Machine Learning and Advanced Data Methods

IIDS67682 Machine Learning and Advanced Data Methods 15 Semester 2 Postgraduate Taught Level 7 Magnus Rattray Unit coordinator Mudassar Iqbal Unit coordinator Sokratia Georgaka 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 Biomedical and health informatics is founded on the usage and application of computational algorithms to high-throughput biological data and patient-level information. With the growing availability of big data in the biomedical domain, machine learning and other advanced approaches are becoming essential. Informaticians need to have a good understanding of these different algorithms, methodologies and analysis pipelines, and their applicability to different data types and research questions.  This unit will cover the following indicative content:<ol> <li>Big data in biomedicine and health – what is big data?, The four Vs of big data, the challenges with working with big data, existing open resources (UK BioBank, MIMIC III, and similar)</li> <li>Unsupervised learning: principal component analysis, non-linear dimensionality reduction (tSNE, UMAP) and clustering (k-means, hierarchical clustering).</li> <li>Supervised machine learning (decision trees, ensembles, logistic regression) including principles of method evaluation and model selection </li> <li>Data in the wild: data wrangling, batch effects, missingness, dataset shift and concept drift</li> <li>Advanced methods (neural networks and deep learning).</li> <li>Ethical aspects of machine learning (algorithmic fairness and explainability)</li></ol> Each topic will cover a theoretical background including introduction to several relevant algorithms, as well as a hands-on labs to gain experience applying methods to data. This unit aims to:<ul> <li class="CxSpFirst">Introduce different types of biomedical and health data, including from high-throughput biological experiments and patient-level information resources</li> <li class="CxSpMiddle">Examine different approaches and pipelines for data analysis</li> <li class="CxSpMiddle">Learn the principles underlying popular machine learning algorithms</li> <li class="CxSpMiddle">Gain experience in applying machine learning to real-world datasets</li> <li class="CxSpMiddle">Critically appraise analysis pipelines and data usage and be able to make suggestions on improvements</li></ul> <ul> <li class="CxSpFirst">Introduce different types of biomedical and health data, including from high-throughput biological experiments and patient-level information resources</li> <li class="CxSpMiddle">Examine different approaches and pipelines for data analysis</li> <li class="CxSpMiddle">Learn the principles underlying popular machine learning algorithms</li> <li class="CxSpMiddle">Gain experience in applying machine learning to real-world datasets</li> <li class="CxSpMiddle">Critically appraise analysis pipelines and data usage and be able to make suggestions on improvements</li></ul> <ul> <li>Demonstrate a critical understanding of advanced machine learning methods and techniques</li> <li>Apply these methods to a range of problems</li> <li>Evaluate different scenarios in which such methods would be applicable</li> <li>Assess the possible ethical implications of Machine Learning methods when applied to biomedical data sets</li></ul> <ul> <li>Critically evaluate the strengths and limitations of different data analysis methods and approaches</li> <li>Apply analytical methods to a range of datasets and research questions</li> <li>Demonstrate an understanding and draw conclusions from the results of analytical methods</li></ul> <ul> <li>Produce and execute a data analysis pipeline using python</li> <li>Carry out exploratory data analysis using visualisation and clustering methods</li> <li>Apply and evaluate the performance of different machine learning algorithms</li></ul> <ul> <li>Improved coding skills and experience with Jupyter notebooks</li> <li>Be able to apply data analysis skills to a broad range of problems</li> <li>Use appropriate software and tools for data analysis in biomedical research</li></ul> This unit will be delivered in a face-to-face format over six days: lectures and open discussions will provide basic and core knowledge and introduce concrete examples and encourage attendees to draw upon their own reading and experience. Practical sessions will provide hands-on experience, with real-world application of the theoretical material. Coding will be carried out using python via web-based jupyter notebooks that will be available to work on throughout the course. All materials will be introduced on Canva. 0 Other 100% 3 x interactive online assessments Three online assessments carried out through interactive notebooks The three assessments will cover:1. Exploratory data analysis2. Supervised machine learning3. Advanced topics (neural networks and ethics) . MSc Health Data Science PT MSc Health Data Science PGDT Taught Component Optional PG Dip Health Data Science PT PG Diploma Health Data Science PGDT Taught Component Optional PG Cert Health Data Science PT PGCert Health Data Sci (18) PGDT Taught Component Optional PG Cert Health Data Science PT PGCert Health Data Sci (12) PGDT Taught Component Optional Health Data Science - CPD Health Data Science - CPD PGDT Taught Component Optional MSc Health Data Science FT MSc Health Data Science FT PGDT Taught Component Optional PG Dip Health Data Science FT PG Dip Health Data Science FT PGDT Taught Component Optional N <ul> <li>Elements of Statistical Learning (Hastie, Tibshirani & Friedman, Springer, 2009, second edition)</li> <li>Machine Learning in Healthcare Informatics (Dua et al., Springer, 2014)</li> <li>Machine Learning and AI for Healthcare: Big Data for Improved Health Outcomes (Panesar, Apress, 2019)</li></ul> Practical classes & workshops 25 Tutorials 25 0 100