THE ISSUE

For low and lower-middle income countries (LLMIC), a ‘global nutrition transition’ exists involving the dual burden of malnutrition and non-communicable chronic diseases, with type 2 diabetes now becoming highly prevalent. For LLMIC risk factors related to dietary intake (e.g. low fruit and vegetable intake) and malnutrition are the top two contributors to burden of disease. In low-income countries malnutrition, is the leading cause of disease burden with dietary risks placed fifth.

The lack of comprehensive, reliable data on the contemporary dietary intakes of individuals in LLMIC hinders the ability to determine nutritional adequacy of intakes and provide timely intervention in these settings. Inherent logistical challenges (e.g. geography, infrastructure) within these settings combined with the resource-intensive nature of the dietary assessment methods commonly used (e.g. interviewer-administered 24-hour recalls and direct observation with weighing of food) can be time and cost prohibitive to regular use, further compounding this problem. In addition, due to a lack of skilled workers in LLMIC, regular and ongoing surveillance of nutritional adequacy of dietary intakes of individuals is often reliant on external organizations and individuals to undertake these activities.

OUR PROPOSED SOLUTION

The project will address this problem by developing the Voice-Image-Sensor technology for Individual Dietary Assessment (VISIDA) system, a novel dietary assessment method which combines voice, imaging and sensor technologies to overcome limitations of current methods used in LMIC settings. In addition to producing an acceptable and valid dietary assessment method, we will develop and evaluate an implementation framework to guide the use of the method in similar settings.

The VISIDA system is modular to facilitate iteration of the components as the technologies and techniques advance. 

The system has been developed with a focus on mothers and their children aged ≤5 years.

The VISIDA system consists of:

1. Image-voice food record smartphone app (Android) to actively collect dietary data on recipes and foods consumed from discrete plates per individual, along with foods served on shared plates and consumed by two or more members of a household. A fiducial marker is placed next the foods at the time of image capture, with onscreen guidance provided to the user to position the phone and the fiducial marker for image capture.

2. Wrist-worn sensor to detect hand-to-mouth gestures to detect eating occasions and other movements associated with eating. Video cameras (spherical and time-lapse) are used at various times in the project as ground truth measures for validating the movement data collected from the wrist sensor.

3. Web-based system to display, process and store dietary intake data collected from the smartphone app. Trained research assistants analyse the Image-voice food records within the web system to derive estimates of nutrient intake. Currently, the analysis is supported by automated and semi-automated processes to assist analysts in identification and quantification of food items contained with the images. Wrist sensor and time-lapse video data will be used to complement the intake data obtained from the smartphone app to account for the combination of discrete and shared plate eating.

The VISIDA system will allow for rapid individualized dietary assessment and also support for the data interpretation and the provision of feedback. The system will allow for identification of individuals with nutritional needs that should be prioritized to optimize health. Further, the VISIDA system can complement population surveillance strategies ensuring that vulnerable groups can be targeted to optimise intake and the evaluation of the effectiveness of nutrition interventions can occur in a timely manner.

An outcome from this project will be to create several dietary training modules for local health workers and nutritionists and an implementation framework for other LLMIC that will be made readily accessible online. Specifically with our partners in Cambodia and Tanzania, we will train local health workers in use of the VISIDA system and a range of research skills to further strengthen capacity at these sites for the ongoing assessment of dietary intake in these settings.