How can bringing together different types of health data help patients and clinicians?

When treating patients, clinicians rely on specific clinical tests like biopsies or scans, the latest scientific knowledge and their experience to inform their diagnosis and treatment decisions. However, the wealth and breadth of information about patients is increasing with data stemming from medical records, scans, genetic tests as well as all the latest scientific literature, we need to find new ways to integrate all this big data into a form that is useful for clinicians.

In the iASiS project, we have developed a novel knowledge-driven framework that uses knowledge graphs and machine learning to understand and predict patterns within these diverse datasets. Moreover, we are utilising advances in artificial intelligence (AI) to not only understand and predict patterns between these diverse data sets, but also to provide a context and rationale around each of the discoveries and elements within the knowledge graph. Collectively, this approach is giving us a more holistic characterisation of the lung cancer and Alzheimer’s disease patient populations.

Our goal is to identify patient characteristics that may enable more accurate diagnosis, indicate the best treatment approaches for the individual and pick-up side effects at an earlier stage. The technical challenge has been to devise the most suitable machine-learning methods for identifying the important characteristics that can predict these diagnostic and treatment outcomes.

Currently, iASiS project partners are identifying new insights on a variety of aspects from drug-drug interactions, disease progression and survival time as highlighted in the figure below.

To find out more about the development of the iASiS knowledge graph and the wider potential of AI for medicine, please click here to read an interview with our partner, Professor Maria-Esther Vidal.