MRC DiMeN Doctoral Training Partnership: Integrated system for rigorous personalised diagnosis for Giant Cell Arteritis

Supervisors:

(1) Prof Scott Ferson is the director of the Institute for Risk and Uncertainty of the University of Liverpool.
https://www.liverpool.ac.uk/engineering/staff/scott-ferson/

(2) Prof Ann Morgan is professor of Molecular Rheumatology at the University of Leeds, and Honorary Consultant Rheumatologist at Leeds Teaching Hospitals NHS Trust.
https://lida.leeds.ac.uk/target/ann-morgan/

(3) Dr Marco De Angelis is PDRA at the Institute for Risk and Uncertainty.
https://www.liverpool.ac.uk/risk-and-uncertainty/staff/marcodeangelis/

(4) Dr Harry Tsoumpas is a Royal Society Industry Fellow & Lecturer in Medical Imaging
https://medhealth.leeds.ac.uk/profile/500/1054/charalampos_tsoumpas

(5) Mr is Louis Clearkin an ophthalmologist with special training and expertise in medical retinal disorders, particularly ocular inflammatory diseases, wet AMD (age-related macular degeneration), retinal vein occlusion and diabetic retinopathy. More about Mr Clearkin can be found on his website: http://www.louisclearkin.com.
 
Project description:
Diagnostic uncertainty is a feature of giant cell arteritis (GCA); which can be challenging to diagnose because of its nonspecific clinical features. Overlooking GCA risks irreversible visual loss. When GCA is suspected, patients are treated presumptively with high dose corticosteroid prior to biopsy result. Biopsy turns out to be negative in 65-85%, so most patients are incorrectly suspected of suffering from GCA, and exposed to the harms of adverse drug effects. It still remains unclear where the balance of benefit -v- harm lies as a result of real-life implementation of this standard practice.
Diagnosis can be considered fundamental in medicine as prognosis and treatment depend on it’s accuracy. Misdiagnosis causes a 30% increase in morbidity and mortality. Misdiagnosis related USA healthcare costs were $340 billion in 2014. NHS paid £197.2 million compensation for misdiagnosis in 2015.

In this project we propose a computer-based Bayesian clinical decision support system (CDSS) to improve diagnostic accuracy and evidence-based care, and to save costs while improving patient safety. Statistical information about diagnostic tests is often inaccessible to patients. Even when available, it may not be correctly communicated or utilised in diagnosis.
The tool will enable clinician to upload images as well as histology results, which will subsequently lead to diagnosis with a certain degree of certainty/confidence. The doctor will then be able to classify the patient with enhanced accuracy. As condition of use, physicians agree to report patients’ results. This creates a feedback loop so the CDSS can be alimented by practicing physicians. Such data also allows quantifying the CDSS’s improvements over traditional approaches in terms of better outcomes and reduced costs, and can detect diagnostician performance to identify better practices. This project is an example of using largely available digital technologies and rigorous methodological analysis to enhance the quality of patient’s experience and help doctors make better GCA diagnosis. The tool that we are developing will be designed to be patient centric, accessible, and inclusive of diversities. Thanks to rigorous uncertainty analysis this information can will aliment the CDSS tool towards ever more accurate GCA diagnosis. This research will look at delivering discovery from data to improve public health, and deliver fundamental insights that grow medical businesses. The project addresses the issue of ageing and disease as GCA occurs over the age of 50yrs. The toxicities from long-term glucocorticoid use have a major impact on quality of life and psychological wellbeing of elderly patients, resulting in a high personal and economic burden.

With our collaborating partner, we have already developed a successful prototype CDSS for diagnosing giant cell arteritis. The CDSS poses questions for a patient whose answers are the signs/symptoms presented. Questions can be answered imprecisely, or left unanswered. The output characterises the patient’s PPV/NPV given the presented signs and symptoms.
Several robust Bayes libraries, along with multiple classifier/confusion matrix as well as machine learning libraries, and many CDSSs have been proposed, but currently no open-source tools exist that tackle the twin problems of bad data and uncertainty communication in diagnosis

Benefits of being in the DiMeN DTP:
This project is part of the Discovery Medicine North Doctoral Training Partnership (DiMeN DTP), a diverse community of PhD students across the North of England researching the major health problems facing the world today. We are very proud of our student-centred ethos and committed to supporting you throughout your PhD. As part of the DTP, we offer bespoke training in key skills sought after in early career researchers, as well as opportunities to broaden your career horizons in a range of non-academic sectors.
Being funded by the MRC means you can access additional funding for research placements, international training opportunities or internships.
Further information on the programme can be found on our website: http://www.dimen.org.uk/