Supporting digital healthcare services using semantic web technologies

Abstract

We report on our efforts and faced challenges in using Semantic Web technologies for the purposes of supporting healthcare services provided by Babylon Health. First, we created a large medical Linked Data Graph (LDG) which integrates many publicly available (bio)medical data sources as well as several country specific ones for which we had to build custom RDF-converters. Even for data sources already distributed in RDF format a conversion process had to be applied in order to unify their schemata, simplify their structure and adapt them to the Babylon data model. Another important issue in maintaining and managing the LDG was versioning and updating with new releases of data sources. After creating the LDG, various services were built on top in order to provide an abstraction layer for non-expert end-users like doctors and software engineers which need to interact with it. Finally, we report on one of the key use cases built in Babylon, namely an AI-based chatbot which can be used by users to determine if they are in need of immediate medical treatment or they can follow a conservative treatment at home. To match user text to our internal AI-models an NLP-based knowledge extraction and logic-based reasoning approach was implemented and evaluation provided with encouraging results.