Project Description

Pharmacowatch: An Adverse Drug Reaction Pharmacovigilance Solution

PharmacoWatch is a PaaS solution that alerts for and predicts drug-to-drug interactions and other adverse drug reactions (ADR) of approved pharmaceutical drugs in the USA. This solution integrates public data from the FDA Adverse Event Reporting System (FAERS) with Social Media mentions (Twitter) of ADRs.

We cleaned, normalized, and integrated the two data sources and created several machine learning models for predictive analytics. Our initial machine learning models were built and tuned to predict accurately previously unknown adverse drug-to-drug reactions.

Our PaaS model supports a public API which will allow for end-users to create various healthcare applications in the pharmacovigilance space.

Key Features

  • Bring together disparate and large data sources of drug and drug ADR reports
  • Use NLP and Unstructured Data Mining and Analytics to extract clinical terms from unstructured data (Twitter)
  • Apply machine learning, data mining, heuristics to build multiple predictive and actionable models
  • Architect solution to work in Big Data Scale
  • Augmented Reality-Enabled Retail Promotions Application

Example Mini-App

My VirtualPillbox

This is a simple application that demonstrates how the PharmacoWatch API can be used to build applications to target different end-users and use-cases in the pharmacovigilance space.

Using this mobile application, patients can query a proposed over the counter (OTC) purchase of a drug against the consumer’s own prescriptions to determine if any adverse reaction is predicted between the drugs in the prescription and the drug about to be purchased.

A similar application targeting physicians is the Prescription Checker where physicians can query proposed prescription against a patient’s existing prescription to determine if any ADR has been predicted.