Tutorials
T1 – Evaluation of Prediction Models in Medicine
Abu-Hanna
The reliable prediction of outcomes from disease and treatment is becoming increasingly important in the delivery and organization of health care. The learning objective of this tutorial is to understand the elements underlying predictive performance and to show how to quantitatively assess the performance of prediction models. In particular, I address different categories of performance measures (including calibration, sharpness, resolution, and discrimination) and valid methods (including bootstrapping and cross validation) for obtaining performance assessments. I will also address the important difference between prediction and causal models, and how prediction models can still play a part in causality. The focus of the tutorial is on conceptual frameworks. Attention will be paid to the various choices in the design of model evaluation procedures, and the relationship between model evaluation and the purpose for which a model has been built. All methods are illustrated with real-world examples.
T2 – Explainable AI: How to create deployable AI in Healthcare
Postill, Rosella
To be updated
T3 – A Tutorial on Causal Models: Estimating the Cardiotoxic Effect of Oncological Treatments in Young Breast Cancer Survivors
Zanga, Bernasconi
This tutorial provides participants with the tools to design and implement a causal network for analyzing healthcare data. A causal network is a probabilistic graphical model that allows to explain the interactions between observed variables. The first step is to construct a causal graph, a directed graph in which a node represents a random variable and an edge represents a cause-effect pair. Participants will learn how to take advantage of clinical experts’ knowledge to derive the cause-effect relationships, shifting from an associational analysis to a causal approach. Then, causal discovery algorithms are applied to the elicited causal graph to learn new relationships. After validating the obtained graph, the parameters of the causal network are estimated from data. Participants will explore different techniques to take care of missing values and low sample size. Finally, the obtained model is used to estimate the causal effects of administered treatments on patients’ outcomes, highlighting the differences across the proposed solutions by comparing biased and unbiased estimates.
T4 – Curating Precision Cohorts: From Long COVID to Unexplained Chronic Conditions
Jonas Hügel, Arianna Dagliati, Spiros Denaxas, Ulrich Sax, Shawn Murphy, Hossein
Post-acute sequelae of COVID-19 (PASC), also called Long-COVID, remains a medical mystery. Many patients experience persistent symptoms such as fatigue, brain fog, and shortness of breath long after recovering from COVID-19, yet these conditions often go unrecognized due to their subtle and variable nature. The need for a more precise, data-driven approach to identifying PASC has never been more urgent. Azhir & Hügel et al. developed an algorithm to identify unexplained, patient-specific conditions after a COVID-19 infection. During the tutorial, attendees will gain hands-on experience using synthetic data to test and apply the algorithm. Additionally, we will explore the algorithm’s expanded capabilities in identifying other unexplained chronic conditions.
T5 – Training Sequential Deep Learning and Machine Learning Model– Hands-on Tutorial using N3C Synthetic Data
Laila Rasmy, Ziqian Xie, Degui Zhi
to be updated.