pith. machine review for the scientific record. sign in

arxiv: 2605.11247 · v1 · submitted 2026-05-11 · 💻 cs.LG

Recognition: 2 theorem links

· Lean Theorem

A Proof-of-Concept Simulation-Driven Digital Twin Framework for Decision-Aware Diabetes Modeling

Zarrin Monirzadeh

Authors on Pith no claims yet

Pith reviewed 2026-05-13 01:54 UTC · model grok-4.3

classification 💻 cs.LG
keywords digital twindiabetes modelingcounterfactual simulationdecision-aware analysiscontinuous glucose monitoringsynthetic data augmentationproof-of-concept framework
0
0 comments X

The pith

A proof-of-concept digital twin framework integrates prediction with counterfactual simulation to support decision-aware diabetes modeling using benchmark and synthetic data.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper sets out to demonstrate that a simulation-driven digital twin can combine standard predictive modeling with the generation of alternative trajectories under different interventions. A sympathetic reader would care because this moves beyond forecasting future glucose levels to exploring how specific decisions might change those levels over time. The work relies on public clinical datasets supplemented by controlled synthetic temporal augmentation and applies the approach to illustrative continuous glucose monitoring examples. It presents this integration as feasible while explicitly avoiding any claim of clinical validation or readiness.

Core claim

The paper establishes that a digital twin framework built on benchmark clinical data and synthetic temporal augmentation can generate interpretable simulated trajectories by merging predictive models with counterfactual simulation, thereby enabling decision-aware analysis of intervention effects in diabetes without asserting clinical accuracy.

What carries the argument

The simulation-driven digital twin framework, which merges predictive modeling with counterfactual simulation to produce decision-aware glucose trajectories from mixed real and synthetic data.

If this is right

  • Prediction models become usable for generating alternative outcome paths rather than single forecasts.
  • Intervention effects can be illustrated through controlled synthetic scenarios overlaid on real data patterns.
  • The approach supplies a reusable structure for future simulation-driven systems in healthcare.
  • Temporal behavior and decision impacts become visible in continuous glucose monitoring examples.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The same structure could be tested on other chronic conditions that rely on time-series physiological data.
  • Linking the twin directly to live sensor streams might allow ongoing updates to the simulated trajectories.
  • Quantitative comparison against long-term outcome registries would be required to assess how well the synthetic augmentations match reality.

Load-bearing premise

Benchmark clinical data plus controlled synthetic temporal augmentation can produce trajectories representative enough for meaningful decision-aware counterfactual analysis.

What would settle it

Demonstration that real patient glucose responses to the same interventions diverge substantially from the framework's simulated counterfactual trajectories.

Figures

Figures reproduced from arXiv: 2605.11247 by Zarrin Monirzadeh.

Figure 1
Figure 1. Figure 1: Overview of artificial intelligence approaches in diabetes, highlighting the transition from prediction-focused models toward decision-oriented and simulation-based frameworks. Despite this progress, an important limitation remains. Most existing studies focus on prediction, classification, or risk scoring. While these approaches estimate future outcomes, they typically do not evaluate alternative actions … view at source ↗
Figure 2
Figure 2. Figure 2: Proposed digital twin architecture illustrating the flow from data ingestion and preprocessing to latent state representation, predictive modeling, and counterfactual simulation for intervention analysis. 4.3. Data Layer The input layer integrates heterogeneous variables across three categories: clinical, physiological, and behavioral. Clin￾ical variables include demographic information, diagnosis type, me… view at source ↗
Figure 4
Figure 4. Figure 4: Illustration of the decision-support workflow within the digital twin framework. The system evaluates multiple intervention scenarios and generates predicted glucose trajectories, enabling comparison of outcomes and ranking of candidate interventions. This figure illustrates the conceptual workflow of the decision-support process. Quantitative results derived from [PITH_FULL_IMAGE:figures/full_fig_p004_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Implementation workflow showing data preprocessing, model training, evaluation, and generation of simulated intervention scenarios for decision-support analysis. The full implementation, including reproducible code, gen￾erated datasets, experimental outputs, and visualization assets, is publicly available in a structured repository [41]. 5. Materials and Experimental Methods 5.1. Proof-of-Concept Objective… view at source ↗
Figure 6
Figure 6. Figure 6: Example 24-hour real-world continuous glucose monitoring (CGM) trajectory extracted from the OhioT1DM dataset, illustrating realistic temporal glucose dynamics. 6. Results 6.1. Benchmark Regression Performance [PITH_FULL_IMAGE:figures/full_fig_p007_6.png] view at source ↗
Figure 7
Figure 7. Figure 7: illustrates the simulated glucose trajectories under different intervention scenarios, while [PITH_FULL_IMAGE:figures/full_fig_p007_7.png] view at source ↗
Figure 8
Figure 8. Figure 8: Illustrative counterfactual trajectories derived from real CGM data. Alternative curves represent estimated effects of reduced carbohydrate intake and postprandial activity for demonstration purposes. The reduced-carbohydrate scenario yields the largest de￾crease in peak glucose, lowering the maximum value from 179 mg/dL to 153 mg/dL and improving time-in-range from 58% to 72%. The addition of postprandial… view at source ↗
read the original abstract

