Listening to the patients

Listening to the patients

Listening to their gut to improve the lives of people with type I diabetes.

New! Highlights 2020

The project's contribution to the Centre for Digital Life Norway annual report 2020.

During 2020 our focus has been on developing the device to be used for collecting and processing abdominal sounds.

One highlight of 2020 has been the successful early identification of meal intake.

For us one added value of being a part of a transdisciplinary centre is the possibility for our PhD and postdoctoral candidates to participate in dedicated transdisciplinary courses.

Project overview

Project lead: Sven Magnus Carlsen
Institution: NTNU
Partner: SINTEF Digital AS
Funding: Research Council of Norway
Duration: 4 years (2019–2022)
Follow in social media: XX
See also Artificial Pancreas Trondheim (APT) and the Double Intraperitoneal Artificial Pancreas (DIAP) project, SINTEF Digital and Prediktor Medical

Publications

 

Pilot Study of Exploiting Abdominal Sound for Early Meal Onset Detection (2020)
Sunilkumar Telagam Setti, Elise Søiland, Øyvind Stavdahl, Anders Lyngvi Fougner

Pilot study of Early Meal Onset Detection from Abdominal Sounds (2020)
Sunilkumar Telagam Setti, Elise Søiland, Øyvind Stavdahl, Anders Lyngvi Fougner

Pilot study of Early Meal Onset Detection from Abdominal Sounds (2019)
Sunilkumar Telagam Setti, Elise Søiland, Øyvind Stavdahl, Anders Lyngvi Fougner

Towards a Safe Artificial Pancreas: Meal Detection and the Intraperitoneal Route (2018)
Konstanze Kölle

Feasibility of early meal detection based on abdominal sound. (2019)
Konstanze Kölle, Anders Lyngvi Fougner, Reinold Ellingsen, Sven Magnus Carlsen, Øyvind Stavdahl

Data driven filtering of bowel sounds using multivariate empirical mode decomposition (2019)
Konstanze Kölle, Muhammad Faisal Aftab, Leif Erik Andersson, Anders Lyngvi Fougner, Øyvind Stavdahl


All results in the CRIStin-database

Research group

Listening to the Patients is developing sound-based meal detection system to improve artificial pancreases. Timing insulin delivery with a meal is an important part of controlling blood glucose levels and avoiding long-term complications. Accomplishing this high-risk goal will be a major step toward improving the lives of people with diabetes.

Many people with type I diabetes are treated with glucose monitors and insulin pumps delivering insulin in subcutaneous tissue. By combining a glucose monitor and an insulin pump one can make a devise delivering insulin based on the measured glucose levels. Such a devise is called an artificial pancreas. However, subcutaneous insulin delivery carries inherent delays in the effect on glucose levels making it hard to achieve good glucose control.

To reduce the delays associated with the subcutaneous approach, the Artificial Pancreas Trondheim (APT) research group work on intraperitoneal glucose monitoring and delivery, i.e. between the intestines. With this “double intraperitoneal” approach for an artificial pancreas the absorption of insulin and the effect on glucose levels are much faster.However, such an artificial pancreas will still struggle to handle increasing glucose levels after meals. It takes at least 30 minutes for the artificial pancreas to detect this rise after a meal, far too long a delay to control glucose properly. Over time, insufficient glucose control can lead to long-term issues like kidney disease, cardiovascular disease or even blindness.

Listening to the Patients’s bold new approach analyzes sounds in the gastrointestinal tract with external microphones for telltale sounds of a meal. The researchers will correlate this information with blood glucose levels to administer the correct dosage of insulin much closer to meal intake than current devices. In a recent pilot study, they were able to detect a meal in as little as 10 minutes after consumption.

This research has great potential for developing a vastly improved artificial pancreas, but it is also highly risky: their detection algorithm must be extremely accurate because administering a meal dose of insulin to a patient that has not eaten can be deadly. To ensure their algorithm meets this high standard, the Listening to the Patients team consists of engineering physicists, sound engineers, cyberneticists, physicians, and endocrinologists. They are using machine learning based on the principles of speech recognition and their knowledge of engineering and physiology to identify meal sounds and eliminate extraneous sounds. They expect to have a proof of concept listening device ready for testing within one year.

The Listening to the Patients team is working closely with people with diabetes to ensure a positive user experience. Privacy is a major concern with audio systems so their algorithm will identify and eliminate speech from the recordings. When fully developed, the system will analyze data real time and not store sound recordings.

Listening to the Patients project is part of the Artificial Pancreas Trondheim (APT) research group at The Norwegian University of Science and Technology (NTNU) in Trondheim. It is funded by the Research Council of Norway and is one of the multidisciplinary research projects within Centre for Digital Life Norway.

By Matthew Davidson

Latest news from the project