-4 attain expert-level scores on medical question-answering benchmarks; and generative models supply synthetic CGM and electronic health record (EHR) data. This review provides a critical, modality-organized synthesis of this transition. We (i) formalize a three-era taxonomy: rule-based risk scores (era I), task-specific deep learning (era II, 2016-2020), and foundation models (era III, from 2021), and map the diabetes literature onto it; (ii) survey foundation-model architectures, pretraining objectives, and adaptation strategies across retinal imaging, CGM, longitudinal EHR, and multimodal data; (iii) expose an evaluation crisis: benchmark contamination of LLM medical evaluations, a near-universal absence in the literature we reviewed of external and prospective validation for retinal and CGM foundation models, miscalibration under dataset shift, and, to our knowledge and within the reviewed literature (January 2016-December 2025; negative findings are time-sensitive), the absence of any randomized evaluation of a foundation model in a diabetes workflow; and (iv) analyze safety risks: hallucinated insulin-dosing advice, automation bias, inherited algorithmic inequity, and privacy exposure during adaptation. We conclude that the decisive open problem is no longer representation learning but grounded, calibrated, causally informed deployment, requiring contamination-resistant evaluation, prospective trials, and hybrid architectures coupling foundation-model representations with treatment-effect estimation.

Related

  1. Assessing Person-Specific Symptom Networks From Brief Narratives: Development and Initial Evaluation of the Narrative-to-Network (N2N) Approach
    Felix Vogel, Carla Paulus, Julian Reichert +2
    PsyArXivPreprintCC BY 4.0Full textPsychiatry & mental healthOSF
  2. Post-Translational Modification Crosstalk in Gastric Cancer: Molecular Switches, Signaling Networks, and Therapeutic Opportunities, with Comparative Insights from Colorectal Cancer
    Jun Wang, Xianying Chen, Yu Wang +1
    CancersPeer-reviewedCC BYFull textOncologyEurope PMC
  3. From Task-Specific AI to Cardiovascular Foundation Models: A New Era of Multimodal Interpretation
    Hirotaka Ieki, Paul Cheng, Andrew P. Ambrosy +5
    Heart (British Cardiac Society)Peer-reviewedLicence not statedFull textEpidemiology & public healthPMC
  4. Benchmarking time series foundation models for multi-country infectious disease forecasting: a retrospective evaluation across over 70 diseases
    Peng Cai, Guoyu Huang, Ximing Xu
    BMC Public HealthPeer-reviewedCC BY-NC-ND 4.0Epidemiology & public healthCrossref
  5. The neuro–metabolic–immune axis in breast cancer: emerging mechanisms and subtype-specific therapeutic opportunities
    Nannan Xia, Renhong Huang, Hanyi Zhong +4
    Frontiers in ImmunologyPeer-reviewedCC BY 4.0Full textPsychiatry & mental healthCrossrefEurope PMCPMCEurope PMC
BibTeX
@article{kishore2026from,
  title = {{From task-specific networks to foundation models in diabetes care: a critical review of opportunities, evaluation gaps, and safety frontiers}},
  author = {M. Nanda Kishore and C. Navaneethan},
  journal = {Frontiers in artificial intelligence},
  year = {2026},
  month = oct,
  doi = {10.3389/frai.2026.1983619},
  url = {https://europepmc.org/article/PMC/PMC13752723},
  note = {CC BY},
}
RIS
TY  - JOUR
TI  - From task-specific networks to foundation models in diabetes care: a critical review of opportunities, evaluation gaps, and safety frontiers
AU  - M. Nanda Kishore
AU  - C. Navaneethan
PY  - 2026
DA  - 2026/10/28
JO  - Frontiers in artificial intelligence
DO  - 10.3389/frai.2026.1983619
UR  - https://europepmc.org/article/PMC/PMC13752723
AB  - Artificial intelligence in diabetes care has historically relied on task-specific supervised learning, such as bespoke networks for retinopathy grading, kidney disease prediction, and glucose forecasting. The emergence of foundation models (FMs: large models pretrained on broad data, typically with self-supervision, and adapted to many downstream tasks) is displacing this paradigm across diabetes-relevant modalities: ophthalmic models such as RETFound, pretrained on 1.6 million retinal photographs, have been reported to outperform task-specific networks on retinopathy grading and systemic disease prediction, especially when labeled data are scarce; transformer-based time-series models are replacing long short-term memory (LSTM) pipelines for continuous glucose monitoring (CGM) forecasting; clinical large language models (LLMs) such as Med-PaLM and GPT-4 attain expert-level scores on medical question-answering benchmarks; and generative models supply synthetic CGM and electronic health record (EHR) data. This review provides a critical, modality-organized synthesis of this transition. We (i) formalize a three-era taxonomy: rule-based risk scores (era I), task-specific deep learning (era II, 2016-2020), and foundation models (era III, from 2021), and map the diabetes literature onto it; (ii) survey foundation-model architectures, pretraining objectives, and adaptation strategies across retinal imaging, CGM, longitudinal EHR, and multimodal data; (iii) expose an evaluation crisis: benchmark contamination of LLM medical evaluations, a near-universal absence in the literature we reviewed of external and prospective validation for retinal and CGM foundation models, miscalibration under dataset shift, and, to our knowledge and within the reviewed literature (January 2016-December 2025; negative findings are time-sensitive), the absence of any randomized evaluation of a foundation model in a diabetes workflow; and (iv) analyze safety risks: hallucinated insulin-dosing advice, automation bias, inherited algorithmic inequity, and privacy exposure during adaptation. We conclude that the decisive open problem is no longer representation learning but grounded, calibrated, causally informed deployment, requiring contamination-resistant evaluation, prospective trials, and hybrid architectures coupling foundation-model representations with treatment-effect estimation.
SN  - 2624-8212
ER  -