This paper presents a proof-of-concept digital twin framework for simulation-driven diabetes modeling using benchmark clinical data, synthetic temporal augmentation, and illustrative continuous glucose monitoring (CGM) analysis. Unlike traditional predictive models, the framework focuses on generating interpretable simulated trajectories rather than clinically validated outcomes. Evaluation is conducted using a public dataset combined with controlled synthetic scenarios to illustrate temporal behavior and intervention effects. Results illustrate the feasibility of integrating prediction with counterfactual simulation for decision-aware analysis. This work does not claim clinical readiness but provides a foundation for future research on simulation-driven digital twin systems in healthcare.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

1 major / 2 minor

Summary. The manuscript presents a proof-of-concept simulation-driven digital twin framework for diabetes modeling. It combines public benchmark clinical data with controlled synthetic temporal augmentation to generate illustrative continuous glucose monitoring (CGM) trajectories, with the goal of integrating predictive modeling and counterfactual simulation to support decision-aware analysis. The work explicitly disclaims clinical validation or readiness and positions itself as a feasibility demonstration and foundation for future research.

Significance. If the synthetic trajectories can be shown to remain statistically representative of real patient CGM data, the framework could provide a useful simulation platform for exploring intervention effects without requiring new clinical trials. The current illustrative results, however, supply no quantitative support for this representativeness, limiting the work to a conceptual starting point rather than a substantive advance in digital-twin methodology for healthcare.

major comments (1)
  1. [Evaluation] Evaluation section (as described in the abstract and results): The central claim that the framework illustrates feasibility of decision-aware counterfactual analysis rests on the premise that benchmark data plus controlled synthetic temporal augmentation produces trajectories sufficiently representative for meaningful analysis. No quantitative fidelity metrics are supplied (e.g., preservation of glucose variability, autocorrelation structure, or time-in-range distributions) to confirm that the generated trajectories lie within the statistical envelope of real CGM data rather than reflecting artifacts of the augmentation rules.
minor comments (2)
  1. [Abstract] The abstract and introduction could more explicitly delineate the predictive model, the augmentation procedure, and the counterfactual generation steps so that readers can assess the integration claimed in the title.
  2. [Methods] A brief discussion of the specific public dataset employed and any preprocessing steps would improve reproducibility of the illustrative scenarios.

Simulated Author's Rebuttal

1 responses · 0 unresolved

We thank the referee for their constructive feedback on our proof-of-concept manuscript. The evaluation concern is addressed point by point below.

read point-by-point responses

  1. Referee: [Evaluation] Evaluation section (as described in the abstract and results): The central claim that the framework illustrates feasibility of decision-aware counterfactual analysis rests on the premise that benchmark data plus controlled synthetic temporal augmentation produces trajectories sufficiently representative for meaningful analysis. No quantitative fidelity metrics are supplied (e.g., preservation of glucose variability, autocorrelation structure, or time-in-range distributions) to confirm that the generated trajectories lie within the statistical envelope of real CGM data rather than reflecting artifacts of the augmentation rules.

    Authors: We agree that the manuscript does not provide quantitative fidelity metrics to assess how well the synthetic trajectories preserve statistical properties of the benchmark CGM data. Given the explicit positioning of the work as a proof-of-concept feasibility demonstration (rather than a clinically validated system), the original focus was on illustrating the integration of benchmark data, controlled augmentation, and counterfactual simulation. However, we recognize that adding such metrics would strengthen the support for the framework's utility. In the revised manuscript, we will include quantitative comparisons of key CGM characteristics, such as glucose variability (standard deviation and coefficient of variation), time-in-range distributions, and autocorrelation structure, between the original benchmark data and the augmented trajectories. These additions will help demonstrate that the generated trajectories remain within a reasonable statistical envelope of real data. revision: yes

