Utilizing Data Clustering for Hypotheses Discovery in Multimodal Exercise and Health Interventions with Limited Sample Size

Abstract

Multimodal exercise (MME) interventions are beneficial for physical fitness, psychosocial health, cognition, or combinations of these aspects of health and wellness in healthy and clinical populations. However, MME intervention studies are laborious to conduct and difficult to assess due to the number of constructs needed to be assessed. The current study is a secondary analysis of the data obtained from the Golf for Healthy Aging (GHA) exercise intervention study. The goal of this work was to develop an analytical framework, using mathematical abstraction and modified K-means clustering, to assess the interrelations of GHA outcome variables, in order to discover novel, testable hypotheses regarding intervention effects for future studies.