A MedEClasses × GROW Society innovation Back to learning.growsociety.in
Peer-reviewed · Hormone Research in Paediatrics 2026

Specialist-grade insulin decisions, in every family's pocket.

DiaBuddy is a point-of-care decision support app for children and adolescents with insulin-treated diabetes. It turns validated paediatric endocrinology algorithms into exact meal doses, basal adjustments and sick-day plans that families can follow at home, between clinic visits.

See the evidence Request clinic access

Developed at Regency Centre for Diabetes, Endocrinology & Research, Kanpur

301,000 children and adolescents in India live with type 1 diabetes, the most of any country (IDF Diabetes Atlas 2025)
6% vs 39% average insulin dose error, DiaBuddy versus families, against a blinded expert
94.5% of families' sick-day errors would have been prevented by following DiaBuddy
+12.6 points in quality of life (PedsQL) after three months' use, with HbA1c down 0.70%
01 · The need

A child with type 1 diabetes needs several insulin decisions a day. The specialist is there a few times a year.

In India, families make almost every one of those decisions alone: how much insulin for this meal, whether to change tonight's dose, what to do when the child is vomiting. Diabetes educators are scarce, clinic visits are months apart and many families live hours from a paediatric endocrinologist. Errors lead to hypoglycaemia, ketoacidosis and admissions that could have been avoided.

39% Average error in families' insulin doses Across 730 dosing decisions by 37 families, 57% of doses fell outside a ±20% safety margin.
52% Of sick-day decisions made wrongly 1,532 errors in 2,960 decisions, including 30% of choices about whether the child needed hospital.
22% Of families count carbohydrates regularly Those who did made smaller bolus errors and had fewer ketoacidosis episodes. Most need the arithmetic done for them.

Family figures: DiaBuddy preclinical validation, 37 families, 20 dosing and 20 sick-day vignettes each (Hormone Research in Paediatrics, 2026).

02 · What DiaBuddy does

The clinic's dosing logic, set up once by the care team and used by the family every day.

At enrolment the team enters the child's regimen, insulin types, doses and honeymoon status. DiaBuddy derives the insulin-to-carbohydrate ratio and sensitivity factor and applies ISPAD 2022-anchored rules to every decision after that.

Meal bolus calculator

Combines pre-meal glucose with the meal's carbohydrate content to give an exact dose, with a clinician-adjustable target.

CarbCal Indian food database

10,245 foods, including regional Indian dishes, so families can count carbohydrates for the meals they actually eat.

Basal titration

Adjusts long-acting insulin from fasting glucose using graded multipliers, from 0.8× when low to 1.2× when high.

Correction with insulin on board

Corrects post-meal highs while subtracting insulin still active from earlier doses, so corrections don't stack into a hypo.

Sick-day guide

Takes symptoms, glucose and ketones and returns a full plan: whether to go to hospital, dose changes, extra insulin, and the type and volume of fluids.

Glucose, food, activity and sleep log

Keeps the record the care team reviews with CGM data at each visit, so advice is based on what actually happened.

STEP 1 The care team sets up the child's profile Regimen, doses and targets entered in clinic; ratios calculated automatically.
STEP 2 The family asks DiaBuddy at each decision Meals, high readings, illness: an exact, guideline-aligned answer in seconds.
STEP 3 Red flags go straight to the hospital Signs of ketoacidosis, or blood ketones above 3.0 mmol/L, trigger referral instead of home management.

Available today as an Android app with its own server, installed for patients by the Regency clinic team. Play Store, iOS and Indian-language versions are on the roadmap below.

03 · Evidence

Peer-reviewed in two phases: accuracy on paper, then outcomes in children.

Original article · Hormone Research in Paediatrics · 2026 Development and Pilot Clinical Evaluation of DiaBuddy: A Point-of-Care Decision Support Tool for Children and Adolescents With Insulin-Treated Diabetes Mahapatra A, Paimode S, Banerjee S, Shukla P, Sachan N, Yadav V, Bajpai A. Published online 1 October 2026. DOI 10.1159/hrp/adaag032 · PMID 42821504
Read the paper
PHASE 1 · PRECLINICAL VALIDATION

DiaBuddy against 37 families and a blinded expert

Each family answered 20 dosing and 20 sick-day vignettes. A third paediatric endocrinologist, blind to both, set the gold standard.