Circularity Check

0 steps flagged

No circularity; framework is self-contained using external public data and synthetic generation without derivations or self-referential fits.

full rationale

The paper is a proof-of-concept description of a digital twin framework that combines benchmark clinical data with controlled synthetic temporal augmentation to illustrate simulated trajectories and counterfactuals. No equations, parameter fittings, or derivation chains are present that would reduce any prediction or result to the inputs by construction. The work explicitly positions itself as non-clinically validated and illustrative, relying on existing public datasets rather than any self-citation load-bearing uniqueness theorems or ansatz smuggling. This matches the default expectation of a non-circular simulation study.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

Abstract provides no explicit free parameters, axioms, or invented entities with independent evidence; the digital twin is presented as a modeling approach rather than a new postulated physical entity.

pith-pipeline@v0.9.0 · 5384 in / 1029 out tokens · 48790 ms · 2026-05-13T01:54:17.602519+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Lean theorems connected to this paper

Citations machine-checked in the Pith Canon. Every link opens the source theorem in the public Lean library.

Reference graph

Works this paper leans on

41 extracted references · 41 canonical work pages

  1. [1]

    Artificial intelligence for diabetes management and decision support: literature review,

    I. Contreras and J. Vehí, “Artificial intelligence for diabetes management and decision support: literature review,”J. Diabetes Sci. Technol., vol. 12, no. 2, pp. 456–461, 2018

    work page 2018

  2. [2]

    Machine learning and data mining methods in diabetes research,

    I. Kavakiotis et al., “Machine learning and data mining methods in diabetes research,”Comput. Struct. Biotechnol. J., vol. 15, pp. 104–116, 2017

    work page 2017

  3. [3]

    Continuous glucose monitoring combined with artificial intelligence: redefining the pathway for prediabetes management,

    C. Ji, T. Jiang, L. Liu, J. Zhang, and L. You, “Continuous glucose monitoring combined with artificial intelligence: redefining the pathway for prediabetes management,”Frontiers in Endocrinology, vol. 16, Art. no. 1571362, 2025, doi: 10.3389/fendo.2025.1571362

    work page doi:10.3389/fendo.2025.1571362 2025

  4. [4]

    Digital twins in Type 1 diabetes: A systematic review,

    G. Cappon and A. Facchinetti, “Digital twins in Type 1 diabetes: A systematic review,”Journal of Diabetes Science and Technology, vol. 19, no. 6, pp. 1641–1649, Nov. 2025, doi: 10.1177/19322968241262112

    work page doi:10.1177/19322968241262112 2025

  5. [5]

    Digital twins and artificial intelligence in metabolic disease research,

    C. Mosquera-Lopez and P. G. Jacobs, “Digital twins and artificial intelligence in metabolic disease research,”Trends in Endocrinol- ogy & Metabolism, vol. 35, no. 6, pp. 549–557, Jun. 2024, doi: 10.1016/j.tem.2024.04.019

    work page doi:10.1016/j.tem.2024.04.019 2024

  6. [6]

    Digital twins in healthcare: a comprehensive review and future directions,

    H. Khoshfekr Rudsari, B. Tseng, H. Zhu, L. Song, C. Gu, A. Roy, E. Irajizad, J. Butner, J. Long, and K.-A. Do, “Digital twins in healthcare: a comprehensive review and future directions,”Frontiers in Digital Health, vol. 7, Art. no. 1633539, 2025, doi: 10.3389/fdgth.2025.1633539

    work page doi:10.3389/fdgth.2025.1633539 2025

  7. [7]

    Digital twins in healthcare: Methodological challenges and opportunities,

    C. Meijer, H.-W. Uh, and S. El Bouhaddani, “Digital twins in healthcare: Methodological challenges and opportunities,”Journal of Personalized Medicine, vol. 13, no. 10, Art. no. 1522, 2023, doi: 10.3390/jpm13101522

    work page doi:10.3390/jpm13101522 2023

  8. [8]

    Personal- izing computational models to construct medical digital twins,

    A. Knapp, D. A. Cruz, B. Mehrad, and R. C. Laubenbacher, “Personal- izing computational models to construct medical digital twins,”Journal of the Royal Society Interface, vol. 22, no. 228, Art. no. 20250055, Jul. 2025, doi: 10.1098/rsif.2025.0055

    work page doi:10.1098/rsif.2025.0055 2025

  9. [9]

    Machine learning- based glucose prediction with use of continuous glucose and physical 10 activity monitoring data: The Maastricht Study,

