Mission

The KCL Centre for Doctoral Training in Data-Driven Health will train the next generation of PhD health data scientists within an active NHS environment with the skills they need to develop new models of data-driven care, leveraging significant recent investment and infrastructure in Health Data Research within the UK.

Scientific Themes

Learning from Big Data for Health

This theme will focus on the use of large, distributed, heterogeneous data sources such as EHRs, patient registries with linked genomic data, and person-produced content (e.g. from smartphone/wearables and social network activities), to address major public health challenges, such as reducing inequalities in life expectancy. Projects will be supported by partners including the Office of the London Mayor, King’s College Hospital (KCH), Harvard and 100k Genomics England. We will research privacy-preserving distributed software engineering approaches that avoid the need to share patient-level data for analysis and ensure full auditability of the analytics task.

Knowledge Representation for Clinical Decision Support

This theme will focus on the translation of data into actionable knowledge, through research into explainable reasoning techniques, common data models, agent-based simulations of knowledge processing environments, data visualisation techniques and frameworks for publishing machine learning algorithms and auditable models into learning health systems. The concept of trust in decision support will also be explored, as prioritised by our partners (Imosphere, NHS Digital), and the use of sensor technologies in smart homes to improve independent living.

Informatics for Next-Generation Clinical Trials

Randomised control trials (RCTs) represent the mainstay of medicine, but are expensive, typically address narrowly defined populations and suffer from a long pathway to translation. This theme will focus on accelerating the pace and reducing the cost of clinical trials through the use of real world data, such as that collected to support clinical care, statistical methods for optimising trial design, workflow modelling for intelligent adaptive trials and use of computable phenotypes and natural language processing for cohort definition. This will involve the use and development of statistical methods for causal inference, and efficacy and mechanisms evaluation as well as the use of wearables for digital transformation of trial endpoints. The partnerships with Janssen, Harvard, Cornell, and Michigan, among others, will help students develop skills to address issues such as the variability across data sets and drift across time in patient definitions.

Translating Informatics Research into Practice

This theme will provide training into the social, legal and ethical challenges of producing innovations that can be deployed into a highly-regulated and information-sensitive domain such as health. These include data governance and security, data provenance to achieve reproducibility, privacy and trust, transparency and explainable AI, human-computer interaction and usability research, data integration and interoperability, ethical issues around data-driven technical innovations and formal validations of black-box machine learning approaches. Imosphere Ltd. will support PhD projects researching democratisation of analytics through improved access, understanding and ability to evaluate and use health information by the general population.