Supporting Meal Management in Type 1 Diabetes (SUMMIT1)

Carbohydrate count marks the cornerstone of Type 1 Diabetes management. Eventhough it is a crucial task, it is burdensome and prone to error. Therefore, the investigators want to explore the effect that SNAQ, a food analyser app would have in glycaemic control by facilitating the task of carbohydrate estimation.

Diet and physical activity are critically important in the lifestyle of people with type 1 diabetes. When diagnosed with the disease, people with type 1 diabetes are educated about nutritional goals and how to estimate nutritional content of food. Carbohydrates are the food component with the greatest impact on blood glucose levels and typical sources in the diet include starches, some vegetables, fruits, dairy products and sugars . Thus, people with type 1 diabetes are primarily being trained to estimate the carbohydrate content of food, a task that is also referred to as carbohydrate counting. Different methods can be used to count carbohydrate in food and drink. These include reading the nutritional labels, consulting reference books or websites, carrying a database on a personal digital assistant or using exchange tables which provides the carbohydrate content for typical serving sizes (e.g. 1 slice of bread). While nutritional information can be accessed through the above mentioned methods, the quantification of the portion sizes (if not indicated on the food package) requires the additional use of scale or measuring vessel. Given the required effort and time investment related to these methods, the great majority of people with type 1 diabetes count carbohydrates by visual estimation and experience. As a consequence, people's estimate often deviate substantially from ground truth values and average carbohydrate estimation errors reported in the literature are 20% or higher. Of note, more than 60% of individuals with diabetes report having trouble with carbohydrate counting, despite their awareness on its importance . Even in patients who are confident in applying carbohydrate counting, the daily task is perceived as major burden of diabetes self-management. Since carbohydrate counting is particularly demanding when eating fresh, non-packaged foods, a concerning trend towards unhealthy dietary choices with preference of prepackaged foods (with accessible nutrition facts) over whole foods is increasingly observed in people with type 1 diabetes. This is paralleled by an increasing prevalence of overweight and obesity in the type 1 diabetes population. Thus, even with the latest hybrid closed-loop insulin delivery technologies, adequate nutrition knowledge remains a cornerstone for satisfactory glucose control, metabolic health, and prevention of diabetes-related complications and comorbidities. With the development of new technologies embedded in modern smartphones (i.e. depth sensors), image-based methods to support food assessment have become widely available. Of particular use is the employment of well-established computer vision methodologies to estimate the quantity of food. When combined with food-recognition technologies and information from nutritional databases, a proposition of the nutritional content (e.g. carbohydrates, fat, proteins, fibres) can be made to the user on the basis of captured images and obviates the need for error prone visual estimations and mental calculations. Several such applications have become available and can support monitoring the diet as part of lifestyle management. Insights from a recent online survey suggest that a high proportion of people with type 1 diabetes believe that such new technologies for meal management could facilitate their daily self-management and would be interested in using such technology. Moreover, according to a recent study, such digital tools may promote diabetes education and food literacy which may particularly benefit those with a lower education level and with a history of depression. Amongst several options (e.g. Foodvisor, Calorie-Mamma, Lifesum) for image-based food tracking and analysis, SNAQ is one of the most commonly used app in people with type 1 diabetes. Up to date, more than 40000 users have downloaded the SNAQ app in their phones, of which 2,500 are living in Switzerland. The investigators have previously demonstrated that the system estimates the macronutrient content of real meals with satisfying accuracy. However, evidence with regards to the effect of the food analysis on daily self-management of people with type 1 diabetes (e.g. glucose control, meal patterns, perceived benefits) is currently lacking. The investigators therefore aim to address these aspects in a randomized-controlled study contrasting the use of the SNAQ app with people's traditional meal management techniques.

Carbohydrates, Continuous glucose measurement, Closed-loop systems, Type 1 Diabetes, Hyperglycaemia, Hypoglycaemia

The study will follow a randomized two-arm parallel design. Study visits will be done remotely via video calls or in-clinic when coinciding with usual care appointments. Following a baseline visit and before randomization, baseline characteristics and medical history of the participants will be collected (as detailed in section 4.3). Following randomization, the intervention group will use SNAQ app for the first 3 weeks while the control group will proceed without any modification/intervention by the study team. After the first 3 weeks, the control group will undergo 3 weeks of SNAQ app use (weeks 4-6). At the end of their respective SNAQ app periods (weeks 4-6 for the intervention group and weeks 7-9 for the control group), both groups will discontinue the use of SNAQ app for 3 weeks to assess sustainability of potential effects. Finally, both groups will be offered to use SNAQ app for 3 additional weeks as per their preference (follow-up period).

The control group will continue estimating the carbohydrate count using their traditional methods for the first three weeks of the study (baseline to V1).

SNAQ is a smartphone food analysis app that estimates the macronutrient content of a meal, based on a single image. The app first determines meal content in terms of food components with input from the user to correct or add further components (e.g. foods, ingredients, sauces, herbs or seasonings). Then, the total macronutrient and energy content of the meal is determined based on the estimated volume and information from a nutritional database. Of note, the application also allows for assessing nutritional content of packaged foods by means of a barcode scanning function. The user can always adapt proposed nutritional contents at their own discretion. Meal macronutrients alongside the food pictures are collected in a detailed log which allows users to review their dietary choices. The product is not conceived by its manufacturer to be used for medical purposes and can thus not be considered a medical device.

Patients will follow their traditional methods of carbohydrate counting during the control period. In addition to assess sustainability of the intervention, following the control period, the control group will also go an intervention period of 3 weeks using the SNAQ App.

Inclusion Criteria: Written informed consent Adults (aged 18 years or older) Type 1 diabetes (as defined by World Health Organization (WHO) for at least 12 month) Current use of a commercial hybrid closed-loop system HbA1c≤12% (measured within the past 3 months) Willing to use the SNAQ app on a daily basis for over 3 weeks The participant is willing to follow study specific instructions and share their treatment data with the study team Exclusion Criteria: Any physical or psychological disease or condition likely to interfere with the normal conduct of the study and interpretation of the study results Previous use of SNAQ app for more than 5 days within the past 3 months Self-reported pregnancy, planed pregnancy within next 3 months or breast-feeding Severe visual impairment Severe hearing impairment Lack of reliable telephone facility for contact Concomitant participation in another trial that interferes with the normal conduct of the study and interpretation of the study results Participant not proficient in German

Department of Diabetes, Endocrinology, Nutritional Medicine and Metabolism (UDEM), Inselspital, Bern University Hospital

Anonymised individual participant data will be shared after inquiry via a validated sharing platform (yet to be defined). Anonymised data packages will be available once the final study results are published in a peer-reviewed journal.