    W. P. T. M. van Doorn, Y . D. Foreman, N. C. Schaper, H. H. C. M. Savel- berg, A. Koster, C. J. H. van der Kallen, A. Wesselius, M. T. Schram, R. M. A. Henry, P. C. Dagnelie, B. E. de Galan, O. Bekers, C. D. A. Stehouwer, S. J. R. Meex, and M. C. G. J. Brouwers, “Machine learning- based glucose prediction with use of continuous glucose and physical 10 act...

    work page doi:10.1371/journal.pone.0253125 2021

  10. [10]

    A hybrid Transformer-LSTM model apply to glucose prediction,

    Q. Bian, A. As’arry, X. Cong, K. A. bin Md Rezali, and R. M. K. bin Raja Ahmad, “A hybrid Transformer-LSTM model apply to glucose prediction,”PLOS ONE, vol. 19, no. 9, Art. no. e0310084, Sep. 2024, doi: 10.1371/journal.pone.0310084

    work page doi:10.1371/journal.pone.0310084 2024

  11. [11]

    A personalized federated learning-based glucose predic- tion algorithm for high-risk glycemic excursion regions in Type 1 diabetes,

    D. Darpit, K. Vyas, J. K. Jayagopal, A. Garcia, M. Erraguntla, and M. Lawley, “A personalized federated learning-based glucose predic- tion algorithm for high-risk glycemic excursion regions in Type 1 diabetes,”Scientific Reports, vol. 15, no. 1, Art. no. 38376, 2025, doi: 10.1038/s41598-025-22316-4

    work page doi:10.1038/s41598-025-22316-4 2025

  12. [12]

    Interpretable glucose forecasting for Type 2 diabetes across traditional, deep, and large language models,

    R. Alredaini, M. Abulkhair, and H. Almisbahi, “Interpretable glucose forecasting for Type 2 diabetes across traditional, deep, and large language models,”Scientific Reports, vol. 16, Art. no. 2421, 2025, doi: 10.1038/s41598-025-32373-4

    work page doi:10.1038/s41598-025-32373-4 2025

  13. [13]

    Technical, ethical, legal, and societal challenges with digital twin systems for the management of chronic diseases in children and young people,

    D. Drummond and A. Coulet, “Technical, ethical, legal, and societal challenges with digital twin systems for the management of chronic diseases in children and young people,”Journal of Medical Internet Research, vol. 24, no. 10, Art. no. e39698, 2022, doi: 10.2196/39698

    work page doi:10.2196/39698 2022

  14. [14]

    S. L. Cichosz, S. S. Olesen, and M. H. Jensen, “Explainable machine- learning models to predict weekly risk of hyperglycemia, hypo- glycemia, and glycemic variability in patients with Type 1 dia- betes based on continuous glucose monitoring,”Journal of Diabetes Science and Technology, vol. 20, no. 3, pp. 836–847, 2024, doi: 10.1177/19322968241286907

    work page doi:10.1177/19322968241286907 2024

  15. [15]

    Detection of viola- tions in credit cards of banks and financial institutions based on artificial neural network and metaheuristic optimization algorithm,

    Z. Monirzadeh, M. Habibzadeh, and N. Farajian, “Detection of viola- tions in credit cards of banks and financial institutions based on artificial neural network and metaheuristic optimization algorithm,”International Journal of Advanced Computer Science and Applications, vol. 9, no. 1, pp. 176–182, 2018, doi: 10.14569/IJACSA.2018.090124

    work page doi:10.14569/ijacsa.2018.090124 2018

  16. [16]

    Available: https://doi.org/10.1038/s41591-018-0300-7

    E. J. Topol, “High-performance medicine: the convergence of human and artificial intelligence,”Nature Medicine, vol. 25, no. 1, pp. 44–56, 2019, doi: 10.1038/s41591-018-0300-7

    work page doi:10.1038/s41591-018-0300-7 2019

  17. [17]

    Scalable and accurate deep learning with electronic health records,

    A. Rajkomar et al., “Scalable and accurate deep learning with electronic health records,”npj Digital Medicine, vol. 1, Art. no. 18, 2018, doi: 10.1038/s41746-018-0029-1

    work page doi:10.1038/s41746-018-0029-1 2018

  18. [18]

    2009.Causality(2 ed.)

