Examining the Validity of Mobile-based Nutrient Assessment Tools for Chinese Patients with Diabetes (Preprint)

BACKGROUND Chinese patients with diabetes do not receive adequate education in dietary management to control blood glucose. Technology-based tools can help them track their dietary intake. However, limited research examined mobile dietary tracking tools with dish-based composition database for Chinese populations. OBJECTIVE The study aimed to examine the validity of mobile-based nutrient analysis tools against nutrient analysis conducted by dietitians on assessment of Chinese dishes. METHODS Registered Dietitian (RD)-performed nutrient analysis was used as the reference. Two mobile-based nutrient analysis tools were selected for the comparison: BD Diabetes Health Coach (an Artificial Intelligence (AI) Coach developed in the current study upon a dish composition database) and Boohoo App (a popular open-access dietary tracking tool in China). Both RDs and mobile-based tools conducted nutrient analysis of the same dishes commonly consumed in China regarding total calories, carbohydrate, fat, and protein content. Descriptive statistics were analyzed. Paired t-tests, Wilcoxon signed rank tests, linear regression and Bland-Altmanwere used to compare results between the RDs and the mobile tools. RESULTS Strong positive correlations between AI Coach- and RD-performed analyses results and moderate positive correlations between Boohoo-performed and RD-performed analyses results were observed for energy and all macronutrients. Bland-Altman plots showed small bias and narrow limit of agreement between AI Coach- and RD-performed analyses, indicating AI Coach-performed analysis is comparable to RD-performed analysis. CONCLUSIONS The AI Coach tool developed in the current study supported by a dish composition database consisting of commonly consumed Chinese dishes was an effective alternative to dietitians in total calorie and macronutrient analysis. This is promising in helping patients with diabetes manage blood glucose through dietary tracking.

DietSkan: Food Volume Estimation for Dietary Intake Analysis Using 3D Mesh Scanning

The prevalence of obesity, found in more than 38% of worldwide adults, is causing dietary measurements to become increasingly important. Most methods for tracking dietary intake utilize estimating the amount of food consumed to determine calories and nutritional content. Currently used methods of dietary tracking are either tedious or inaccurate. Our proposed method for dietary tracking is called DietSkan. It combines an off the shelf 3-Dimensional (3D) scanner, the Structure Sensor, with a smartphone application to produce a 3D reconstructed mesh scan of food items. The DietSkan process requires the desired food item to be scanned and exported for volume calculation. Then, using a 3D mesh manipulation tool, a 3D mesh, enclosing the consumed food, is constructed to obtain volume. The volume measurements achieved using the DietSkan algorithm average only 6% error and allow a user to track their dietary intake simply and effectively. The DietSkan system simplifies the estimation process and improves measurement accuracy when compared to current common practices.

The Effect of Adherence to Dietary Tracking on Weight Loss: Using HLM to Model Weight Loss over Time

The role of dietary tracking on weight loss remains unexplored despite being part of multiple diabetes and weight management programs. Hence, participants of the Diabetes Prevention and Management (DPM) program (12 months, 22 sessions) tracked their food intake for the duration of the study. A scatterplot of days tracked versus total weight loss revealed a nonlinear relationship. Hence, the number of possible tracking days was divided to create the 3 groups of participants: rare trackers (<33% total days tracked), inconsistent trackers (33–66% total days tracked), and consistent trackers (>66% total days tracked). After controlling for initial body mass index, hemoglobin A1c, and gender, only consistent trackers had significant weight loss (−9.99 pounds), following a linear relationship with consistent loss throughout the year. In addition, the weight loss trend for the rare and inconsistent trackers followed a nonlinear path, with the holidays slowing weight loss and the onset of summer increasing weight loss. These results show the importance of frequent dietary tracking for consistent long-term weight loss success.