Modelling future population requirements for energy and water

Background:
A major challenge today is to provide the global human population with adequate supplies of food-energy and safe drinking water (1,2). Moreover, this challenge will inevitably change in the future, due to interacting factors such as demographic trends and changes in human nutritional status, in particular the burgeoning global obesity epidemic. Predicting likely future population requirements for energy and water is extremely difficult, however, due to the lack of objective data on per capita food-energy and water intakes. It is well established that dietary intake records are highly prone to error, and also relate to what individuals actually consumed, rather than what would have been optimal for them to consume. Food sales data are widely available, but only indicate what food-energy was purchased, not consumed, while large proportions of the food produced globally go to waste. Meanwhile, global food-energy requirements are increasing in association with the rapid increase in overweight/obesity, and are also influenced by trends to more sedentary lifestyles. Both energy intake and water turnover can be measured accurately and objectively in individuals of any age using a stable isotope probe, the double-labelled water (DLW) method. The International Atomic Energy Agency has recently opened a global database of such measurements, already exceeding 6500 individual measurements from over 25 countries (3). Further data collection is ongoing.

Aims/Objectives:
The project will use these objective data on energy and water turnover to predict population food-energy and water requirements, by integrating needs across the demographic profile of individual countries. The student will initially focus on developing an appropriate prediction model of food energy and water requirements, using machine learning techniques. This model will then be applied to construct training and validation sets based on simulations using data from high-income countries.  The primary aim is to predict changes in food-energy and water requirements associated with different levels of success, and underlying strategies, for resolving the obesity epidemic. For example, tackling obesity by promoting dietary constraint is expected to reduce food-energy requirements, whereas promoting physical activity may reduce weight and improve health, but maintain energy turnover closer to baseline levels. The model will then be validated and tested with the latest DLW observations available.  Aside from the scientific objectives, the project provides an excellent training opportunity for a researcher with interests in computational global health.

Methods:
The IAEA database contains data on energy and water turnover and body composition in individuals across the entire age-span from birth to >90 years, allowing population values to be integrated in association with age, sex and nutritional status. The machine learning models will be fitted in the Bayesian framework.  Resampling methods, e.g. bootstrapping and cross-validation, will be used to address overfitting, generalisability, and trade-off between bias and variance. Later versions of models will incorporate estimates of the associations of childhood obesity with adult obesity, and child obesity in the next generation, in order to improve understanding of how contemporary efforts to reduce childhood and maternal obesity may impact food and water requirements in subsequent decades as populations age and reproduce.

Timeline:
Year 1: Training in Bayesian machine learning procedures using R or python.  Development of a basic prediction model and construction of training, validation on datasets from the UK, and evaluation of its predictive properties.
Year 2: Analysis of data from high-income countries, modelling potential changes to population food-energy and water requirements under different simulated scenarios for tackling obesity, and simulating life-course and inter-generational effects.
Year 3: Analysis of data from low-/middle-income countries, to improve predictions of global food-energy and water requirements in association with global demographic trends and changes in nutritional status.

References:
1.  Godfray HC et al. Food security: the challenge of feeding 9 billion people. Science 2010; 327, 812-8.
2.  Boretti A, Rosa L. Reassessing the projections of the World Water Development Report. NPJ Clean Water 2019; 2: 15.
3.  Speakman JR et al. The International Atomic Energy Agency International Doubly Labelled Water Database: Aims, scope and procedures. Ann Nutr Metab. 2019;75(2):114-118.
4.  James G et al. An introduction to statistical learning, with applications in R. New York: Springer. 2013.
5.  Wickham, H. Advanced R (2nd edition). Boca Raton: CRC Press. 2019.