The majority of cancer patients are diagnosed in primary care, and the majority of their interactions are with general practitioners. Ensuring primary care is optimised from both a health-outcomes and patient-experience perspective is the objective of numerous researchers in this field. This is being done by integrating statisticians, digi-trials, infectious disease, clinical informatics, behavioural science and sociology.
Oxford is home to over 60 research groups focusing on data science and its application to cancer research. Ranging from electronic patient record epidemiology and health-economic assessment, to computer vision, modelling and molecular informatics.
The importance of data science has grown with the exponential expansion of the digitisation in every aspect of our lives. 90% of the worlds 44 zettabytes of data has been created in the last two years, and this astonishing growth continues globally at 1.7MB per person every second. The resulting increase in volume and complexity of information, and the analytical technologies to extract meaning from them, represent exciting opportunities for revolutionising the scale and efficiency of cancer research.
The field of data science focusses on the analysis of this information through the application of novel statistical, computational, simulation, machine learning and artificial intelligence techniques. Oxford researchers specialise in diverse data types, including health records in patient registries, molecular phenotype, and images. This information can be collected from anything as small as individual molecules, to as large as an entire nation’s population.
Oxford’s 60 data science research groups are spread across the city with hubs at the Big Data Institute, Mathematical Institute, and Departments of Population Health and Primary Care. As cancer is a multi-faceted disease that is ever-changing, the more we know, the better we can predict and treat it. By integrating data scientists from various backgrounds into our programmes, new meaning from large clinical and molecular datasets can be extracted and our ability to understand, predict and treat the disease improved. See our Cancer Big Data theme for more information.
Images are routinely collected in the clinic (radiology, endoscopy and pathology) and laboratory (atoms, cells and tissues). Developing novel approaches to image acquisition, processing, reconstruction, analysis and interpretation are the subject of intense study in Oxford. Applying the right technique in the correct setting can be critical in predicting and understanding tumour behaviour, as well as detecting and treating cancer patients.
Drawing on skills in biology, computer science, information engineering and statistics, researchers in this field specialise in developing new tools and algorithms to interpret high-throughput-high-content genome-wide or single-cell experiments. This seeks to address how tumours evolve and react to environmental changes. As the costs of these investigations continue to fall with advancing technology, the volume of available data grows, driving the need for standardised and efficient analytical platforms. By introducing AI-based approaches, researchers are able to both guide clinical decision making and lead the discovery of fundamental molecular processes in cancer biology through the integration and statistical analysis of new and publicly available data.
Mechanistic and multiscale modelling allow us to simulate biological process and interactions (such as hypoxia, immune cell infiltration, DNA dynamics), as well as functions that vary within space and time (tumour evolution, cell signalling). In doing so, we can make predictions based on hypothetical mechanisms in the context of experimental observations, and novel hypotheses can be derived for further testing. In addition, novel methods for analysing biological data are being generated (such as topological data analysis), which can be tailored to the noisy and often incomplete datasets generated during laboratory analysis and clinical research.
Health economics is a branch of economics which addresses issues related to efficiency, effectiveness, value and behaviour in the production and consumption of health and healthcare. Ensuring that cancer interventions, such as new diagnostic tools, drugs or surgical approaches, are cost-effective and are likely to improve patient health outcomes is central to Oxford health economics research. This is undertaken by working on clinical trials and observational data with clinical and laboratory colleagues. Oxford researchers also carry out interviews and surveys with patients, clinical staff, policy makers and the general public to ascertain their views on cancer care and funding options. A particular area of interest and expertise for health economists in Oxford is the evaluation of novel genomic technologies in cancer.
Electronic health records collected as part of standard clinical care, national biobanks and targeted cohorts contain detailed information about a patient’s clinical history. Analysis of these on a large-scale enables us to identify the individuals with the highest risk of disease, or those likely to respond positively to specific treatments. This research requires a range of expertise such as the ability to ethically and securely collect, store and curate the data, link it with other datasets and analyse it using advanced statistical or machine learning methods. This needs to be combined with on-the-ground clinical knowledge, both in primary and secondary care, to direct the research questions and to evaluate how best to implement the findings for patient benefit.
Examples of the unique resources Oxford cancer researchers support include very large cohort datasets (e.g. The Million Women Study, Our Future Health), major biobanks (UK Biobank and China Kadoorie) and groups drawing together electronic health data (NCIMI, QResearch, Oxford-Royal College of General Practitioners Research & Surveillance Centre, CORECT-R, NIHR Health Informatics Collaborative, Clinical Trial Service Unit & Cancer Epidemiology Unit, the Turing Institute, Health Data Research UK and NHS Digital.)
In this theme
• Computer Vision
• Molecular Informatics
• Mathematical Biology
• Health Economics
• Primary Care
• Electronic Health Record Epidemiology & Risk Stratification