Using natural language processing to develop layperson summaries of clinical trials.

The aim of the project is to develop a Machine Learning (ML) system to summarise clinical trials technical reports into lay summaries. This is timely because in January 2022 a new EU regulation will come into force requiring the sponsors of clinical trials to post a trial lay summary onto a new EU online portal. A key goal is to make healthcare information more accessible to the general population with a view to empowering them to make more informed health-related decisions.

The main technical innovation is the application of a combination of Natural Language Processing (NLP) text summarisation approaches to clinical trial reports. There are two main approaches of NLP-based text summarisation: extraction and abstraction. In extraction, the content from the corpus is consolidated into a summary, while in abstraction, the information from the corpus is transformed into new sentences. These initial algorithms have not been developed with technical clinical trials reports in mind, hence the will investigate a hybrid approach which involves a combination of extractive and abstractive summarisation. Specifically, as part of the project delivery the work will use Transformer based NLP methods within a text summarisation context.