Using Mining Association Rules for the Dietary Assessment of Adults in Rural Area

Shim, J E; Paik, H Y; Shin, W S; Kim, K C

Abstract

This study was performed to find out a nutrient which could be used as an index for a quick and easy dietary assessment using mining association rules, the automated and intelligent database analysis. A Dietary survey with 24 hour recall method was conducted to 2037 adults over 30 years of age in rural area of Korea. Daily nutrient intake as the percent of the Recommended Daily Allowances(RDA) was applied to the association rules and nutrient intake over 75% of RDA was considered enough intake. Association rules in nutrient database show relation-ships among nutrients, and they are statements as this form, "50% of person that take protein under 60% of RDA also take calcium." Applying these rules between nutrients over 75% of RDA with 25% level of minimum support, we found that protein, phosphorus, niacin had 7 association rules and association rules of protein had higher level of confidence than the others. Applying between nutrients below 60% of RDA, calcium and vitamin A had the highest number of association rules, however, the levels of confidence were low. Protein had 5 association rules, 2nd higher number, and the levels of confidence were higher than others. From these results protein indicated nutrient intake patterns better than energy did, which is generally used as a representative nutrient. This results could be used in the dietary assessment and nutrition education of korean adults in rural area.