Department of Computer Science and Engineering

Project Abstract

In today’s medical community there is a prevalence of medications that are being pushed as a treatment for symptoms or through other means that are putting patient’s at risk of addiction. There is no way to determine a patients risk of addiction, and no standardized means of ensuring a patients level of pain that is accepted as accurate. Because of this there is an opioid epidemic that has made part of this country addicts to the medications that should be curing them. 

Prospect is a clinician's means of preventing opioid substance abuse, when prescribing medication. Based on six metrics that track a patient’s various statistics, a doctor can suggest that patient’s risk of addiction. With Prospect, that doctor can use the metrics and a machine learning model to classify that patients risk based on learned trends in the model. It is a product that is designed to predict opioid addiction using a neural network that is learning off of the trends of patients records. These records comprise of clinical evaluation metrics that vary based on each patents interactions.

The goal is to model a 7-dimensional problems space into a risk category for the patient. This uses a deep neural network and classifies the patient’s risk down to three categories (low, med, high). These categories can be adjusted and since the network is trained on current patient data these parameters allow cliexa the freedom to give clinician's the insight to help their patients.

Project presentation

Project poster

Prospect project poster