Webinar: How can NLP solve data challenges in healthcare?

The Topol Review cited Natural Language Processing (NLP)-based AI as one of the top five technological advances set to impact the provision of services delivered by the NHS.

Earlier this month we attended a presentation which explored how NLP can solve data challenges in healthcare, by examining tried and tested ways in which it is already being deployed today, and the exciting new use-cases in which it could be equally well applied tomorrow.

The session was hosted by a team from Inspirata, a specialist provider of cancer informatics, artificial intelligence and data structuring services. With a survey sent out in advance to all those who registered, 73% of respondents said there were not using NLP tools today. The aim of the meeting was to demystify NLP by spotlighting how healthcare professionals could utilise artificial intelligence to automate many of the manual data tasks they undertake each day, in order that they might expend greater effort on higher-value analysis activities.

Inspirata presented their technology: “More than 70% of healthcare data is locked in clinical documents, reports, patient charts, clinician notes and discharge letters.  Our NLP-engine offers countless opportunities to dive into the vast amount of clinical and associated data that is currently inaccessible within documents and reports of various types, and leverage it to improve outcomes, optimise costs, and deliver a better quality of care.”

The team presented cancer reporting and data abstraction use-cases in which NLP is being used to help identify instances of reportable cancer, automatically surface patients eligible for clinical trials and formulate registry datasets.

The session also explored how the Inspirata team have applied their engine to the COVID-19 Open Research Dataset (CORD-19) as a way of further illustrating the power of NLP.  The CORD-19 archive of over 180,000 articles presented the opportunity to extract concepts, and process the information quickly and at scale, to understand the terms and connections between articles. For example, a user could explore the relationship between COVID-19 and a particular protein or drug, and then quickly in turn, explore any connections between research papers that would help inform decision making.

Watch a recording of the session here: