(We originally posted this in 2021. You can read more of our original ideas in our archive. You can order a business plan of this idea here.)
Problem: Despite the younger generation’s health awareness, the population of Diabetics in America (and globally) is increasing at a derivative rate. According to the CDC, Diabetes has a diagnosis rate of about 13% of the population. Youths diagnosed with diabetes are more susceptible to increased risk due to the likelihood of poor eating and management habits that lead to long term mortality increases of about 20%, or 10-20 years off the average lifespan. In terms of the health crisis, the direct and indirect cost of diabetes in America is over $327 billion.
Two major barriers exist in Diabetes Management: Technological capability and human psychology. Technologically, diabetics can now employ a “closed loop” mechanic where continuous blood sugar readings are in a feedback loop with a pump system to automatically deliver insulin. Manual estimation of food eaten, as well as manual calibration of the algorithm used in a closed loop, is still required.
Across the board, A1C readings (a measure of management & state of health for diabetics) declined by an average of 17% for diabetics using closed loops. Enter the next barrier: counting carbs. Save for the strictest of individuals, no one counts carbs correctly. After years of treatment, diabetics “get a feel” for the impact of food on their body and just guess what they are intaking. This leads to incorrect dosage and poor management of diabetes, and despite further transparency from restaurants on nutrition and labels provided for cooking, it is strenuous to continuously calculate how much you are eating.
Solution: An App (or tool to be deployed within app) to enhance diabetes management through AI Image recognition and Blood Glucose monitoring to reward positive eating habits in youths and provide trend analysis for both healthcare professionals and families.
With open source image recognition tools such as MASK-RCNN and image repositories like MS COCO, building a program to recognize food is relatively trivial. Counting carbs would require density and volumetric measurements, which is possible with the iPhone's new LIDAR capable camera (originally designed for AR apps). In tandem with lookup tables for carb density, the app would be able to report with more precision how much food is being eaten and log what food was consumed with its average nutrition & classification to provide you solid data on your eating habits.
With the goal of integrating into the closed loop system, the app would analyze the picture and send its carb count directly to the Loop app, thus only requiring you to take a picture of your food instead of count.
Monetization: The average cost of diabetes per person per year is +$8,000 USD. Subscription to an app capable of providing this level of detail could be $60 a year (same price as Strava, and less than Headspace, two incredibly popular wellness apps).
Further expansion of the target audience or refinement of the value offered could be:
Gamifying food eaten and blood sugar readouts to encourage positive eating habits and management for young diabetics (gamification, or “childification” of medical therapies has been shown to increase the likelihood of enthusiastic engagement from youth and reduce psychological impact of illness)
Expansion of audience to health-conscious individuals to serve a growing market of weight loss programs & food/wellness. Most tracking is still manual, and this would provide a way to automate your data collection
Integration of photo taking with food to share to social media/year in review montages
Contributed by: Alex Kessel (Intern at Billion Dollar Startup Ideas)