A Study of the Relationship Between Driving and Health Based on Large-Scale Data Analysis Using PLSA and t-SNE

Abstract

The purpose of this study is to facilitate knowledge discovery about the relationship between driving and health among the elderly. In the Iwaki Health Promotion Project that is an annual project conducted by Hirosaki University, we have included a survey on driving for the first time in 2019. After linking the data obtained from the survey with four years of health data for 2016–2019, we have utilized PLSA as a machine learning method to cluster those data in an integrated manner. As a result, we have found latent classes broadly classified according to whether the health level has been generally high or low. Also, when we have focused on a specific health item, for example, cognitive function, we have found some people with higher and lower maintenance of cognitive function over four years, even if they have belonged to a same latent class. To characterize these differences in detail, we have utilized t-SNE as a machine learning method. As a result, we have found that “I like driving” as a factor related to the Kansei (sensitivity) may characterize the high maintenance of cognitive function. For those who like driving, it is considered that the high maintenance of cognitive function may be occurred because they enjoy driving, have a wider range of activities, and increase the possibility of multitasking.

Keywords

• Cognitive function
• driving
• health
• machine learning
• PLSA
• t-SNE