Type2 Diabetes Mellitus
Conditions
Keywords
AI-powered Thai food analysis
Brief summary
Study Title: The Effectiveness of an AI-powered Thai food analysis (SnapD) and Continuous Glucose Monitoring on Glycemic Control in Patients with Type 2 Diabetes and Overweight or Obesity: A Randomized Controlled Pilot Study Rationale: Effective dietary management is the cornerstone of treating Type 2 Diabetes (T2DM) and obesity. However, traditional manual food logging is often inaccurate and burdensome. While digital tools and Continuous Glucose Monitoring (CGM) have shown promise internationally, there is a lack of validated AI-powered tools specifically designed for Thai cuisine. This study introduces SnapD, an AI-powered platform (utilizing Gemini 2.5 Flash) designed to recognize Thai food, estimate nutritional values, and integrate with CGM data to provide personalized feedback. The primary goal of this pilot study is to evaluate the efficacy of the SnapD application, both as a standalone tool and in combination with CGM, compared to Standard of Care in improving glycemic control (HbA1c) over 8 weeks. Additionally, the study aims to assess the feasibility, participant adherence, and safety of these digital interventions to inform a future, fully powered randomized controlled trial. Study Design: This is an 8-week, randomized, open-label, parallel-group, superiority pilot study with a 1:1:1 allocation ratio. A total of 45 participants will be enrolled and assigned to one of three arms: 1. Intervention Arm 1: SnapD application + Real-time CGM + Diabetes Self-Management Education and Support (DSMES) 2. Intervention Arm 2: SnapD application standalone + DSMES 3. Control Arm: DSMES alone Inclusion Criteria Highlights: Adults (18-65 years) diagnosed with T2D with BMI \> 23 kg/m² (overweight/obesity) with HbA1c between 6.5% and 9.0% with Must possess a compatible smartphone/tablet Procedures: Baseline (Visit 1): All participants receive 20-30 minutes of DSMES. Intervention groups receive training on SnapD. Arm 1 receives a 15-day CGM sensor.During Study: Intervention arms log meals via SnapD (at least twice daily). Nutritionists conduct bi-weekly follow-up calls to address technical issues and provide support. End-of-Study (Week 8): Assessment of HbA1c, body weight, waist circumference, lipid profile, and patient-reported outcomes (self-care activities and user satisfaction) Primary Outcome: Mean change in HbA1c from baseline to 8 weeks Secondary Outcomes: Changes in Fasting Plasma Glucose (FPG), body weight, waist circumference, and lipid profiles, Diabetes self-management scores (SDSCA questionnaire), User satisfaction with the SnapD application, Incidence of adverse events (hypoglycemia/hyperglycemia). Significance: This study will provide preliminary evidence on the synergistic benefits of AI-driven nutritional feedback and CGM in a Thai-specific context, supporting the development of scalable, culturally adapted digital health technologies for diabetes management.
Interventions
DEVICESnapD
SnapD, developed by the Division of Endocrinology and Metabolism, Department of Medicine, Ramathibodi Hospital, is a Progressive Web App. It is built using React 18.3.1, TypeScript, and Vite for responsive performance on both mobile and desktop platforms. The application utilizes Supabase for database management, which operates on a PostgreSQL backend.
DEVICELinx CGM
The Linx CGM system, manufactured by Connect Diagnostics, is a real-time device that measures glucose concentrations in the interstitial fluid. It is an all-in-one device, integrating the glucose sensor, applicator, and transmitter into a single unit. The device has a diameter not exceeding 22 mm and a weight not exceeding 2.2 g. The sensor has a maximum operational life (wear time) of 15 days. It demonstrates a Mean Absolute Relative Difference (MARD) not exceeding 8.67%, which meets the standard accuracy requirement of \<10%. This device was registered as a medical device by the Thai Food and Drug Administration (Thai FDA), Ministry of Public Health, in January 2025, for the indication of management of diabetes in adults age 18 and older (as shown in the attached document) .