Sick-day management

DiaBuddy was at least 90% accurate in every domain. Families erred on 30% of hospital decisions, 74% of dose changes, 68% of extra-insulin doses and 35% of fluid choices. Following DiaBuddy would have prevented 94.5% of these errors, including every wrong hospital decision.

PHASE 2 · PILOT CLINICAL STUDY

Three months of use by 20 children, aged 5 to 18

Prospective single-arm pilot at Regency CDER, Kanpur. 45% of the children had a past episode of ketoacidosis.

HbA1c 9.18 → 8.48% −0.70 points, P = 0.049; 55% fell by ≥0.5
Quality of life (PedsQL) 76.5 → 89.1 +12.6 points, P < 0.001
New ketoacidosis None during the study period
Time in range (CGM) 41.6 → 43.5% not significant at 3 months

A small, uncontrolled pilot: we report it as hypothesis-generating. Its variability data set the sample size for the multicentre randomised trial below.

04 · Key observations

What we learned

  1. Experience is not enough Families who had managed diabetes for years still got about half of sick-day decisions wrong. The gap is arithmetic and rules, not motivation.
  2. The biggest gains are in the riskiest moments DiaBuddy would have corrected every wrong decision about going to hospital, the error most likely to end in ketoacidosis.
  3. Families feel the difference first Quality of life rose sharply within three months, ahead of measurable change in CGM metrics.
  4. Built for Indian kitchens A local food database and simple dosing removed the need for families to count carbohydrates by hand, which only 22% did.
05 · Future plan

From one clinic to every family that needs it

DONE · 2026
  • Algorithms validated against families and expert
  • Pilot outcomes published in HRP
  • AI companion MedEDiabuddy validated and presented at APPES 2026
NEXT · 6 MONTHS
  • MedEDiabuddy chat built into the DiaBuddy app
  • Google Play release
  • Hindi interface, then other Indian languages
  • Copyright and trademark protection
2027 · TRIAL
  • Five-centre randomised trial of DiaBuddy with MedEDiabuddy
  • ICMR ANVESHAN funding application submitted
  • Trial designed as regulatory evidence for CDSCO software-as-medical-device review
SCALE
  • iOS and offline mode for low-connectivity areas
  • Free for families who cannot pay; clinic and state programme contracts
  • Adaptation for countries with similar diets and care gaps
Help us reach more families: we are looking for partners in state diabetes programmes, insulin-access charities and health-AI governance, and support for the iOS build, Indian languages and the trial. Partner with us
06 · Developers

Built by the clinicians who use it

Dr Anurag Bajpai

Lead developer and principal investigator · FRACP

Consultant Paediatric and Adolescent Endocrinologist at Regency Centre for Diabetes, Endocrinology & Research, Kanpur, and President of the Indian Society for Pediatric and Adolescent Endocrinology (ISPAE) 2025–26. Trained at AIIMS New Delhi and the Royal Children's Hospital, Melbourne.

  • Founder of MedEClasses Private Limited (2018, DPIIT-recognised) and the GROW Society; leads a hybrid paediatric endocrinology fellowship with about 200 fellows
  • Series Editor, Advanced Pediatric Endocrinology; Section Editor, Indian Journal of Pediatrics
  • h-index about 24, more than 1,800 citations; developer of a suite of validated decision-support tools for paediatric endocrinology
Development and research team
Dr Alapan MahapatraFirst author, HRP 2026
Dr Sandeep PaimodeCo-author, HRP 2026
Dr Sayan BanerjeeIndependent expert reviewer · PGIMER Chandigarh
Prashant ShuklaDietitian · nutrition and CarbCal
Dr Nimisha SachanPaediatric endocrinology, Regency
Vibha YadavCo-author, HRP 2026
SISTER PROJECT

MedEDiabuddy: the AI companion that answers families' questions

Guideline-locked AI built on DiaBuddy's algorithms. 96% agreement with a blinded expert, ahead of ChatGPT and Gemini (90%) and caregivers (56%).

Visit MedEDiabuddy