Opioid Prescription Predictive Analysis

Team: The Cavalry

Submission By

  • Akshay Tambe (apt321@nyu.edu)
  • Aditya Bhatt (apb462@nyu.edu)

Source Code

  • Complete Source Code along with hyperparameter tuning: apt321_apb462_Project_Notebook_Final.ipynb
  • Hyperparameter tuning script: hpc_tuning.py
  • Batch script to run in HPC: hpc_tuning.sbatch

Data Science for Social Good

Problem: Opioid Prescription helps to treat moderate to severe pain but it also leads to addiction and hence, people misuse it by consuming it at higher rate. Overdose of Opioids leads to death.

  1. How to improve ways the opioids are prescribed?
  2. How to reduce the death toll due to drug overdose in North America?

Goal: Detect opium components in the data and predict prescribers with opioid prescriptions which may lead to drug addiction.

Data: Prescription data with drug components, medicare domain, opium drug list, state-wise overdose deaths.

Target Variable: Opioid Prescriber or not?

Impact: Model is able to predict prescriber with opioid prescription and individual chemicals and features of high importance causing opioid addiction.