Virginia is currently under a public health emergency as a result of the opioid addiction crisis. Our state has been severely impacted by opioid abuse, and the situation has been escalating surprisingly fast in recent years. In 1999, the first year for which such data is available, approximately 23 people died from abuse of prescription opioids. However, by 2017, the most recent year for which complete data is available, 1445 people died of the overdose of Fentanyl, Heroin or Prescription Opioid, an epidemic increase of roughly 6200%. Moreover, the data of 2015-2017 alone, showed us that there was a staggering increase of more than 60% in death attributed to drug overdose. Drug-related deaths have risen unrelentingly, and the drugs kill more people annually in Virginia than either car crashes or gunfire. This situation needs to be controlled and we should contribute our efforts as community members.Our project is intended to study the drug abuse problem here in Virginia using machine learning methods, and bring awareness to this widespread public health issue by presenting a thorough analysis of the problem, and possible solutions for different parties to reverse the epidemic of opioid drug overdose deaths and protect the public from overdose and other harms.
This project examines the opioid addiction crisis in the state of Virginia through related indicators such as Overdose Deaths, ED Visits, Hepatitis C and Diagnosed HIV. By examining the occurrence, we visualized the Virginia Opioid dataset and discovered correlations to opioid addiction in the state of Virginia. These correlations include age group, case count, rate and year for each related opioid addiction indicator. In this project, we use two major strategies: (1) Analyzing and Visualizing the data-set to discover and understand patterns and re-occurrences for the case being studied and (2) Carrying out Machine Learning Techniques in order to find hidden structures and anomalies on data. This projects utilizes machine learning techniques such as Isolation Forest, K-Means, Mean Shift and Dimensionality reduction techniques such as Principal component analysis(PCA) to better represent and understand the data-set being worked on.
see it in report directory.
This work has been a part of the Machine Learning for Virginia project at the University of Virginia in Fall 2018. We would also like to thank the Virginia Department of Health for providing the Virginia Opioid Dashboard Dataset.