A novel biomarker for Type 2 Diabetes based on PhotoPlethysmoGraphy (PPG) data and Machine Learning.
Diabetes is a chronic condition where high blood glucose levels present for extended times impacts the cardiovascular system and also cause damages to the central and peripheral nervous system.
Instant Diabetes Test (IDT) uses high time-resolution PPG signal taken from a person's finger with a smartphone camera. PPG data is further processed to enhance signal-to-noise ratio and remove various artifacts.
Pre-processed PPG data points serve as the input to the sequence-to-sequence machine learning model, that along with other demographics data provide a probability of having Type 2 diabetes or pre-diabetes.
To develop the IDT machine learning model we used several years of longitudinal PPG heart pulse data from more than 150,000 people, out of which 25,000 have Type 2 diabetes.
With the enhanced PPG dataset, the specificity and sensitivity of the IDT biomarker matches the Fasting Plasma Glucose method used today in clinical practice.
IDT is a non-invasive biomarker for diabetes that can operate on any iOS or Android smartphone. IDT is easy to use from the comfort of your home without additional devices. It is therefore accessible to anyone, anywhere and anytime.
Our IDT model is currently in the process of clinical validation. We compare the IDT's model predictions with medical datasets from Retinopathy exams and the Fasting Plasma Glucose exams. We further compare these datasets to IDT scores of healthy individuals without Type 2 diabetes or pre-diabetes.
If you are interested in participating in the clinical study and contributing to the development of our novel biomarker for diabetes, apply here.