Diabetes Analytics and Recommendation Engine

Can machine learning predict glucose fluctuations and deliver personalized, context-driven recommendations to help diabetic patients manage their condition daily?

2020 IEEE Eighth International Conference on Communications and Networking (ComNet 2020)

IEEE · 2020

Read Paper
Machine LearningAnomaly DetectionRecommendation SystemsIoTCloud Native

Lixar and Carleton University Data Day Award

Lixar / Carleton University · 2020

Top Project Award — Data Day 6.0

Carleton University · 2020

The Gap

With the widespread use of ever-personal technologies such as smartwatches and voice assistants, new techniques need to be explored to enable the management of an individual's glucose levels. Relevant health factors that contribute to an individual's glucose levels need to be determined in order to accomplish this.

which led us to ask
?The Question

Can machine learning predict glucose fluctuations and deliver personalized, context-driven recommendations to help diabetic patients manage their condition daily?

The Approach

We propose the Diabetes Analytic and Recommendation Engine (DARE) Architecture to harness personal technologies in assisting type two diabetic patients to manage their glucose levels through a rule-based system coupled with anomaly detection and threat forecasting in a context-driven environment. DARE takes a modular, cloud-native approach in applying machine learning to predict glucose levels and provide context-driven recommendations.

The Transformation

DARE assists diabetic patients in interactively managing their glucose levels, receive relevant health tips through correlations of their daily routines, and harness smart IoT devices for more accurate input of their health data.