Current public health guidelines on physical activity and sleep duration are limited by a reliance on subjective self-reported evidence. Using data from simple wrist-worn activity monitors, we developed a tailored machine learning model, using balanced random forests with Hidden Markov Models, to reliably detect a number of activity modes. We show that physical activity and sleep behaviours can be classified with 87% accuracy in 159,504 minutes of recorded free-living behaviours from 132 adults. These trained models can be used to infer fine resolution activity patterns at the population scale in 96,220 participants. For example, we find that men spend more time in both low- and high- intensity behaviours, while women spend more time in mixed behaviours. Walking time is highest in spring and sleep time lowest during the summer. This work opens the possibility of future public health guidelines informed by the health consequences associated with specific, objectively measured, physical activity and sleep behaviours.

Background The provision and over-consumption of foods high in energy, saturated fat, free sugars or salt are important risk factors for poor diet and ill-health. In the UK, policies seek to drive improvement through voluntary reformulation of single nutrients in key food groups. There has been little consideration of the overall progress by individual companies. This study assesses recent changes in the nutrient profile of brands and products sold by the top 10 food and beverage companies in the UK. Methods The FSA/Ofcom nutrient profile model was applied to the nutrient composition data for all products manufactured by the top 10 food and beverage companies and weighted by volume sales. The mean nutrient profiling score, on a scale of 1–100 with thresholds for healthy products being 62 for foods and 68 for drinks, was used to rank companies and food categories between 2015 and 2018, and to calculate the proportion of individual products and sales that are considered by the UK Government to be healthy. Results Between 2015 and 2018 there was little change in the sales-weighted nutrient profiling score of the top 10 companies (49 to 51; p = 0.28) or the proportion of products classified as healthy (46% to 48%; p = 0.23). Of the top five brands sold by each of the ten companies, only six brands among ten companies improved their nutrient profiling score by 20% or more. The proportion of total volume sales classified as healthy increased from 44% to 51% (p = 0.07) driven by an increase in the volume sales of bottled water, low/no calorie carbonates and juices, but after removing soft drinks, the proportion of foods classified as healthy decreased from 7% to 6% (p = 33). Conclusions The UK voluntary reformulation policies, setting targets for reductions in calories, sugar and salt, do not appear to have led to significant changes in the nutritional quality of foods, though there has been progress in soft drinks where the soft drink industry levy also applies. Further policy action is needed to incentivise companies to make more substantive changes in product composition to support consumers to achieve a healthier diet.

Objective Current polysomnography-validated measures of sleep status from wrist-worn accelerometers cannot be used in fully automated analysis as they rely on self-reported sleep-onset and -end (sleep-boundary) information. We set out to develop an automated, data-driven approach to sleep-boundary detection from wrist-worn accelerometer data. MethodsOn three separate occasions, participants were asked to wear a GENEActiv ® wrist-worn accelerometer for nine days and concurrently complete sleep diaries with lights-off, asleep and wake-up information. We developed and evaluated three data-driven methods for sleep-boundary detection: a change-point detection based method , a thresholding method and a random forest classifier based method. Mean absolute errors between automatically-derived and self-reported sleep-onset and wake-up times were recorded in addition to kappa statistics for the minute-by-minute performance of each of the methods. ConclusionOur methods provide a data-driven approach to detect sleep-onset and -end times without the need for self-reported sleep-boundary information. The methods are likely

Current public health guidelines on physical activity and sleep duration are limited by a reliance on subjective self-reported evidence. Using data from simple wrist-worn activity monitors, we developed a tailored machine learning model, using balanced random forests with Markov confusion matrices, to reliably detect a number of activity modes. We show that physical activity and sleep behaviours can be classified with 87% accuracy in 84,616 minutes of recorded free-living behaviours from 57 adults. These trained models can be used to infer fine resolution activity patterns at the population scale in 96,609 participants. For example, we find that men spend more time in both low-and high-intensity behaviours, while women spend more time in mixed behaviours. Walking time is highest in spring and sleep time lowest during the summer. This work opens the possibility of future public health guidelines informed by the health consequences associated with specific, objectively measured, physical activity and sleep behaviours.

Background Excess consumption of salt is linked to an increased risk of hypertension and cardiovascular disease. The United Kingdom has had a comprehensive salt reduction programme since 2003, setting a series of progressively lower, product-specific reformulation targets for the food industry, combined with advice to consumers to reduce salt. The aim of this study was to assess the changes in the sales-weighted mean salt content of grocery foods sold through retail between 2015 and 2020 by category and company. Methods and findings Information for products, including salt content (g/100 g), was collected online from retailer websites for 6 consecutive years (2015 to 2020) and was matched with brand-level retail sales data from Euromonitor for 395 brands. The sales-weighted mean salt content and total volume of salt sold were calculated by category and company. The mean salt content of included foods fell by 0.05 g/100 g, from 1.04 g/100 g in 2015 to 0.90 g/100 g in 2020, equivalent to −4.2% (p = 0.13). The categories with the highest salt content in 2020 were savoury snacks (1.6 g/100 g) and cheese (1.6 g/100 g), and the categories that saw the greatest reductions in mean salt content over time were breakfast cereals (−16.0%, p = 0.65); processed beans, potatoes, and vegetables (−10.6%, p = 0.11); and meat, seafood, and alternatives (−9.2%, p = 0.56). The total volume of salt sold fell from 2.41 g per person per day to 2.25 g per person per day, a reduction of 0.16 g or 6.7% (p = 0.54). The majority (63%) of this decrease was attributable to changes in mean salt content, with the remaining 37% accounted for by reductions in sales. Across the top 5 companies in each of 9 categories, the volume of salt sold decreased in 26 and increased in 19 cases. This study is limited by its exclusion of foods purchased out of the home, including at restaurants, cafes, and takeaways. It also does not include salt added at the table, or that naturally occurring in foods, meaning the findings underrepresent the population’s total salt intake. The assumption was also made that the products matched with the sales data were entirely representative of the brand, which may not be the case if products are sold exclusively in convenience stores or markets, which are not included in this database. Conclusions There has been a small decline in the salt content of foods and total volume of salt sold between 2015 and 2020, but observed changes were not statistically significant so could be due to random variations over time. We suggest that mandatory reporting of salt sales by large food companies would increase the transparency of how individual businesses are progressing towards the salt reduction targets.