CoChat

Gamified optimized diabetes management with artificial intelligence-powered rural telehealth intervention (GODART): protocol for an optimization pilot and feasibility trial

Open access

Published version of a preprint

The DOI matches its Crossref record; 84 of the 86 references it cites were found in a registry or an index, and 2 are not checked yet; 17 of the 70 sources its claims cite were checked for support.

Abstract

Background: Type 2 diabetes mellitus (T2DM) is highly prevalent in the United States and represents a significant public health challenge. Telehealth interventions have shown promise for improving T2DM outcomes, but their effectiveness is often limited by disparities in digital literacy and access, especially in rural areas. To address this gap, we propose an innovative, individualized lifestyle modification intervention delivered via phone call to support glycemic control. Objective: This paper outlines the protocol for a pilot study designed to assess the feasibility and preliminary effectiveness of an artificial intelligence–assisted intervention for T2DM self-management in rural settings. Methods: The study uses the preparation phase of the MOST (Multiphase Optimization Strategy) framework to evaluate two components: (1) automated versus human health coaching and (2) fixed versus adapted gamified financial incentives, based on participants’ daily engagement with automated monitoring calls. We aim to enroll 88 adults with T2DM and hemoglobin A1c (HbA1c) levels between 6.5% and 11.5%. Participants receive daily interactive voice response calls tracking diet, physical activity, medication adherence, and blood glucose, and weekly coaching based on randomization. In the fixed-reward arm, participants earn US $0.60 per day; in the adaptive arm, rewards start at US $0.20 and increase weekly, with penalties for missed days. Primary outcomes include feasibility metrics and preliminary changes in HbA1c. Semistructured interviews will assess patient experience. Results: This study was funded by the National Institute of Diabetes and Digestive and Kidney Diseases. As of October 2025, we have enrolled and completed data collection for 88 participants. We expect to complete the feasibility analysis by December 2025. Conclusions: This pilot and feasibility study evaluates a low-tech, artificial intelligence–assisted T2DM intervention designed to reduce digital barriers and inform a future MOST optimization trial. Trial Registration: ClinicalTrials.gov NCT05344859; https://clinicaltrials.gov/study/NCT05344859 International Registered Report Identifier (IRRID): DERR1-10.2196/70271

Verified 2026-10-07 · 2 references not checked yet

Overview

CoChat Summary

TL;DR

GODART is a protocol for a MOST preparation-phase pilot/feasibility trial testing AI-assisted interactive voice response daily behavior monitoring plus either automated or human weekly coaching and fixed or adaptive gamified rewards in 88 rural adults with type 2 diabetes. As of October 2025, 88 participants completed 6-month follow-up, with feasibility analysis expected by December 2025.

Key findings

  • The study enrolled 88 adults with T2DM and HbA1c 6.5%-11.5%, randomized in a 2×2 factorial design to four arms: adaptive reward + automated coaching, adaptive reward + human coaching, fixed reward + automated coaching, or fixed reward + human coaching.
  • Recruitment involved 1938 phone contact attempts, 1192 completed phone screenings, 162 potentially eligible participants consented to screen for HbA1c, and 88 were randomized/enrolled and completed the 6-month follow-up as of October 2025.
  • All participants received daily IVR calls for 24 weeks tracking diet, physical activity, medication adherence, and fasting blood glucose monitoring, plus weekly coaching; fixed rewards were US $0.60/day, while adaptive rewards started at US $0.20/day, increased to US $0.25 and US $0.30 in weeks 2 and 3, and reached US $0.65/day from week 4, with up to US $100.80 possible over 24 weeks in either arm.
  • Primary outcomes are feasibility metrics (recruitment, retention, adherence); secondary outcomes include HbA1c and weight, and the trial is powered to detect a Cohen d of 0.572 (0.86% HbA1c difference) with 80% power assuming 15% attrition and a pre-post correlation of 0.5.

Methods

  • Design: MOST preparation-phase optimization pilot and feasibility trial using a 2×2 factorial experiment with two components—automated versus human health coaching and fixed versus adaptive gamified rewards—among 88 English-speaking adults aged ≥18 years with T2DM and HbA1c 6.5%-11.5%, recruited largely from rural/nonmetropolitan areas via the UAB Health System i2b2 system.
  • Data: participants received daily automated behavior monitoring system calls via IVR using AI-assisted natural language processing and Nutritionix for dietary intake, weekly coaching, baseline and 6-month LabCorp blood draws for HbA1c/lipids/biometrics, validated surveys, and semistructured exit interviews.
  • Analysis: feasibility outcomes will be summarized descriptively, with logistic regression for recruitment predictors, linear mixed models using intent-to-treat for preliminary effectiveness, 75%/85%/95% CIs, and qualitative interview analysis using a phenomenological approach with inductive coding.

Limitations

  • The paper states that daily phone-based engagement may limit generalizability because enrolled participants were likely more motivated at baseline; non-English speakers were excluded; behavioral data were self-reported and subject to recall bias; no social desirability scale was included; and actual glucose values were not collected.
  • The paper also states that automated coaching recommendations were not generated by AI but were developed by a human coach and delivered asynchronously via IVR, and that coaching fidelity was not comprehensively evaluated.
  • Evident from the text: this is a pilot/feasibility protocol not powered for definitive effectiveness, and effectiveness results are not yet available; the paper reports enrollment completion but no outcome data.

Model-written summary. Check it against the paper before relying on it.

Integrity signals

Registry

Crossref · 10.2196/70271

Cites

84 of 86 references verified0 citation errors · 0 not checkable

Statistics

Nothing to recompute.

