Carbohydrate Estimation Supported by the GoCARB System

Diabetes Mellitus, Type 1 Clinical Trial

Official title:

Carbohydrate Estimation Supported by the GoCARB System in Individuals With Type 1 Diabetes

Clinical Trial Details — Status: Completed

Administrative data

• NCT number: NCT02546063
• Other study ID #: 100/15
• Status: Completed
• Phase: N/A
• First received: September 4, 2015
• Last updated: August 9, 2016
• Start date: August 2015
• Est. completion date: December 2015

Study information

• Verified date: August 2016
• Source: University Hospital Inselspital, Berne
• Contact: n/a
• Is FDA regulated: No
• Health authority: Switzerland: Ethikkommission
• Study type: Interventional

Clinical Trial Summary

The standard method for determining the carbohydrate content of a meal in patients with diabetes mellitus is the weighing of individual foods. However, in daily life, the weighing is not practical at all times. Inaccurate estimation of meal's CHO content, leads to wrong insulin doses and consequently to poor postprandial glucose control. Fact is that even well trained diabetic individuals find it difficult to estimate CHO precisely and that especially meals served on a plate are prone to false estimations underlining an emergent need for novel approaches to CHO estimation.

GoCarb is a computer vision-based system for calculating the carbohydrate content of meals. In a typical scenario, the user places a credit card-sized reference object next to the meal and acquires two images using his/her smartphone. A series of computer vision modules follows: the plate is detected and the different food items on the plate are automatically segmented and recognized, while their 3D shape is reconstructed. On the basis of the shape, the segmentation results and the reference card, the volume of each item is then estimated. The CHO content is calculated by combining the food types with its volumes, and by using the USDA nutritional database. Finally, the results are displayed to the user.

A preclinical study using the GoCarb system indicates that the system is able to estimate the meal's CHO content with higher accuracy than individuals with T1D. Furthermore, the feedback gathered by the participants showed that the system is easy to use even for non-smartphone users.

The aim of this randomized, cross-over pilot study is to investigate the benefits of an automated determination of the carbohydrate content of meals on glycemic control in subjects with type 1 diabetes mellitus with sensor-augmented insulin pump therapy.

Description:

Background

For individuals with type 1 diabetes (T1D) the current gold standard to evaluate the carbohydrate (CHO) amount of a meal is by carefully weighing its different components and calculating the CHO content using reference nutritional tables. The resulting CHO amount is then used to define the insulin dose needed to avoid an abnormal postprandial glucose profile. Since this is a cumbersome procedure in real life, diabetic individuals often estimate the CHO amount based on their personal experience. Especially for food served on a plate CHO estimates are often significantly over or underestimated leading to high variation in postprandial blood glucose. Besides the immediate risk of hypoglycaemia, there is emerging evidence that suboptimal control of postprandial glucose is affiliated with increased risk for long-term complications (e.g. diabetic micro- and macro-vascular diseases).

The effect of CHO counting in T1D control has been increasingly recognized and investigated. A meta-analysis including five studies on individuals with T1D has shown, that improved CHO counting accuracy reduces the HbA1c significantly (0.64% reduction in HbA1c compared to a control group). Teaching adult individuals with T1D to count CHO reduces HbA1c significantly and also leads to an improvement in quality of life. Similar findings are reported in children, in whom higher CHO counting accuracy is associated with a lower HbA1c. Even a short educational intervention of 4 weeks can still result in a significant and sustained effect on HbA1c reduction 9 months after without having an increase in hypoglycaemia. Lower CHO counting accuracy is a significant predictor of prolonged time in hyperglycaemic state. In one study with adults only 31% of the participants estimated the CHO content with an error of less than 20 grams per day and accurate CHO estimation were correlated with the lowest HbA1c values. In line with these findings another study has shown that individuals on intensive insulin therapy count CHO content of meals with an average error in the order of 16 grams or 21%. In general there is overestimation of small meals and a substantial underestimation of large meals. While breakfast (+8.5%) and snacks (-5%) were estimated fairly accurately, lunch (-28%) and dinner (-23%) are more prone to errors leading to an underestimation in the order of 30 grams. In children, an inaccuracy of ±10grams does not deteriorate the postprandial glycaemic control, whereas a ±20 grams variation significantly impacts the postprandial glycaemia.

The debate how to optimally estimate CHO intake is on-going and controversial. Fact is that even well trained diabetic individuals find it difficult to estimate CHO precisely and that especially meals served on a plate are prone to false estimations underlining an emergent need for novel approaches to CHO estimation. The investigators hypothesize that computer vision supported CHO estimation can have a beneficial impact on postprandial glucose control, ultimately leading to reduced episodes of hypoglycaemia and reduction in long term complications.

The recent advances in smartphone technologies and computer vision permitted the development of applications for the automatic dietary assessment through meal image analysis. The applications are using either a number of images or a short video of the upcoming meal, as captured by the user's smartphone. Although several systems have been proposed in the past decade, none of them is designed for individuals with diabetes, while they rely on strong assumptions, which often do not hold in real life, or require too much user input. The GoCARB system provides CHO estimations to individuals with T1D, by using only two meal images. The current version GoCARB has been designed to deal with

• elliptical plates with a flat base;

• single-dish images;

• fully visible food items. The system works without assumptions on the food shape, while it supports the following food classes: Pasta, green salad, meat, breaded food, beans, carrots, mashed potato, rice, potatoes, bread, cheese, egg, couscous, mushrooms and spätzli.

