The aim of the CDT is to train a new generation of responsible researchers and innovators with the expertise and knowledge to transform the pace and precision of medical diagnosis and care through the application of Artificial Intelligence.
Students complete taught modules in the first 18 months of the integrated programme, equivalent to a full-time UK MSc. These modules cover:
- The fundamentals of AI in the context of applications from the health domain;
- The wider perspective of digital systems in the health service, and ethical and legal issues associated with health data;
- Specialist topics from over 100 potentially relevant modules, according to individual background and chosen PhD research topic;
After six months, students select a topic area for their PhD research. The research is jointly supervised by AI and clinical researchers, and culminates in the presentation of a PhD dissertation at the end of the fourth year.
As part of the MSc, students complete a Masters-level project that leads into the chosen PhD research topic.
Alongside the MSc modules and PhD research, students follow an extensive programme of activities and events:
- Participation in an annual CDT Conference, CDT seminar series, and annual Turing CDT conference;
- Participation in masterclasses organised jointly with The Alan Turing Institute on topics such as shaping policy for government, and journalistic communication.
- Training on developing impact from research, responsible innovation, and generic research skills.
- Weekly journal club
There is the opportunity for a 3-month placement with one of our industry or public-sector partners.
Where necessary for your research, we will fund visits to our international university partners.
Example PhD topics:
Screening and Early Detection:
- Using the CORECT-R UK wide colorectal cancer data repository curated at Leeds to identify high-risk digital phenotypes for targeted screening;
- Training recognition of skin cancer from 5 million images held at LTHT, to create a smartphone app that can be used by patients for rapid self-screening;
- Automated analysis of the free text and structured data within electronic patient records to identify optimal patient pathways and ensure prompt diagnosis referrals;
- Identifying those at low risk of having cancer from demographic, clinical and biochemical data, and developing pathways of care that deliver a stratified response.
- Using machine learning to identify which digital phenotypes could be fed into automatic image analysis of tissue scans to better highlight and categorise suspected tumours;
- Analysing images automatically within digital pathology, digital radiology and digital photography to accelerate and improve accurate diagnosis;
- Video analysis to aid in navigating a steerable endoscope through the gastrointestinal tract and to detect possible abnormal polyps in situ;
- Real-time optimization of image acquisition to maximise diagnostic efficacy, for example, varying resolution and acquisition time across the image plane.
Therapy and Care:
- Risk-stratifying patients into those who would benefit from major surgery or those where minor interventions would be a better option, using machine learning from pathology images linked to clinical outcome and genetic data;
- Developing intelligent tools for a patient to find their own best clinical pathway, given their understanding of the risks and life trade-offs;
- Assistance in correct dressing choice based on learning from skin lesion data linked to prescription data;
- Using AI to develop individualized risk-stratified models for cancer survivor surveillance after treatment, based on supported self-management and patient-initiated follow-up.