Toward Concurrent Identification of Human Activities with a Single Unifying Neural Network Classification: First Step.

The characterization of human behavior in real-world contexts is critical for developing a comprehensive model of human health. Recent technological advancements have enabled wearables and sensors to passively and unobtrusively record and presumably quantify human behavior. Better understanding human activities in unobtrusive and passive ways is an indispensable tool in understanding the relationship between behavioral determinants of health and diseases. Adult individuals (N = 60) emulated the behaviors of smoking, exercising, eating, and medication (pill) taking in a laboratory setting while equipped with smartwatches that captured accelerometer data. The collected data underwent expert annotation and was used to train a deep neural network integrating convolutional and long short-term memory architectures to effectively segment time series into discrete activities. An average macro-F1 score of at least 85.1 resulted from a rigorous leave-one-subject-out cross-validation procedure conducted across participants. The score indicates the method's high performance and potential for real-world applications, such as identifying health behaviors and informing strategies to influence health. Collectively, we demonstrated the potential of AI and its contributing role to healthcare during the early phases of diagnosis, prognosis, and/or intervention. From predictive analytics to personalized treatment plans, AI has the potential to assist healthcare professionals in making informed decisions, leading to more efficient and tailored patient care.

The effects of snack foods of different energy density on self-served portions and consumption in preschool children.

It is recommended that preschoolers serve themselves their own food portions; however, it is unclear what factors influence the amount they select for consumption, and particularly how their selected portions are influenced by food properties such as energy density, volume, and weight. We offered preschool children snacks differing in energy density (ED) and investigated the effects on the amounts they served and then consumed. In a crossover design, 52 children aged 4-6 y (46% girls; 21% overweight) ate an afternoon snack on 2 days in their childcare classrooms. Before each snack time, children served the amount they would like to eat of 4 snacks presented in equal volumes but differing in ED (higher-ED: pretzels, cookies; lower-ED: strawberries, carrots). Across the 2 sessions, children were given their self-served amount of either pretzels (3.9 kcal/g) or strawberries (0.3 kcal/g) and intake was measured. Later, children tasted all 4 snacks and rated liking. Results showed that the portions children served themselves were influenced by their liking ratings (p = 0.0006), but after accounting for liking, the volumes they served were similar for all 4 foods (p = 0.27). At snack time, children ate a greater proportion of self-served strawberries (92 ± 4%) than pretzels (73 ± 4%; p = 0.0003), but because of the ED difference they consumed 55 ± 4 kcal more from pretzels than strawberries (p < 0.0001). The difference in snack intake by volume was not attributable to liking ratings (p = 0.87). That children served a consistent volume of similarly-liked snacks suggests that their portions were affected more by visual cues than by weight or energy content. Despite eating a greater volume of lower-ED strawberries, children consumed more energy from the higher-ED pretzels, highlighting the contribution of energy density to children's energy intake.

Children's Portion Selection Is Predicted by Food Liking and Is Related to Intake in Response to Increased Portions.

BACKGROUND: When children choose amounts of food to eat, it is unclear what influences the portions they select and whether their selections are related to the amounts they consume. OBJECTIVES: Using a computer survey, we investigated the effect of food liking on portion selection in middle childhood and examined how children's selections were related to measured intake at meals in which portions of all foods were varied across 4 test days. METHODS: Fifty-one children aged 7-10 y completed a computer survey of 20 common foods with a range of energy density. For each food, the survey presented sliding scales with 5 images varying in portion size and children indicated their liking and the amount they would eat at a specified meal or snack. On 4 test days in a randomized crossover design, children were served a meal of 6 foods from the survey with portions of 100%, 133%, 167%, or 200% of baseline amounts. Data were analyzed using random coefficients models. RESULTS: Across the 20 foods used in the survey, portion selection ratings were predicted by food liking ratings (P < 0.0001). After accounting for liking, portion selection ratings did not vary by food energy density (P = 0.50). At the meals, intake of all 6 foods increased when larger portions were served (P = 0.002). Furthermore, the selected portion of a food on the survey was positively related both to intake of that food at the 100%-portion meal (P = 0.014) and to increased intake as larger portions were served (P < 0.0001). CONCLUSIONS: Children aged 7-10 y were able to use a computer survey to choose food portions that predicted their measured intake in response to increased portions. The relation of liking to selection and intake indicates that to promote children's consumption of larger portions of low-energy-dense foods such as fruits and vegetables, these foods must be well liked by the children. This trial was registered at clinicaltrials.gov as NCT02759523.

Promoting vegetable intake in preschool children: Independent and combined effects of portion size and flavor enhancement.

This study investigated the independent and combined effects on preschool children's vegetable intake of serving a larger portion of vegetables and enhancing their flavor. In a crossover design, lunch was served in childcare centers once a week for four weeks to 67 children aged 3-5 y (26 boys, 41 girls). The meal consisted of two familiar vegetables (broccoli and corn) served with fish sticks, rice, ketchup, applesauce, and milk. Across the four meals, we varied the portion of vegetables (60 or 120 g total weight, served as equal weights of broccoli and corn) and served them either plain or enhanced (6.6% light butter and 0.5% salt by weight). All meals were consumed ad libitum and were weighed to determine intake. Doubling the portion of vegetables led to greater consumption of both broccoli and corn (both p < 0.0001) and increased meal vegetable intake by 68% (mean ± SEM 21 ± 3 g). Enhancing vegetables with butter and salt, however, did not influence their intake (p = 0.13), nor did flavor enhancement modify the effect of portion size on intake (p = 0.10). Intake of other meal components did not change when the vegetable portion was doubled (p = 0.57); thus, for the entire meal, the increase in vegetable consumption led to a 5% increase in energy intake (13 ± 5 kcal; p = 0.02). Ratings indicated that children had similar liking for the plain and enhanced versions of each vegetable (both p > 0.31). All versions of vegetables were well-liked, as indicated by ≥ 76% of the children rating them as "yummy" or "just okay". Serving a larger portion of vegetables at a meal was an effective strategy to promote vegetable intake in children, but when well-liked vegetables were served, adding butter and salt was not necessary to increase consumption.

Evaluating the Validity of the PortionSize Smartphone Application for Estimating Dietary Intake in Free-Living Conditions: A Pilot Study.

OBJECTIVE: Evaluate the validity of the PortionSize application. METHODS: In this pilot study, 14 adults used PortionSize to record their free-living food intake over 3 consecutive days. Digital photography was the criterion measure, and the main outcomes were estimated intake of food (grams), energy (kilocalories), and food groups. Equivalence tests with ±25% equivalence bounds and Bland-Altman analysis were performed. RESULTS: Estimated gram intake from PortionSize was equivalent (P < 0.001) to digital photography estimates. PortionSize and digital photography estimated energy intake, however, were not equivalent (P = 0.08), with larger estimates from PortionSize. In addition, PortionSize and digital photography were equivalent for vegetable intake (P = 0.01), but PortionSize had larger estimates of fruits, grains, dairy, and protein intake (P >0.07; error range 11% to 23%). CONCLUSIONS AND IMPLICATIONS: Compared with digital photography, PortionSize accurately estimated food intake and had reasonable error rates for other nutrients; however, it overestimated energy intake, indicating further application improvements are needed for free-living conditions.