    J. Pearl,Causality: Models, Reasoning, and Inference, 2nd ed. Cambridge, UK: Cambridge University Press, 2009, doi: 10.1017/CBO9780511803161

    work page doi:10.1017/cbo9780511803161 2009

  19. [19]

    M. A. Hernán and J. M. Robins,Causal Inference: What If. Boca Raton, FL, USA: Chapman & Hall/CRC, 2020

    work page 2020

  20. [20]

    The ‘digital twin’ to enable precision cardiol- ogy,

    M. Corral-Acero et al., “The ‘digital twin’ to enable precision cardiol- ogy,”European Heart Journal, vol. 41, no. 48, pp. 4556–4564, 2020, doi: 10.1093/eurheartj/ehaa159

    work page doi:10.1093/eurheartj/ehaa159 2020

  21. [21]

    Digital twins in health care: ethical implications of an emerging engineering paradigm,

    K. Bruynseels, F. Santoni de Sio, and J. van den Hoven, “Digital twins in health care: ethical implications of an emerging engineering paradigm,”Frontiers in Genetics, vol. 9, Art. no. 31, 2018, doi: 10.3389/fgene.2018.00031

    work page doi:10.3389/fgene.2018.00031 2018

  22. [22]

    In silico clinical trials: how computer simulation will transform the biomedical industry,

    M. Viceconti, A. Henney, and E. Morley-Fletcher, “In silico clinical trials: how computer simulation will transform the biomedical industry,” Int. J. Clin. Trials, vol. 3, no. 2, pp. 37–46, 2016, doi: 10.18203/2349- 3259.ijct20161408

    work page doi:10.18203/2349- 2016

  23. [23]

    Personalized blood glucose prediction for Type 1 diabetes using evidential deep learning and meta- learning,

    T. Zhu, K. Li, P. Herrero, and P. Georgiou, “Personalized blood glucose prediction for Type 1 diabetes using evidential deep learning and meta- learning,”IEEE Transactions on Biomedical Engineering, vol. 69, no. 10, pp. 3107–3118, 2022, doi: 10.1109/TBME.2022.3150729

    work page doi:10.1109/tbme.2022.3150729 2022

  24. [24]

    Deep multi- output forecasting: learning to accurately predict blood glucose trajec- tories,

    I. Fox, L. Ang, M. Jaiswal, R. Pop-Busui, and J. Wiens, “Deep multi- output forecasting: learning to accurately predict blood glucose trajec- tories,” inProc. 24th ACM SIGKDD Int. Conf. Knowl. Discovery Data Mining (KDD), 2018, pp. 1387–1395, doi: 10.1145/3219819.3220102

    work page doi:10.1145/3219819.3220102 2018

  25. [25]

    Dilated recurrent neural networks for glucose forecasting in Type 1 diabetes,

    T. Zhu, K. Li, J. Chen, P. Herrero, and P. Georgiou, “Dilated recurrent neural networks for glucose forecasting in Type 1 diabetes,”Journal of Healthcare Engineering, vol. 2020, Art. no. 5847940, 2020, doi: 10.1155/2020/5847940

    work page doi:10.1155/2020/5847940 2020

  26. [26]

    Digital twin for healthcare systems,

    A. Vallée, “Digital twin for healthcare systems,”Frontiers in Digital Health, vol. 5, Art. no. 1253050, 2023, doi: 10.3389/fdgth.2023.1253050

    work page doi:10.3389/fdgth.2023.1253050 2023

  27. [27]

    Benchmarking deep learning models on large healthcare datasets,

    S. Purushotham, C. Meng, Z. Che, and Y . Liu, “Benchmarking deep learning models on large healthcare datasets,”Journal of Biomedical Informatics, vol. 83, pp. 112–134, 2018, doi: 10.1016/j.jbi.2018.04.007

    work page doi:10.1016/j.jbi.2018.04.007 2018

  28. [28]

    Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining , pages =

    T. Chen and C. Guestrin, “XGBoost: A scalable tree boosting system,” inProc. 22nd ACM SIGKDD Int. Conf. Knowledge Discovery and Data Mining (KDD), San Francisco, CA, USA, 2016, pp. 785–794, doi: 10.1145/2939672.2939785

    work page doi:10.1145/2939672.2939785 2016

  29. [29]

    Machine Learning, V ol

    L. Breiman, “Random forests,”Machine Learning, vol. 45, no. 1, pp. 5–32, 2001, doi: 10.1023/A:1010933404324

    work page doi:10.1023/a:1010933404324 2001

  30. [30]

    LeCun, Y

    Y . LeCun, Y . Bengio, and G. Hinton, “Deep learning,”Nature, vol. 521, no. 7553, pp. 436–444, 2015, doi: 10.1038/nature14539

    work page doi:10.1038/nature14539 2015

  31. [31]