Objective

The purpose of this study is to investigate the effect of GoCARB-supported CHO estimation in the postprandial glucose control of individuals with T1D on sensor augmented insulin pump therapy.

In a typical scenario, the user places a credit card-sized reference object next to the meal and acquires two images using her/his smartphone. A graphical user interface guides the user in choosing the optimal angles for image acquisition based on the smartphone's built-in sensors. The images are then transmitted to a dedicated server via Wi-Fi or the mobile network. Then, the following computer vision algorithms are activated: the plate is detected and the different food items on the plate are automatically segmented and recognized, while their 3D shape is reconstructed. On the basis of the shape, the segmentation results and the reference card, the volume and CHO content of each item is then calculated by combining the food types with its volumes. Finally, the results are displayed to the user.

Methods

This study's overall objective is to perform a prospective randomized controlled clinical pilot study comparing postprandial glucose control assisted by the GoCARB system against the usual care approach by the participants. Scope of the study will be to investigate whether the postprandial glucose profile is improved by using the GoCARB prototype for CHO estimation. Postprandial glycaemia is evaluated using continuous glucose monitoring.

Recruitment information / eligibility

• Status: Completed
• Enrollment: 20
• Est. completion date: December 2015
• Est. primary completion date: December 2015
• Accepts healthy volunteers: No
• Gender: Both
• Age group: 18 Years and older

Eligibility

Inclusion Criteria:

• Type 1 diabetes

• Minimum age of 18 years old

• Sensor-augmented pump therapy for at least six months

• HbA1c levels within the last 4 months = 8.5%

• Familiar with carbohydrate (CHO) counting (e.g. CHO counting training in the past)

• Normal insulin sensitivity (reflected by a daily insulin requirement of 0.3-1.0 U/kg body weight)

• Able to comprehend German or English

• Written informed consent

Exclusion Criteria

• Relevant diabetic complications

• Hypoglycemia unawareness

• More than one episode of severe hypoglycemia as defined by American Diabetes Association in preceding 12 months

• Pregnancy

• Relevant psychiatric disorder

• Active neoplasia

• Participation in another study

• Other individuals especially in need of protection (according to the guidelines of the Swiss Academy of Medical Sciences)

Study Design

Allocation: Randomized, Endpoint Classification: Efficacy Study, Intervention Model: Crossover Assignment, Masking: Open Label

Related Conditions & MeSH terms

• Carbohydrates
• Diabetes Mellitus
• Diabetes Mellitus, Type 1

Intervention

Other:
• Smartphone App
The GoCARB system is a smartphone application designed to support type 1 diabetic patients with carbohydrate counting by providing automatic, accurate and near real-time CHO estimation for non-packed foods.

Locations

Country	Name	City	State
Switzerland	Division of Endocrinology, Diabetes and Clinical Nutrition, Bern University Hospital	Bern

Sponsors (1)

Lead Sponsor	Collaborator
University Hospital Inselspital, Berne

Country where clinical trial is conducted

Switzerland, 

References & Publications (3)

Agianniotis A, Anthimopoulos M, Daskalaki E, Drapela A, Stettler C, Diem P, Mougiakakou S. GoCARB in the Context of an Artificial Pancreas. J Diabetes Sci Technol. 2015 May;9(3):549-55. doi: 10.1177/1932296815583333. Epub 2015 Apr 21. — View Citation

Anthimopoulos M, Dehais J, Shevchik S, Ransford BH, Duke D, Diem P, Mougiakakou S. Computer vision-based carbohydrate estimation for type 1 patients with diabetes using smartphones. J Diabetes Sci Technol. 2015 May;9(3):507-15. doi: 10.1177/19322968155801 — View Citation

Anthimopoulos MM, Gianola L, Scarnato L, Diem P, Mougiakakou SG. A food recognition system for diabetic patients based on an optimized bag-of-features model. IEEE J Biomed Health Inform. 2014 Jul;18(4):1261-71. doi: 10.1109/JBHI.2014.2308928. — View Citation

Outcome

Type	Measure	Description	Time frame	Safety issue
Primary	Average of the postprandial area under the glucose curve (AUC) measured over three hours after each meal's start using Continuous Glucose Monitoring		14 days	No
Secondary	Composite of insulin-related parameters	Calculated from daily bolus, basal and correction insulin	14 days	No
Secondary	Glucose-related parameters	Postprandial glucose levels within, below and above target	14 days	No
Secondary	Daily nutritional behavior in individuals with T1D	Measured in kcal and documented by photographs	14 days	No
Secondary	User satisfaction	Composite measured by questionnaire: Participants will be invited to complete three questionnaires: i) a carbohydrate counting specific questionnaire including questions concerning educational level and former training of carbohydrate counting, and ii) a questionnaire related to GoCARB's features.	14 days	No

External Links

•   Click here for more information about GoCARB