Figures

Figure 1 Overview of the GODART intervention structure; all study participants receive the GODART core technology (ABMS), but are randomized to one of two optimization experimental conditions: the Technology Optimization Experimentation, comparing human versus automated health coaching, or the Resource Optimization Experimentation, comparing fixed versus adaptive gamified rewards. ABMS: automated behavior monitoring system; AI: artificial intelligence; GODART: Gamified Optimized Diabetes Management with Artificial Intelligence-Powered Rural Telehealth.
Figure 1 Overview of the GODART intervention structure; all study participants receive the GODART core technology (ABMS), but are randomized to one of two optimization experimental conditions: the Technology Optimization Experimentation, comparing human versus automated health coaching, or the Resource Optimization Experimentation, comparing fixed versus adaptive gamified rewards. ABMS: automated behavior monitoring system; AI: artificial intelligence; GODART: Gamified Optimized Diabetes Management with Artificial Intelligence-Powered Rural Telehealth.CC BY · PubMed Central
Figure 2 Flow diagram outlining screening, randomization (n=88), and intervention arms comparing fixed versus adaptive incentives and automated versus human coaching in the GODART study. EMR: electronic medical record; GODART: Gamified Optimized Diabetes Management with Artificial Intelligence-Powered Rural Telehealth; UAB: University of Alabama at Birmingham.
Figure 2 Flow diagram outlining screening, randomization (n=88), and intervention arms comparing fixed versus adaptive incentives and automated versus human coaching in the GODART study. EMR: electronic medical record; GODART: Gamified Optimized Diabetes Management with Artificial Intelligence-Powered Rural Telehealth; UAB: University of Alabama at Birmingham.CC BY · PubMed Central
Figure 3 Overview of the automated behavior monitoring system: 8-step process used in the GODART study to support diabetes self-management. See our YouTube demo [49] for step 5. API: application programming interface; GODART: Gamified Optimized Diabetes Management with Artificial Intelligence-Powered Rural Telehealth; NLP: natural language processing; UAB: University of Alabama at Birmingham.
Figure 3 Overview of the automated behavior monitoring system: 8-step process used in the GODART study to support diabetes self-management. See our YouTube demo [49] for step 5. API: application programming interface; GODART: Gamified Optimized Diabetes Management with Artificial Intelligence-Powered Rural Telehealth; NLP: natural language processing; UAB: University of Alabama at Birmingham.CC BY · PubMed Central

Verification

Claims the paper makes

7 Valid · 2 Not valid · 27 Not checked · 54 Own results and interpretations

33 more claims

Create a free account to see all 90 claims and to verify the 53 references still open.

See all claims — free account
  • Probably not a substantial claim. 39
  • It may state two separate claims as one. 23

Create a free account to see these claims.

Claims other papers make about this one

Each is checked against this paper’s own text: does it say what the citing paper says it does?

Self-check

Percentages, sums and reported differences re-computed from the paper’s own numbers.

Each contradiction between two passages listed here was confirmed by an independent check.

2 of 2

Citations

Cites

Show all 86 references

Research works printed but not linked: 3

  1. World Health Organization Global Report on Diabetes 2016 Geneva, Switzerland World Health Organization
  2. Polkosky MD Toward a Social-Cognitive Psychology of Speech Technology: Affective Responses to Speech-Based e-Service 2005 Tampa, FL, United States University of South Florida
  3. Radford A Narasimhan K Salimans T Sutskever I Improving Language Understanding by Generative Pre-Training 2018 San Francisco OpenAI

Cited by

See all 2

Metadata

Authors
Mehta, T., John, T., El Zein, A., Faught, V., Nawshin, T., Chilke, T. S., Cohen, C. W., Cherrington, A., & Thirumalai, M.
Affiliations
University of Alabama at Birmingham (US)
Corresponding author
Mehta, Tapan
Venue
JMIR Research Protocols
Published
2025-12-05
Volume
14
Pages
e70271
ISSN
1929-0748 (ISSN-L)
OpenAlex
W4413775962
Cited by
2
References
63
Citations by year
2026: 2
FWCI
1.36
Type
journal article
Language
en
Open access
Licence
cc-by
Locations
  • JMIR Research Protocols (publishedVersion, cc-by)
  • PubMed (publishedVersion)
  • PubMed Central (publishedVersion, cc-by)
  • DOAJ (DOAJ: Directory of Open Access Journals) (submittedVersion, cc-by-sa)
Funders
National Institute of Diabetes and Digestive and Kidney Diseases (R01DK129378)
Topics
  • Health Sciences › Medicine › Endocrinology, Diabetes and Metabolism › Diabetes Management and Education
  • Health Sciences › Health Professions › General Health Professions › Mobile Health and mHealth Applications
  • Health Sciences › Medicine › Public Health, Environmental and Occupational Health › Telemedicine and Telehealth Implementation
Keywords
type 2 diabetes, telemedicine, gamification, financial incentives, artificial intelligence, interactive voice response, rural health, diabetes self-management, glycemic control, HbA1c, digital health disparities, medication adherence, physical activity, diet
MeSH terms
Self-Management, Artificial Intelligence, Blood Glucose, Diabetes Mellitus, Type 2, Feasibility Studies, Female, Glycated Hemoglobin, Humans, Male, Middle Aged, Pilot Projects, Rural Population, Telemedicine

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

The claim verdicts this report counts are experimental: their accuracy is not yet measured.

A language model drew the claims from the paper on 2026-10-07, under version 3 of the claim definition (model deepseek-v4.1-flash, prompt extract-v3).

A second model labelled each claim on 2026-10-07 (model jev-1.13-20260917, definition revision gold-guideline-v0.2).

Cited sources are checked with a language model’s help (qwen3.8-omni-flash and gpt-6-sol).

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review

Data sources

Bibliographic data: Semantic Scholar, OpenAlex, Crossref.

Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).

Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.