Diabetes: Calculating Insulin by Automated Carb Counting Using AI, ML
summary
A Progressive Web App automates carb counting and computes insulin by combining Clarify food model image recognition, Azure OCR, USDA FoodDataCentral lookups and an optional Bluetooth scale for gram-accurate portions, then applies stored user formulas and a linear regression model to predict next glucose. Back end runs as microservices in Kubernetes using Oracle Autonomous Database (ORDS/SODA) and Micronaut/Micronaut Data; demo shows the pipeline works but experienced Bluetooth pairing failures during the live demo.
AI-written summary · about
chapters (22)
- 0:00Introduction
- 0:45Personal Story
- 1:32Avas Story
- 3:17What is Diabetes
- 5:25How Diabetes Works
- 6:27Managing Highs and Lows
- 7:42Glucose Monitors
- 9:36Limitations
- 10:36Demo
- 11:24Insulin Helper
- 12:10Formulas
- 14:28Bluetooth Scale
- 22:54Architecture
- 24:45Nutrition Service
- 26:59Behind the Scenes
- 30:05Serverless Function
- 35:44Formula Service
- 39:34Glucose Prediction
- 40:36Model
- 41:21Conclusion
- 44:05Other uses
- 46:11Research
related talks
36mAI for human and the planet: AI in nutrition
01h 00mTake Control of the Data of You
25mAI for healthcare: Scaling Access and Quality of Care for Everyone
17mAI Innovation in Health
53mImplementing AI using Azure Machine Learning Services
55mTake Control of the Data of You
01h 04mTake Control of the Data of You
07mHow AI is Enabling Healthcare's Biggest Blindspot