RESUMEN
Nutrition is a key contributor to health. Recently, several studies have identified associations between factors such as microbiota composition and health-related responses to dietary intake, raising the potential of personalized nutritional recommendations. To further our understanding of personalized nutrition, detailed individual data must be collected from participants in their day-to-day lives. However, this is challenging in conventional studies that require clinical measurements and site visits. So-called digital or remote cohorts allow in situ data collection on a daily basis through mobile applications, online services, and wearable sensors, but they raise questions about study retention and data quality. "Food & You" is a personalized nutrition study implemented as a digital cohort in which participants track food intake, physical activity, gut microbiota, glycemia, and other data for two to four weeks. Here, we describe the study protocol, report on study completion rates, and describe the collected data, focusing on assessing their quality and reliability. Overall, the study collected data from over 1000 participants, including high-resolution data of nutritional intake of more than 46 million kcal collected from 315,126 dishes over 23,335 participant days, 1,470,030 blood glucose measurements, 49,110 survey responses, and 1,024 stool samples for gut microbiota analysis. Retention was high, with over 60% of the enrolled participants completing the study. Various data quality assessment efforts suggest the captured high-resolution nutritional data accurately reflect individual diet patterns, paving the way for digital cohorts as a typical study design for personalized nutrition.
RESUMEN
The automatic recognition of food on images has numerous interesting applications, including nutritional tracking in medical cohorts. The problem has received significant research attention, but an ongoing public benchmark on non-biased (i.e., not scraped from web) data to develop open and reproducible algorithms has been missing. Here, we report on the setup of such a benchmark using publicly available food images sourced through the mobile MyFoodRepo app used in research cohorts. Through four rounds, the benchmark released the MyFoodRepo-273 dataset constituting 24,119 images and a total of 39,325 segmented polygons categorized in 273 different classes. Models were evaluated on private tests sets from the same platform with 5,000 images and 7,865 annotations in the final round. Top-performing models on the 273 food categories reached a mean average precision of 0.568 (round 4) and a mean average recall of 0.885 (round 3), and were deployed in production use of the MyFoodRepo app. We present experimental validation of round 4 results, and discuss implications of the benchmark setup designed to increase the size and diversity of the dataset for future rounds.
