AI-assisted dietary assessment and personalized “nudges” in urban Ghana: Preparing for scale-up

Diet-related risk factors cause 11 million deaths annually, making them the largest single factor included in the Global Burden of Disease analyses (Afshin et al. 2019). Recent trends associated with urbanization and the “nutrition transition”—which is characterized by shifts to unhealthy diets and increased consumption of energy dense and nutrient poor processed foods and decreased physical activity—have led to increases in rates of overweight and obesity (Popkin et al. 2020). Regular data collection to document population-level dietary intake is essential for effective, evidence-based nutrition actions (Hargreaves et al. 2022). However, collecting and analyzing dietary data is complex and expensive (Bell et al. 2017). Dietary assessments typically involve the use of the multi-pass 24-hour recall (24HR) method, which has been validated in adults reporting their intake and/or that of their young children (Gibson and Ferguson 2008), and to some degree in adolescents. The costs associated with conducting a 24HR are approximately $500 per recall (Adams et al. 2022). Mobile phone-based tools have potential to lower the cost of diet assessment; however, evidence is lacking on the validity and feasibility of conducting phone-based assessments in low- and middle-income countries (LMICs) (Bell et al. 2017).