Evidence-based Decision Assistance in Cancer Diagnosis

Human decision-making performance can be sub-optimal and deteriorate with increasing complexity of available data. In clinical diagnosis, a huge amount of multi-modal information (structured, semi-structured, unstructured) is available (e.g. clinical narrative, imaging, genetic data, etc.). Integrating all these data is key to assist decision making in support of clinical/translational applications. Recent workshops [1, 2] have highlighted the fundamental need for high quality, evidence-based decision making within a data science/analytics framework, with particular reference to healthcare.

This project will develop a novel methodology to integrate multi-modal healthcare information to assist cancer clinicians and researchers in diagnosis and decision-making. Specifically, the methodology should support explanatory and auditable assistance for clinical description and diagnosis, and design mapping and integration between different data sources to accommodate temporal and spatial data representation (e.g. tumour change over time). Data analytic approaches will be used to identify relevant similar cases either in a local setting or externally (e.g. published case reports), and support making and justifying a decision through transparent interaction between the user and the data space.

The successful candidate should have an excellent first degree in Computer Science or related discipline (e.g. mathematics, statistics), with interests in decision making, big data analytics and machine learning. Excellent programming skills are essential. Additional experience in health informatics, text mining and semantics integration are desirable, but are not mandatory.

We have an established successful collaboration with Christie Hospital, one of Europe’s leading cancer centres and an international leader in research and development, and the Royal Manchester Children’s Hospital (RMCH), the UK’s largest single-site children’s hospital. Over the last 10 years, we have developed several components to support cancer diagnostics, including a prototype decision assistant imaging system [3, 4] and methods to extract key clinical data from EHRs (’patient journey’ [5, 6]). SCS work has developed novel approaches to consistency and traceability to improve decision quality [7]. This work has been supported by the National Institute for Health Research (NIHR), the Christie Charity and the Christie NHS Foundation Trust.

The project will be supervised by Keane and Nenadic, with clinician involvement from Christie (Dr Laasch; Consultant Radiologist and Interventional Lead) and academic/clinician involvement from RMCH/ Faculty of Medical and Human Sciences (Dr Stivaros; NIHR Clinician Scientist & Senior Lecturer, Honorary Consultant Paediatric Neurologist). Stivaros did his PhD in Medical Informatics (2009), co-supervised by Keane with input from Nenadic; Keane is formally associated with Stivaros’ NIHR 5-year Clinical Scientist Award (2011-6).