Design and Analysis to Estimate Diagnostic Test Accuracy Without Universal Verification

This PhD project at the University of Bristol, within the Bristol Medical School, offers an exciting opportunity to advance statistical methods for estimating diagnostic test accuracy in clinical settings. As the number of available disease tests grows, understanding their accuracy—especially when no perfect 'gold standard' exists—becomes increasingly important for public health and clinical practice. The project focuses on developing and applying Bayesian modelling techniques to address challenges in diagnostic test accuracy estimation, particularly when universal verification is not feasible. Traditional approaches rely on comparing test results to a gold standard, but in many cases, such a standard is unavailable or impractical. Latent class models have been proposed to jointly estimate test accuracy and disease prevalence, but these require large datasets and multiple tests, which are not always possible. This project will explore alternative study designs, including selective and conditional testing pathways, and investigate how external information (such as prior evidence on test accuracy) can improve robustness and reduce data requirements. The research will also assess the value of routinely collected real-world test result data to supplement information from diagnostic accuracy studies. Working in collaboration with the UK Health Security Agency (UKHSA), the project will apply these methods to infections of public health importance, such as Lyme disease, where no true gold standard exists and conditional testing is common. The Bayesian statistical framework, using software like JAGS or Stan, will allow for flexible modelling of complex data structures and synthesis of evidence across studies. Simulation studies and value of information approaches will be used to optimize future study designs based on current evidence. The Bristol Medical School is a leading centre for research and teaching in Population Health Sciences and Translational Health Sciences, with a collaborative and multidisciplinary environment. The studentship is fully funded for three years by the Health Protection Research Unit in Evaluation and Behavioural Science (HPRU EBS), covering tuition fees for home students and a stipend at the UKRI rate (£20,780 in 2025/2026). Overseas students are welcome to apply but must pay the difference between home and overseas fees. Applicants should have, or expect to obtain, at least a 2.1 degree (or equivalent) in statistics, medical statistics, mathematics, or a similar highly quantitative field. A relevant Master's degree or research experience is advantageous, and some experience with Bayesian statistical modelling is desirable but not essential. The programme welcomes applicants from diverse backgrounds and those who have experienced challenges or disadvantages. To apply, candidates must select the Population Health PhD programme and enter the supervisor names as listed under the project title. Please state 'Health Protection Research Unit in Evaluation and Behavioural Science' in the funding box. Refer to the Admissions Statement for full application details and provide a personal statement outlining your training, motivation, and suitability for the programme. The application deadline is 16:00 GMT on Monday, March 16, 2026, with interviews scheduled for May 2026 and an anticipated start date in September 2026. For further information, contact Prof Hayley Jones at [email protected].