Text and Data Mining Meets the Pharmaceutical Industry: Markus Bundschus Speaks

Abstract

EDITOR'S SUMMARY

Text and data mining have proven to greatly impact the world of biomedical research, especially for Roche Diagnostics in Penzberg, Germany. Taking information from such sources as patient literature, genomic cancer samples and PubMed articles, researchers at Roche Diagnostics are able to structure the data in a way that lends itself to creating personalized healthcare. Text mining used to build structured databases tends to yield the most relevant information for biomedical research, so Roche uses unstructured data to build a knowledge base automatically. This knowledge base, the disease marker association database, offers search capabilities for full text, abstracts or curated data. The database is made up of 50-million scientific abstracts and leans on rule-based engines as well as machine learning engines. By combining information from patient care, diagnoses and treatment, the healthcare industry can see a shift to digitization and more efficient care.