P11-07

Data-driven search for diseases whose patient numbers are associated with weather variability using LIFE study data

Kensei ORITA *1SUN KIM1Koichiro KATO1, 2Haruhisa FUKUDA3

1Department of Applied Chemistry, Graduate School of Engineering, Kyushu University
2Center for Molecular Systems, Kyushu University
3Department of Health Care Administration and Management, Graduate School of Medical Sciences, Kyushu University
( * E-mail: orita.kensei.423@s.kyushu-u.ac.jp )

In recent years, the climate has been changing rapidly, and these variations have become increasingly noticeable year by year. The relationship between weather and health status has long been studied, but only a limited number of diseases, such as sinusitis and asthma, have been identified due to the huge amount of data required. Therefore, this study overcame the lack of data by using a database produced by the Longevity Improvement & Fair Evidence (LIFE) study, a longitudinal cohort database that collected and linked administrative claims data for residents of participating municipalities.
In this study, monthly data on temperature, humidity, and precipitation were obtained from the Japan Meteorological Agency website for weather data. For medical data, monthly data on the number of patients with various diseases were calculated using claims data from the LIFE study. The number of patients was adjusted for age and then analyzed by sex. Seven years of data from April 2015 to March 2022 were used for the analysis.
A vector autoregressive (VAR) model was employed for the time series analysis. This model was chosen because it captures the dynamic interdependencies between multiple time series data and allows for a comprehensive analysis of the impact of changing weather patterns on health status.
The results of the analysis suggest an association between some time series of weather data and those of the patient counts. For example, the time series of average temperature and that of the number of patients in schizophrenia could be related. The results of this study may contribute to the development of models based on weather data to predict the number of patients with various diseases, which could lead to improved health care and public health measures in local communities. In the future, more practical models will be developed by analyzing a wider range of data sets and examining their applicability to other regions.