*2.1. Materials*

We collected data from ten cities/counties in central China, Yichun City (Yuanzhou Municipal District), Gao'an City, Wanzai County, Tonggu County, Pingxiang City (Anyuan Municipal District), Shangli County, Ji'an County, Xingan County, Liling City, and Chaling County, from May 2015 to September 2018 (178 weeks). These cities/counties have similar dietary habits and levels of health services. The dataset consists of two parts:


We constructed a data tuple per week for each city/county; the total number of tuples is 1780. If an indicator was measured more than once in a week, we took the mean value in the tuple.


**Table 1.** Types of food for morbidity prediction [21].


**Table 2.** All 227 contaminants used for morbidity prediction [21].
