The long-standing quest in public health research has been focused in trying to find ways to improve the health of the individuals. Considering the ubiquity of smartphones and wearable devices, as well as their capabilities of providing continuous streams of data, it is now possible to build very powerful models which can be used to predict the health of the individuals in the population. Different from the conventional studies, this study represents multi-disciplinary effort to use the advancements in machine learning algorithms on a large set of various physiological, lifestyle and environmental factors to predict different health outcomes with a higher accuracy.
The project will last for three years and is organized into four phases. In the first phase, we plan to explore the data and extract the features of interest for analyzing various diseases (e.g., cardiovascular diseases, diabetes etc.) and mortality. In the second phase, we plan to evaluate different machine learning algorithms in training and validating the predictive models. In the third stage, we plan to construct a single and intuitive data-driven health index value. In the fourth phase, we plan to document our findings and prepare 1-2 journal papers which will be published in the peer review literature.
We believe that there are many important outcomes from this project, as the methodology devised in this project will lay the foundation for designing more efficient algorithms that can be used to assess the health condition of any population. Moreover, it will also allow us to generate relevant and intuitive messages which can positively impact the promotion of public health (e.g. sufficient and good quality of sleep and having at least 30min of moderate-to-vigorous physical activity per day is associated with better life quality and decreased risk of cardiometabolic disease).