Inspiration

As I'm a diabetic myself I deeply know what the challenges are. As developer I want to challenge the current state of the industry.

What it does

The solution analyses the glucose continuous monitoring reports as a time series and highlights the anomalies. These can then be documented and further detailed by the patient, i.e. what might have caused the anomaly. This way, when a patient visits the hospital for a regular checkup, these interesting data points can be discussed instead. The medical staff can save time by not having to run through long reports of e.g. the last six months.

How we built it

We use Azure Anomaly Detector API and Power Automate with Excel Online to visualise the data in an adaptive card which is sent to Outlook. A SharePoint list is used as the repository for the data points with the gather feedback from the patient. Alternatively patient can contact a bot using a case number to document the data points via a conversational flow. During this conversation the bot also provides information about the carbohydrate contents of foods which was trained using QnA maker and a public data set (FoodData Central) by the U.S. Department of Agriculture.

Challenges we ran into

It was the first time to use the Azure Anomaly Detector API, so I was unsure if it was a good match. It was also quite a challenge to find the time to work on the project.

Accomplishments that we're proud of

I'm proud of having an end-to-end solution for analysing as well as sharing and enriching the data by patients.

What we learned

I certainly learned a lot during the project, but was never really blocked by anything. There a lot of tools that can be used within the Microsoft stack.

What's next for Blood Glucose Analysis

Since we're enriching the data with user feedback, my hope is that this can be used to improve blood glucose level predictions.

Built With

  • botframework
  • cognitive-services
  • composer
  • power-automate
  • qnamaker
Share this project:

Updates