The FDA requires drug companies to monitor the safety of their drugs for many years after they have approval for general availability. Pharmacovigilance is the activity which tracks; in multiple geographies; the use of drugs and reports suspected adverse reactions. Our client was responsible for sifting through millions of prescriptions, hospital records and FDA forms to decide on the valid cases and prepare Case Summaries suitable to be submitted to the FDA.
Alumnus automated the workflow to reduce time and cost. The automation used NLP to ‘read’ forms (structured but non-uniform) and plain English (unstructured) text to extract information such as patient details, diseases, medications and adverse reactions. It looked up medical databases of symptoms, diseases and drugs and extracted their medical codes. Events were collated into a timeline and a textual Case Summary was generated.
The NLP-accelerated intelligent workflow made a significant increase in the efficiency of using expert hours, the most expensive human cost used in pharmacovigilance.
Technology Stack:
Java
