Adverse Drug Reaction Detection

Automatically detect Adverse Drug Reactions or Events (ADR / ADE) from multichannel unstructured data, notes, transcripts and literature​
# 1
Peer-reviewed State-of-the-art Accuracy
Schedule a Call

Auto-detect and Extract Key Facts of Adverse Drug Reactions at Scale

Live Demo
Python Notebook
01
Collect
  • Multichannel unstructured data
  • Transcription of calls pharmacist, doctors, patients, and medical affairs​
  • CRM notes, customer support notes​
  • Clinical notes from EMR’s or PDF’s​
  • Social media posts
  • Biomedical literature​
02
Filter
  • Search
  • Pre-processing
03
Classify & Extract
  • Text Classification: does this text describe an adverse event?
  • Entity Recognition: Identity and normalize the drugs and symptoms
  • Relation Extraction: Which symptoms are related to which drugs?

We have established a new state-of-the-art accuracy:

  • ADR and Drug entity extraction
  • Relation Extraction (RE) models when enriched with a supplementary dataset
  • Text classification model, for deciding if a conversation includes an ADR
download csv
Read the paper

Available as a software or fully managed solution

Data
Collection
we build the data integration pipelines
  • batch or streaming data
  • structured, unstructured, or image/PDF files
  • data quality testing
Information
Extraction
we build the NLP pipelines
  • text classification
  • entity recognition
  • relation extraction
Ongoing
monitoring & Turing
we ensure uptime, performance and accuracy
  • model tuning
  • model retraining over time
  • regular service and software upgrades
Schedule a Call