To assess whether machine learning can improve CVD risk prediction, compared to traditional statistical methods and to identify the minimum informative set of self-reported questionnaire data that can accurately predict CVD outcome events in the Australian 45 and Up Study.