This project exploits big data sets from population-based surveys existing worldwide with machine learning to track dementia and its determinants. Population-based surveys are a rich source of information about the health and economic well-being of adults over age 50. Twelve very similar surveys exist worldwide, in North and South America, in Europe, in Africa, and in Asia. Together, they constitute the HRS family studies. They collect information through interviews about demographics, health, mental status, wealth, lifestyle, healthcare, and employment. Yet, because the diagnosis of dementia is rarely mentioned in the HRS family studies, these studies have been poorly exploited so far in the field of dementia.
In preliminary studies, we have shown that the technique of machine learning could properly identify persons at risk of dementia in US and European HRS family surveys. Here, we expand the use of machine learning to all HRS family studies to provide a global perspective upon dementia. Such innovative use of survey data opens new avenues of research with important insights into the differential and comparative impact of culture, diet, environment, education, healthcare, wealth, gender, ethnicity, and race on the risk of dementia worldwide.
New information about dementia and its determinants in low- and middle-income countries will be offered while much of the current data comes from high-income countries. This project ultimately aims at influencing primary dementia care by providing simple algorithms to identify individuals at high risk of dementia in resource-poor areas where neurological expertise is lacking.