Bias-free Regression Analysis in Nutrition (BRAiN) Index: A Predictive Model of Cognitive Function

Evidence suggests the development of age-related dementia is strongly associated with numerous modifiable risk factors, particularly lifestyle behaviours such as diet, sleep, and physical activity. Yet, when exploring the dietary predictors of cognitive function, many scoring systems apply linear scales to eating patterns in a way that prioritizes simplicity over predictive accuracy. The present study employed a cross-validated elastic net regression algorithm on a large dataset of self-reported dietary data (n=28,968 individuals, mean age 56.3, age range 16-100) to identify the greatest nutritional predictors of participants performance on a validated online cognitive function test. Bias-free Regression Analysis in Nutrition (BRAiN) scores were computed, and their predictive power was found to outperform a well-recognized cognition-focused scoring system (Mediterranean-DASH Intervention for Neurodegenerative Delay, MIND) in held-out data. Non-linear relationships between intake level and cognition were noted for most foods, with greater consumption of animal and vegetable proteins, vegetables, nuts and seeds, and wholegrains, being positively associated with cognitive function, while the regular consumption of both refined fats and refined carbohydrates was negatively associated with cognitive function. Shifts in dietary patterns by age suggested that some relationships between diet and cognitive function in nutritional epidemiology may be confounded by age-related dietary trends. Our data-driven model can be easily adjusted over time to reflect potential shifts in dietary practices at the population level and inform relevant health authorities in the timely design of effective nutritional recommendations and guidelines to prevent or slow down the onset of age-related cognitive decline.