M6 | Machine Learning for Precision Medicine

Teaching machines to learn and draw conclusions from data is key to the development of intelligent assistants for many tasks in clinical institutions. Machine learning (ML) tools allow users to analyse enormous amounts of clinical data and elaborate precise methods to help clinicians in their practice, helping them to handle the complexity of integrating information from different data sources. The goal of this module is to improve personalised healthcare and individual outcomes by giving participants the tools for more patient-centred care decisions by improving diagnosis, treatment and monitoring of diseases, and streamlining administrative processes in hospitals.

Today, hospitals manage huge amount of clinical data that joined with ML/AI tools allow to perform real digital transformation in healthcare. Teaching machines to learn from data is the key to develop intelligent assistants for many tasks in the clinical institutions. Today, machine learning tools allow to process large amounts of clinical data and elaborate highly precise diagnostic methods to help experts to get high precision medicine and significantly better care quality.

The module will focus on the following main topics: 

• Online resources for medical information about commercial or scientific Machine Learning (ML) offerings
• Introduction to data science topics: efficient data storage, missing data, training, testing, validation data, parameter spaces, dimensionality reduction, etc.
• ML methods for clinical diagnosis: impact, potential, and limitations
• Modelling tasks like measuring, classification, detection, and prediction in clinical environment
• Probability and uncertainty of predictive ML algorithms
• Accuracy, risks, and tools in predicting diagnoses with ML
• Key elements of operational data in hospital management
• Confidence in predictive risk outcomes: accuracy and implications for safety, ethics, and privacy
• Key General Data Protection Regulation (GDPR) elements impacting on training in AI/ ML algorithms
• Applying ML for clinical data profiling
• Anomaly detection to find pathological vs. healthy data
• Regression models for predicting clinical resources
• Accuracy and confidence of predictive risk outcomes: implications on safety, ethics, and privacy
• Black box effect: transparency, explainability and uncertainty
• Deep learning and its interpretability

Petia Ivanova

Full professor at the Department of Mathematics and Computer Science, Universitat de Barcelona, Senior Researcher at Computer Vision Center (CVC), Head of “Computer Vision and Machine Learning at the University of Barcelona” consolidated research group (www.ub.edu/cvub) SGR 1742, Collaborator of the State Agency of Research (Agencia Estatal de Investigación) area TIC (INF), División de Coordinación, Evaluación y Seguimiento Científico Técnico, November 2019.

Visit LinkedIn Profile

Markus Wenzel

Markus holds a degree in Computer Science. He is a researcher with Fraunhofer MEVIS since 2005 and has practical experience with machine learning for many differenz clinical applications. His research focusses on novel machine learning methods for clinical decision support and hypothesis generation. Besides, he develops and teaches machine learning for professionals and students.

Visit LinkedIn Profile

Javier Ródenaz

Deep Learning researcher and associate professor at the University of Barcelona. Master in Artificial Intelligence from the International University of La Rioja. +2 years of experience in Artificial Intelligence solutions.