Data Collection of GlucoCheck and the Usability of the Mobile App
Disciplines
Computer Engineering | Hardware Systems | Investigative Techniques | Medicine and Health Sciences
Abstract (300 words maximum)
There is a need for a non-invasive method for monitoring blood glucose concentration. In seniors, infection and tissue damage risks increase with reduced skin elasticity. Thus, we are working on a prototype called GlucoCheck. The GlucoCheck model enables needle-free blood glucose estimation. The device analyzes the light that passes through the skin. Upon participants' arrival, height, weight, and demographic data such as race, gender, and age are collected. Skin tone, temperature, and humidity are all collected using sensors. Certified phlebotomists insert a flexible catheter into the participant’s arm to obtain comparable blood samples. GlucoCheck camera is placed on the finger to collect 12 pictures of the extremity. The Raspberry Pi processes these images using machine learning to determine the blood sugar level. Images from fingers are gathered at the same time points using the camera. Blood from a finger prick will also be collected. The participant then ingests a 75-gram glucose beverage. The mobile app aids diabetic users in logging pre- and post-meal glucose values, providing an accurate graphical representation of their blood sugar levels. Pictures of their meal(s) can be taken to log the type of food consumed and their caloric intake. A periodically retrained Convolutional Neural Network (CNN), based on transfer learning for Machine Learning (ML), ensures consistent optimal accuracy. Currently, the focus is on an American diet. We plan to include cuisines ranging from neighboring countries and eventually worldwide. The GlucoCheck is still in the prototyping phase of the study and more machine learning is necessary. We expect to observe a positive impact on the well-being of individuals in society living with diabetes and metabolic syndrome by enabling them to manage their condition without the pain of finger pricks.
Academic department under which the project should be listed
CCSE - Computer Science
Primary Investigator (PI) Name
Maria Valero de Clemente
Data Collection of GlucoCheck and the Usability of the Mobile App
There is a need for a non-invasive method for monitoring blood glucose concentration. In seniors, infection and tissue damage risks increase with reduced skin elasticity. Thus, we are working on a prototype called GlucoCheck. The GlucoCheck model enables needle-free blood glucose estimation. The device analyzes the light that passes through the skin. Upon participants' arrival, height, weight, and demographic data such as race, gender, and age are collected. Skin tone, temperature, and humidity are all collected using sensors. Certified phlebotomists insert a flexible catheter into the participant’s arm to obtain comparable blood samples. GlucoCheck camera is placed on the finger to collect 12 pictures of the extremity. The Raspberry Pi processes these images using machine learning to determine the blood sugar level. Images from fingers are gathered at the same time points using the camera. Blood from a finger prick will also be collected. The participant then ingests a 75-gram glucose beverage. The mobile app aids diabetic users in logging pre- and post-meal glucose values, providing an accurate graphical representation of their blood sugar levels. Pictures of their meal(s) can be taken to log the type of food consumed and their caloric intake. A periodically retrained Convolutional Neural Network (CNN), based on transfer learning for Machine Learning (ML), ensures consistent optimal accuracy. Currently, the focus is on an American diet. We plan to include cuisines ranging from neighboring countries and eventually worldwide. The GlucoCheck is still in the prototyping phase of the study and more machine learning is necessary. We expect to observe a positive impact on the well-being of individuals in society living with diabetes and metabolic syndrome by enabling them to manage their condition without the pain of finger pricks.