November 6, 2023

Wearable technologies help fight diabetes

Continuous innovation in the field of medical devices aims to improve user-friendliness and reduce the burden on patients. This also applies to people with diabetes, which is why CSEM and the Department for Diabetes, Endocrinology, Nutritional Medicine and Metabolism (UDEM) at Inselspital, University Hospital Berne aim to develop non-invasive, non-intrusive hypoglycemia detection solutions based on voice and other physiological signals.

Woman speaking in her cellphone
© CSEM - Like other vital signs, voice can potentially harbor relevant information about the state of health of patients.

According to the World Health Organization, by 2021, more than 500 million people were living with diabetes, and this figure is expected to increase to more than 1 billion people by 2050 due to factors such as an ageing population, a sedentary lifestyle and unhealthy eating habits. Despite ongoing and important developments in diabetes care, hypoglycemia (low blood glucose) remains one of the most dangerous and frequent complications of this chronic disease.

"Because of the stigma still attached to people with diabetes, there is a strong demand for discreet, real-time, and non-invasive solutions that enable the continuous monitoring of blood glucose levels and the anticipation of hypoglycemia situations so that they can be quickly remedied," explains Professor Christoph Stettler, Director of UDEM (Dept of Diabetes, Endocrinology, Nutritional Medicine and Metabolism) at Inselspital, University Hospital Bern.

People with diabetes traditionally have to monitor their blood sugar several times a day and adjust medications, activity, and food intake to help achieve blood sugar targets and avoid low blood glucose. The monitoring can be performed by the patient using a blood glucose meter (SMBG – Self-monitoring of blood glucose) or a special sensor inserted beneath the skin (CGM – Continuous Glucose Monitoring). SMBG is limited by its cumbersome procedure and lack of information on glucose dynamics. CGM offers permanent glucose control but is invasive, costly, and subject to a time lag, especially in hypoglycemia where rapid detection is vital. Therefore, there is an unmet need for non-invasive and novel approaches to detect hypoglycemia.


While SMBG requires patients to perform several finger-sticks during the day, CGM devices include a small disposable sensor inserted into the skin from the stomach or arm. The sensor measures glucose levels every few minutes, throughout the day and night, and sends the information to an attached transmitter and to a separate receiving device. Based on the information from the sensor, patients will either inject the appropriate level of insulin themselves or use a so-called hybrid loop system with semi- automatic insulin injection via a pump. The readings are relayed in real time to a device which can also be read remotely by the caregiver or health-care provider, allowing these to adjust their patient’s treatment.

Different to traditional blood glucose measurement systems, CGM measure glucose in interstitial fluid (ISF), assuming that glucose levels in blood and ISF are similar, and that the information provided can be used interchangeably. Most CGMs report blood sugar levels every 5 minutes (for a total of 288 glucose readings per day). Also, it can take a CGM device about 5 to 25 minutes longer to show a variation in glucose levels compared to venous blood glucose. This time gap is due to the compartmental difference and can be critical since prompt reaction is key in the treatment of a hypoglycemia situation.


Digital biomarkers extracted from human voice hold a potential to significantly improve public and individual health. Voice can be acquired by personal devices and the same devices can be used to provide rapid diagnostics or report the progression of a disease to the clinicians.

Like other vital signs, voice can potentially harbor relevant information about the state of health of patients. Vocal biomarker systems analyze how a person talks and focuses on prosody, intonation, pitch, etc. to draw conclusions about his/her state of health, using machine learning/artificial intelligence. CSEM and IDIAP Research Institute in Martigny have been actively collaborating on projects where they combined their expertise in physiological signal monitoring and speech processing since 2018. They developed the integrated sensing platform ICARUS, which synchronously measures multiple physiological signals and records speech.


To encourage the deployment of such technologies, clinical validation is key. This is why in 2022 Inselspital’s UDEM approached IDIAP and CSEM to establish a joint research collaboration focusing on diabetes. To establish that speech acquisition systems offer a quality equivalent to the gold standard references readily used in the clinic for diagnostic purposes, UDEM, CSEM and IDIAP are currently conducting a clinical study in which voice samples are collected by the integrated sensing platform ICARUS along with clinically relevant measurements such as electrocardiogram, breathing patterns, etc. on large patient cohorts. Should the results from the pilot research project confirm the hypothesis, a larger clinical research project by the consortium will follow.

What is diabetes?

Diabetes is a rapidly growing chronic disease characterized by an abnormal increase and variation in blood sugar (glucose) levels. It can lead to serious chronic complications such as cardiovascular problems, and damage nerves, kidney, and eyes but also to relevant acute complications like low blood sugar (hypoglycemia).

To find out more about diabetes, please visit the webpage of Diabetes Switzerland.