    Hochreiter and J

    S. Hochreiter and J. Schmidhuber, “Long short-term memory,” Neural Computation, vol. 9, no. 8, pp. 1735–1780, 1997, doi: 10.1162/neco.1997.9.8.1735

    work page doi:10.1162/neco.1997.9.8.1735 1997

  32. [32]

    Attention is all you need,

    A. Vaswani, N. Shazeer, N. Parmar, J. Uszkoreit, L. Jones, A. N. Gomez, Ł. Kaiser, and I. Polosukhin, “Attention is all you need,” inProc. 31st Conf. Neural Information Processing Systems (NeurIPS), Long Beach, CA, USA, 2017, pp. 6000–6010

    work page 2017

  33. [33]

    R. S. Sutton and A. G. Barto,Reinforcement Learning: An Introduction, 2nd ed. Cambridge, MA, USA: MIT Press, 2018

    work page 2018

  34. [34]

    Weihao Tan, Ziluo Ding, Wentao Zhang, Boyu Li, Bohan Zhou, Junpeng Yue, Haochong Xia, Jiechuan Jiang, Longtao Zheng, Xinrun Xu, Yifei Bi, Pengjie Gu, Xinrun Wang, B ¨orje F

    D. Silveret al., “Mastering the game of Go with deep neural networks and tree search,”Nature, vol. 529, no. 7587, pp. 484–489, 2016, doi: 10.1038/nature16961

    work page doi:10.1038/nature16961 2016

  35. [35]

    A unified approach to interpreting model predictions,

    S. M. Lundberg and S.-I. Lee, “A unified approach to interpreting model predictions,” inProc. 31st Conf. Neural Information Processing Systems (NeurIPS), Long Beach, CA, USA, 2017, pp. 4765–4774

    work page 2017

  36. [36]

    "Why Should

    M. T. Ribeiro, S. Singh, and C. Guestrin, “Why should I trust you?: Ex- plaining the predictions of any classifier,” inProc. 22nd ACM SIGKDD Int. Conf. Knowledge Discovery and Data Mining (KDD), San Francisco, CA, USA, 2016, pp. 1135–1144, doi: 10.1145/2939672.2939778

    work page doi:10.1145/2939672.2939778 2016

  37. [37]

    2015 , isbn =

    R. Caruana, Y . Lou, J. Gehrke, P. Koch, M. Sturm, and N. Elhadad, “In- telligible models for healthcare: Predicting pneumonia risk and hospital 30-day readmission,” inProc. 21st ACM SIGKDD Int. Conf. Knowledge Discovery and Data Mining (KDD), Sydney, NSW, Australia, 2015, pp. 1721–1730, doi: 10.1145/2783258.2788613

    work page doi:10.1145/2783258.2788613 2015

  38. [38]

    Deep learning: A critical appraisal,

    G. Marcus, “Deep learning: A critical appraisal,” arXiv preprint arXiv:1801.00631, 2018, doi: 10.48550/arXiv.1801.00631

    work page doi:10.48550/arxiv.1801.00631 2018

  39. [39]

    Stem cell-derived beta-cell therapies: Encapsulation advances and immunological hurdles in diabetes treatment,

    S. Waris, H. H. Begam, M. P. Kumar, Z. H. I. Abdulrasool, M. Avudaiappan, A. E. Butler, and M. Nandakumar, “Stem cell-derived beta-cell therapies: Encapsulation advances and immunological hurdles in diabetes treatment,”Cells, vol. 15, no. 2, Art. no. 191, 2026, doi: 10.3390/cells15020191

    work page doi:10.3390/cells15020191 2026

  40. [40]

    The OhioT1DM dataset for blood glucose level prediction: update 2020,

    C. Marling and R. Bunescu, “The OhioT1DM dataset for blood glucose level prediction: update 2020,” inProc. 5th Int. Workshop Knowl. Discovery Healthcare Data (KDH), vol. 2675, pp. 71–74, 2020

    work page 2020

  41. [41]

    A Proof-of-Concept Simulation-Driven Digital Twin Framework for Decision-Aware Diabetes Modeling,

    Z. Monirzadeh, “A Proof-of-Concept Simulation-Driven Digital Twin Framework for Decision-Aware Diabetes Modeling,” GitHub repos- itory, 2026. [Online]. Available: https://github.com/zarrinmonirzadeh/ diabetes-ai-ml-digital-twin

    work page 2026