Discovering Hidden Patterns in US Health-Related Open Data with Machine Learning

Harvey Alférez, Ph.D

Data Scientist, School of Engineering and Technology, Montemorelos University, Mexico
www.harveyalferez.com

 
Open data is data that can be freely used, re-used and redistributed by anyone - subject only, at most, to the requirement to attribute and share alike [1]. Do you see the relationship between open data and what the Bible says in Matthew 10:8 - “Freely you have received; freely give”? 
 
At the School of Engineering of Technology, Montemorelos University (Mexico), we are especially interested in discovering new knowledge from open data by means of machine learning, which is a key component of data science. Basically, machine learning is the sub-field of computer science that “gives computers the ability to learn without being explicitly programmed” [2]. 
 
In my artificial intelligence course, Myriam Torres and I studied open data related to diabetes in the US by means of machine learning. I believe that health-related research is important in order to understand the relevant needs of people in North America. In fact, through the ministry of Ellen G. White, God indicated that the “gospel of health” was to be the “right arm” of the third angel’s message [3].
 
In this study, we analyzed the Community Health Status Indicators (CHSI) open dataset provided by the Centers for Disease Control and Prevention (CDC) on Data.Gov [4]. Analysis was carried out by means of Weka [5], which is a suite of machine learning software. Weka is free software. The CHSI dataset provides key health indicators for local communities. It contains over 200 measures for each of the 3,141 United States counties [4]. In this study, we were specifically interested in the following indicators: number of people with diabetes, primary care physicians per 100,000 people, medicare beneficiaries, community/migrant health centers located in the county, and number of dentists per 100,000 people. 
 
Some of our initial findings are as follows:

1. An increasing number of primary care physicians correlates with a decreasing number of people with diabetes per county.
2. An increasing number of medicare beneficiaries correlates with a decreasing number of people with diabetes per county.
3. An increasing number of community/migrant health centers correlates with a decreasing number of people with diabetes per county.
4. An increasing number of dentists correlates with a decreasing number of people with diabetes per county.

The first three results indicate that medical attention can play a key role to prevent diabetes in communities. The last result is surprising because it shows that dentists can also help to prevent diabetes. Maybe their recommendations on the consumption of sugar and junk food may not only promote dental health but also may help to reduce risk factors for diabetes. Along with another group of students, I am carrying out additional experiments to mine deeper in this and other health-related datasets. 
 
Are you interested in applying data science to understand the needs of your community? If so, do not hesitate to contact Social Media + Big Data Services. Comment below.
 
References
 
1. Open Knowledge International, “What is Open Data,” http://opendatahandbook.org/guide/en/what-is-open-data/.
 
2. P. Simon, “Too Big to Ignore: The Business Case for Big Data,” Wiley (2013). ISBN 978-1-118-63817-0.
 
3. E.G. White, “The Canvassing Work,” Review and Herald (1899), https://text.egwwritings.org/publication.php?pubtype=Periodical&bookCode=RH&lang=en&year=1899&month=June&day=20
 
4. Centers for Disease Control and Prevention, “Community Health Status Indicators (CHSI) to Combat Obesity, Heart Disease and Cancer” (2016), https://catalog.data.gov/dataset/community-health-status-indicators-chsi-to-combat-obesity-heart-disease-and-cancer.
 
5. The University of Waikato, “Weka 3: Data Mining Software in Java,” (n.d.), http://www.cs.waikato.ac.nz/ml/weka/.