Clinical Prediction Models

IIDS67802 Clinical Prediction Models 15 Semester 2 Postgraduate Taught Level 7 Jamie Christopher Sergeant Unit coordinator David Jenkins Unit coordinator Division of Informatics, Imaging and Data Sciences 1 FHEQ level (Framework for Higher Education Qualifications) ' Masters/Integrated Masters P4 ' European Credit Transfer & Accumulation System Rating : 7.5 Healthcare generates vast quantities of data that remain under-utilised for decision-making, with analyses often focused on descriptions of healthcare delivery rather than on prediction to support clinical decisions. Clinical prediction modelling offers a framework for translating routinely collected health data into actionable risk estimates that can inform diagnosis, prognosis, and treatment planning.This unit develops the core statistical and machine-learning skills required to design, implement, and evaluate clinical prediction models. Building on the concepts introduced in the pre-requisite unit Fundamentals in Mathematics and Statistics in Health Data Science, students will learn modern modelling approaches for risk prediction, including survival analysis and handling missing data, alongside principles of model development, validation, and reporting. The unit will also address challenges posed by complex health data structures, as well as issues of bias, fairness, and interpretability in clinical models.The unit is application-driven, with case studies and practical examples drawn from clinical and health research across the University of Manchester, enabling students to develop skills directly applicable to real-world clinical prediction problems. Students will undertake their own prediction modelling and also read, appraise and discuss published research studies. Healthcare generates vast quantities of data that remain under-utilised for decision-making, with analyses often focused on descriptions of healthcare delivery rather than on prediction to support clinical decisions. Clinical prediction modelling offers a framework for translating routinely collected health data into actionable risk estimates that can inform diagnosis, prognosis, and treatment planning.This unit develops the core statistical and machine-learning skills required to design, implement, and evaluate clinical prediction models. Building on the concepts introduced in the pre-requisite unit Fundamentals in Mathematics and Statistics in Health Data Science, students will learn modern modelling approaches for risk prediction, including survival analysis and handling missing data, alongside principles of model development, validation, and reporting. The unit will also address challenges posed by complex health data structures, as well as issues of bias, fairness, and interpretability in clinical models.The unit is application-driven, with case studies and practical examples drawn from clinical and health research across the University of Manchester, enabling students to develop skills directly applicable to real-world clinical prediction problems. Students will undertake their own prediction modelling and also read, appraise and discuss published research studies. This unit aims to <ul><li>Consolidate and build on the core statistical and machine learning content in Semester 1 modules </li><li>Introduce the more advanced statistical topics of missing data, survival analysis and clinical prediction models </li><li>Develop students’ literacy in the strengths, characteristics and correct application of modelling techniques, and how to interpret results </li><li>Equip students to implement analyses in an appropriate scripting language </li><li>Enable students to engage with and critically appraise published research </li><li>Promote reasoned discussion and respectful debate</li></ul> Students should be able to: LO1: Explain and discuss modelling techniques and appraise their application in healthcareLO2: Appraise the strengths and weakness of modelling methods  Students should be able to: LO3: Critically appraise research studies LO4: Assess the effectiveness and fitness for purpose of a modelling tool or technique LO5: Apply modelling techniques/methods to healthcare data and methodological when designing studies LO6: Interpret analytical results Students should be able:  LO7: Design and write scripts to implement statistical and machine learning methods to analyse health data LO8: Access and extract relevant information from research publications LO9: Apply reporting guidelines to research studies and use critical appraisal tools Students should be able to:LO10: Effectively communicate findings in written work and presentations LO11: Develop problem solving skills LO12: Demonstrate a critical understanding of technical descriptions of statistical/mathematical analysis methods LO13 Work effectively as an individual and in a group The unit will be delivered using a blended-learning approach designed to promote both conceptual understanding and applied skills. Core knowledge, directed reading, and preparatory materials will be provided via eLearning to introduce key concepts and methods in advance of face-to-face sessions. This pre-reading will enable students to engage more actively in taught sessions and to develop confidence in working with the research literature on clinical prediction and advanced statistical modelling. Face-to-face teaching will consist of twice-weekly sessions combining lectures, interactive discussions, and supervised computer-based practical work. Lectures will be used to recap and further develop concepts introduced through pre-reading, highlight key areas of the methodological and applied research literature, and provide context for important clinical and statistical research questions. A central focus of these sessions will be discussion and debate, enabling students to explore current methodological challenges, interpret contrasting approaches in the literature, and engage with areas of ongoing debate in clinical prediction research. Supervised practical sessions will allow students to apply methods using appropriate software and programming tools, with structured exercises associated with each key concept to support learning and assess understanding. Individual and group activities, including literature-based tasks and presentations, will be scaffolded during face-to-face sessions to support the development of critical appraisal, communication, and collaborative skills. To support inclusive participation, electronic tools (such as Padlet and Mentimeter) will be used to facilitate both verbal and non-verbal engagement during discussions. Wherever possible, teaching sessions will be recorded and made available online. Ongoing academic support will be provided through scheduled tutorials, online discussion forums, email, and regular office hours with academic staff, delivered either face-to-face or online. <figure class="table"><table><thead><tr><th>Assessment</th><th>Length</th><th>Weighting</th></tr></thead><tbody><tr><td>Summative 1 × individual coursework assignment withwritten report The assignment will involve developingstatistical scripts demonstrating theanalysis of data and a written report tojustify methods and explanation of work.There is a focus on selecting appropriatemethods to solve applied problems.The assignment will also requireengagement with the research literatureand could include, for example,performing a critical appraisal of and/or arisk of bias assessment on a publishedarticle.</td><td>Approx, 2000 words</td><td>80%</td></tr><tr><td>Summative 1 × individual multiple choice exam The exam will be multiple choice, focusingon the survival and missing data elementsof the module</td><td> </td><td>20%</td></tr><tr><td>Formative Formative feedback will primarily beprovided in the face-to-face sessions.Verbal feedback will be provided as part offacilitation of discussions and debates.Feedback on student group presentationswill include elements of both tutorfeedback and peer feedback.</td><td> </td><td> </td></tr><tr><td>Formative.Formative assessment and feedback tostudents is a key feature of the on-linelearning materials for this unit. Studentswill be required to engage in a wide rangeof interactive exercises to enhance theirlearning and test their developingknowledge and skills.In addition, there will be a series ofsupervised practical hands-on exercisesthat will allow for verbal feedback</td><td> </td><td> </td></tr></tbody></table></figure> Scores and written feedback available to students after marking. During face-to-face sessions, verbally and via written comments (e.g. on Padlet). Real-time feedback, and available through regular office hours. IIDS67631 Statistics for Health Data Science Pre-Requisite Compulsory IIDS67901 Machine Learning for Health Data Science Pre-Requisite Compulsory N Kirkwood, B. R., & Sterne, J. A. (2010). Essential medical statistics. John Wiley & Sons.<a href="https://www.prognosisresearch.com" target="_blank">https://www.prognosisresearch.com</a><a href="https://www.clinicalpredictionmodels.org/homepage" target="_blank">https://www.clinicalpredictionmodels.org/homepage</a> Lectures 18 Practical classes & workshops 24 Tutorials 2 106