OTHERDiabetes Self-Management Education (DSMES)
Participants will receive 1 single session of DSMES, 20-30 minutes/session. This single session will be delivered at the baseline visit only. All personnel providing DSMES are Certified Dietitians (CD) and/or have passed the Certified Diabetes Educator (Thai CDE) examination
Sponsors
Ramathibodi Hospital
Study design
Allocation
RANDOMIZED
Intervention model
PARALLEL
Primary purpose
TREATMENT
Masking
NONE
Masking description
This is an open-label study. Blinding of participants and investigators is not feasible due to the distinct procedural nature of the assigned interventions (e.g., the visible use of a wearable CGM device and a specific mobile application). However, to minimize bias in the assignment, allocation concealment is ensured by using sequentially numbered, opaque, sealed envelopes prepared by a non-investigator.
Intervention model description
This is a randomized, open-label, parallel-group, superiority pilot study designed with a 1:1:1 allocation ratio. The study compares two digital intervention arms against a standard of care control group over an 8-week period. Arm 1 (SnapD + CGM): Participants use the SnapD AI-powered application for food logging combined with a 15-day session of real-time Continuous Glucose Monitoring (CGM). Arm 2 (SnapD only): Participants use the SnapD application as a standalone digital food diary. Control Arm: Participants receive standard Diabetes Self-Management Education and Support (DSMES) only. The randomization uses a permuted variable block size design to ensure balanced allocation. Allocation concealment is maintained using sequentially numbered, opaque, sealed envelopes. The study aims to evaluate the feasibility and preliminary efficacy of these tools in patients with type 2 diabetes and overweight or obesity.
Eligibility
Sex/Gender
ALL
Age
18 Years to 65 Years
Healthy volunteers
No
Inclusion criteria
1. Aged 18 to 65 years, male or female at birth 2. Diagnosed with type 2 Diabetes Mellitus with overweight or obesity (BMI\>23 kg/m²) 3. Hemoglobin A1c (HbA1c) 6.5-9% measured within 3 months prior to the screening date 4. Willing to maintain their current antidiabetic medication regimen without dose adjustment for the entire 8-week study duration 5. Must possess an internet-enabled devices e.g. smartphone, tablet compatible with the SnapD application 6. Able and willing to adhere intervention, including using snapD and CGM
Exclusion criteria
1. Currently pregnant, plan pregnancy or breastfeeding during the 8-week study period 2. Current participation in another interventional clinical trial 3. Current use of insulin or incretin-based therapies (e.g., GLP-1 Receptor Agonists, GIP/GLP-1 Receptor Agonists) 4. Presence of severe hearing or visual impairment that, in the investigator's judgment, would preclude the participant from safely and effectively using the SnapD application or the CGM device 5. known contraindication to CGM usage e.g., a history of severe hypersensitivity to the device's materials or adhesive, planing to go on CT-contrasted imaging etc.
Design outcomes
Primary
| Measure | Time frame | Description |
|---|---|---|
| To evaluate the efficacy of two intervention arms on glycemic control | baseline, 8 weeks | HbA1c level(%) |
Secondary
| Measure | Time frame | Description |
|---|---|---|
| To assess changes in other glycemic parameter | baseline, 8 weeks | Fasting plasma glucose(mg/dL) |
| To assess changes in anthropometric measurements | baseline, 8 weeks | Body Mass Index(Kg/m2) |
| To evaluate changes in metabolic parameters | baseline, 8 weeks | • change in systolic and diastolic blood pressure (mmHg) |
| To evaluate changes in diabetic self-management activities via questionnaire | Baseline, 8 weeks | A modified version of the SDSCA tool, developed by Rattanaporn Jeerawattana, encompassing domains of diet, exercise, self-monitoring, foot care, and medication. The total score ranges from 0 to 133, where higher scores indicate better diabetes self-care activities. |
| • To evaluate user satisfaction scores for the SnapD application | 8 weeks | Name of the questionnaire is User satisfaction with the SnapD and CGM application. It was an a self-developed questionnaire validated by experts. The total score ranges from 7 to 35, where higher scores indicate better user satisfaction. |
| • To assess adverse events about glycemic control | up to 8 weeks | • Incidence of hypoglycemia and/or hyperglycemia |
| Change in metabolic parameters | baseline, 8 weeks | change in lipid profile including serum Total Cholesterol, Triglyceride-Cholesterol, HDL-Cholesterol, LDL-Cholesterol |
Countries
Thailand
Contacts
CONTACTSupasuta Wongdama, MD
PRINCIPAL_INVESTIGATORSupasuta Wongdama
Ramathibodi Hospital
Outcome results
None listed