Artificial intelligence (AI) improves healthcare by helping to deliver earlier diagnosis and better treatment based on accumulated learning far beyond even the most experienced clinician.
AI will help the NHS increase efficiency, improve therapies and reduce costs – all of which benefit patients, medical staff and wider society. This is currently one of the most exciting and inspiring areas of healthcare being researched and developed, with organisations such as the AI Centre for Value Based Healthcare spearheading new innovations to improve patient care and facilitate earlier diagnosis.
With a team of AI, data science, research and clinical experts based in St Thomas’ Hospital MedTech Hub, the AI Centre for Value Based Healthcare is currently working on several projects in a range of medical fields.
We’ve compiled a list of the five most exciting and innovative projects set to reshape the future of healthcare.
- Atrial Fibrillation: Led by Alistair Young in partnership with Siemens Healthineers
The most common cardiac arrhythmia, atrial fibrillation, is commonly treated with catheter ablation – a treatment with a high recurrence rate. However, identifying the geometry and volume of the left atrial can be hugely beneficial when looking at diagnosis and prognosis. This is currently estimated with multiple measurements, a limited method that can lead to inaccurate calculations.
The AI Centre for Value Based Healthcare is developing a tool alongside Siemens Healthineers that will automatically analyse the atria and ventricles using MRI images. This tool infers the 3D shape, volume and surface area of the left atrial with greater accuracy. Looking ahead, the results will also include risk prediction scores and highlight optimal sites for ablation treatment.
- Radiotherapy planning: Led by Dr Marc Modat (academic) and Dr Teresa Guerrero Urbano & Dr Tony Greener (clinical) in partnership with Mirada Medical
AI based medical image contouring allows clinicians to easily see suspected disease, but its performance greatly depends on the quality of the data provided. However, challenging access to high-quality contoured data alongside policies that lead to that data not being widely deployed within the NHS mean that we are currently not harnessing the benefits of AI in healthcare at scale.
Mirada Medical is seeking to address the above issues with its DCLExpert™, a CE marked deep learning auto-contouring solution for radiotherapy treatment planning that produces computer-generated contouring outputs set to achieve greater accuracy and improve generalised models.
Once embedded within the NHS, it will improve current contouring practices used for treatment planning in prostate radiotherapy and will reduce variability among radiologists performing assessments, ensuring patients receive more consistent and accurate treatment.
Myocardial Ischemia: Led by Dr Amedeo Chiribiri in partnership with Siemens Healthineers, Philips Healthcare, Circle cardiovascular Imaging, Medis and GE Healthcare
Heart disease is the second most common cause of death for NHS patients over 75, and coronary artery disease (CAD) was listed as the most frequent cause of these acute heart disease instances.
CAD develops when the blood vessels supplying the heart become damaged, limiting oxygen and blood supply. This can be challenging to diagnose, as test results can sometimes be conflicting, and the interpretation of diagnostic tests often relies on the availability of trained experts who are often inaccessible to more rural communities.
This project focuses specifically on the use of AI to better understand the complex relationship between medical images and patient symptoms, eventually providing personalised care that doesn’t rely on local resources or expertise.
Using the large scale historical dataset at St Thomas’ Hospital, the project will develop novel algorithms that can guide physicians in selecting right from the start the best treatment pathway.
- Prostate cancer: Led by Professor Vicky Goh in partnership with Siemens Healthineers
Current best practice guidelines for prostate cancer suggest active surveillance as the preferred monitoring strategy. This can avoid or delay the need for further treatment until absolutely necessary, with only patients whose cancer shows signs of progressing considered for radical therapies.
However, there is a need for improved and objective criteria when determining which patients should remain on active surveillance and which should switch to definitive therapy. In this project, the AI Centre for Value Based Healthcare is developing an AI-based assessment of longitudinal changes in multi-parametric magnetic resonance imaging of the prostate in patients on active surveillance.
- HD-Neuro: Led by Dr Jorge Cardoso (academic) and Professor James Teo (clinical) in partnership with the UCL Institute of Neurology
Incredibly individually complex, computational constraints have traditionally limited our understanding of neurological illnesses to generic, weakly predictive models. This leads to an inability to provide personalised treatment and delayed reactions to patient needs – rather than early anticipation of them.
By applying novel machine learning to the biological information brain imaging provides, the AI Centre for Value Based Healthcare can create complex models with high individuating power, to be deployed within existing healthcare pathways.
Care is then personalised without the burden of new investigations – improving clinical outcomes, enhancing operational efficiency and catalysing the development and evaluation of new interventions across healthcare.
Learn more about the AI Centre for Value Based Healthcare’s latest projects here.
