Development and Preclinical Validation of DiaBuddy, a Point-of-Care Decision Support Tool for Children With Type 1 Diabetes: A Pilot Clinical Study

Background Optimal management of Type 1 Diabetes Mellitus in children requires frequent insulin dose adjustments based on pre-meal glucose concentrations, carbohydrate content of meals, and personalized parameters including insulin-to-carbohydrate ratio and insulin sensitivity factor. Intercurrent illness requires structured sick-day management to prevent diabetic ketoacidosis. Consistent implementation of these tasks by families in real-world settings is poor, contributing to suboptimal glycemic control and preventable acute complications. Bolus calculators and digital carbohydrate-counting programs have been shown to reduce postprandial glycemic excursions and hypoglycemia in Western populations but no validated tool addressing both insulin dosing and sick-day management existed for Indian children prior to DiaBuddy™.

DiaBuddy™ Application DiaBuddy™ is a comprehensive point-of-care decision support application for pediatric T1DM with seven integrated modules. At enrollment, the clinical team enters the child's honeymoon status, insulin regimen, insulin type, doses, and timing. The Insulin Wizard computes the insulin-to-carbohydrate ratio (500/TDD) and insulin sensitivity factor (1800/TDD for rapid-acting insulin; 1500/TDD for regular insulin). Bolus doses integrate pre-meal glucose, meal carbohydrate content from CarbCal - a proprietary India-specific nutrition database of 10,245 foods - and a default meal target of 120 mg/dL. Basal dose titration follows fasting glucose-linked multipliers. The Sick-Day Guide requests clinical features, capillary blood glucose, and blood or urine ketone values and generates ISPAD-aligned recommendations on hospital admission, total daily dose modification, supplemental rapid-acting insulin dose, and oral fluid type and volume.

Phase 1 - Preclinical Vignette-Based Validation Thirty-seven families of children with T1DM attending the Regency Center for Diabetes, Endocrinology and Research outpatient clinic completed 20 standardized insulin-dosing vignettes (12 bolus, 8 basal) and 20 sick-day management vignettes independently and using DiaBuddy™, blinded to each other's outputs. Vignettes were drafted by two pediatric endocrinologists and reviewed for ISPAD 2022 guideline concordance. A pediatric endocrinologist from a different institution with no financial interest in DiaBuddy™ served as the blinded independent gold standard. Insulin dosing accuracy was quantified using absolute relative deviation (ARD). Sick-day accuracy was evaluated across four domains: hospitalization decision, total daily dose modification, supplemental rapid-acting insulin dose, and fluid type recommendation. Domain-specific weights were pre-specified (hospitalization: 4; supplemental rapid-acting insulin: 3; total daily dose modification: 2; fluid type: 1). Weighted net benefit and prevention index were calculated.

Phase 2 - Pilot Clinical Study A prospective single-arm study enrolled 25 children and adolescents with T1DM (diabetes duration greater than 1 year, age 5 to 18 years) from the same outpatient clinic. The Phase 2 cohort was independent of Phase 1; no participant was enrolled in both phases. At baseline, participants underwent HbA1c measurement, 15 days of intermittently scanned continuous glucose monitoring with FreeStyle Libre 2, and quality of life assessment using the PedsQL questionnaire. Families received structured training on DiaBuddy™ use by a trained research nurse and were provided unrestricted access to the application for home use. After three months, HbA1c, CGM metrics, and quality of life scores were reassessed. Application satisfaction was evaluated using a bespoke 10-item Likert-type rating scale (maximum 50 points). Families were asked whether they wished to continue using the application after study completion. Two participants with incomplete follow-up data and three who withdrew voluntarily were excluded, leaving 20 participants for analysis.

Statistical Analysis Continuous variables were summarized as mean ± standard deviation. Insulin dosing accuracy was quantified using ARD and Bland-Altman analysis. Baseline and post-intervention outcomes were compared using paired t-tests or Wilcoxon signed-rank tests as appropriate. Effect sizes were calculated using Cohen's dz for paired samples. A clinically meaningful improvement in glycemic control was predefined as HbA1c reduction of at least 0.5%. All analyses were performed in SPSS version 23.

Ethics The study was approved by the Institutional Ethics Committee of Regency Hospital Limited, Kanpur (Approval Nos. 16084, 160152, 160151) and conducted in accordance with the Declaration of Helsinki. Written informed consent was obtained from parents or guardians and, where applicable, assent from participants prior to enrollment.

Funding This study received no external funding. All resources were provided by Regency Hospital Limited, Kanpur as institutional support. No commercial funding, application licensing fee, or industry contribution was received.

Conflict of Interest Dr. Anurag Bajpai is a co-developer of DiaBuddy™ and holds equity in MedEClinics Private Limited, the company in which DiaBuddy™ is registered. To mitigate potential bias, Dr. Bajpai was not involved in vignette construction, family data collection, or the selection and engagement of the independent expert gold standard. The independent gold standard investigator was affiliated with a different institution and had no financial interest in DiaBuddy™.