Skip to content

Class project (15.071 Analytics Edge, taught by Prof. Alexandre Jacquillat) on the detection of opioid overselling in the US. In collaboration with Joey Aramouni, Girish Govindarajan and Rihab Rebai.

Notifications You must be signed in to change notification settings

shmodal/Detecting_opioid_overprescription_in_the_US

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

5 Commits
 
 
 
 
 
 

Repository files navigation

Detecting opioid overprescription in the US

The US suffers from an opioid epidemic, in which there is an extensive overuse of opioid drugs (10.8M people misused opioid prescriptions in 2018). There are large costs of this including loss of lives, health care, criminal justice, and lost economic productivity. Our objective will be to identify pharmacies that have filled an excessive number of opioid prescriptions in the states of Kentucky, Tennessee, and West Virginia between the years 2006 and 2012. This will point to pharmacies who may be illegally distributing opioids or neighboring doctors who overprescribe opioids. It is also a proxy for understanding which areas are suffering the most opioid abuse. Thus, we aim to inform senior decision-makers (1) which pharmacies should be investigated for potential malpractice and (2) which counties to focus interventions on.

The main data source is the Automation of Reports and Consolidated Orders System (ARCOS) which was established by the Drug Enforcement Agency (DEA). The dataset tracks the path of every opioid pain pill (oxycodone and hydrocodone), from manufacturer to pharmacy, in the United States between 2006 and 2012. Also, secondary sources of data used are US census data (population, age, income, population), and health indicators (percentage insured and number of physicians per county)

Methods used :

(1) Visualization - Calculate metrics such as number of pills purchased per capita at the zip code level and visualize on heat maps.

(2) Time series - Graph trend lines to study the rate of opioid purchases by individual pharmacies. A peak in the rate of opioid purchases may indicate the pharmacy has started to oversell opioids.

(3) Clustering - Cluster pharmacies based on demographics, density of pharmacy in the areas, number of doctors, and other external environmental factors. This serves as a proxy for regrouping pharmacies with a similar market for opioid pills. Within each group of similar pharmacies, we looked for outliers in number of sales.

About

Class project (15.071 Analytics Edge, taught by Prof. Alexandre Jacquillat) on the detection of opioid overselling in the US. In collaboration with Joey Aramouni, Girish Govindarajan and Rihab Rebai.

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published