Identifying risk patterns in Public Health data through Association Rules

Rao, Rohini R and Makkithaya, Krishnamoorthi (2016) Identifying risk patterns in Public Health data through Association Rules. The Journal of BMESI. pp. 30-34.

[img] PDF
1851.pdf - Published Version
Restricted to Registered users only

Download (580kB) | Request a copy

Abstract

The impact of Public Health services can be tremendously improved with the proper use of Information and Communication Technologies. The patient’s health status is recorded effectively, into sufficiently large data sets, these public health data records could be analyzed to identify risk factors for many diseases. The primary objective of this paper, is to determine if data mining techniques can be used on public health data records, to discover previously unknown health related risk factors or risk patterns, from the data. A risk pattern is a frequently occurring data pattern in large data sets, which identifies cohorts of patients who are vulnerable to a risk outcome. The second objective is to study the impact of family history, lifestyle and socioeconomic conditions on the mothers’ health status and also the health and growth patterns of children below the age of five. The data mining tool used was WEKA and two well known data mining techniques, tree based classifiers & association rule miners was applied on the data. Classification trees are generated to find the set of attributes which are correlated to the health condition or status. Association Rule Mining was performed on the relevant data attributes to find frequent risk patterns. The association rules generated have been further enhanced and validated with a statistical measure called relative risk. The results obtained, prove that association rule mining is a useful tool for exploratory analysis of large public health data sets. The method could help epidemiologists find risk patterns. The authors found a strong association between age of menarche and a pregnancy related symptom edema. The results also found that the girl child from poor socio-economic families tend to be underweight. Based on the risk patterns, various interventions (medical and non medical) and preventive health policies to tackle disease at the population level can be effectively designed and implemented.

Item Type: Article
Uncontrolled Keywords: risk patterns, risk factors, data mining, association rule mining, public health surveillance, epidemiology
Subjects: Engineering > MIT Manipal > Computer Science and Engineering
Engineering > MIT Manipal > MCA
Depositing User: MIT Library
Date Deposited: 12 Jan 2017 11:33
Last Modified: 12 Jan 2017 11:33
URI: http://eprints.manipal.edu/id/eprint/148042

Actions (login required)

View Item View Item