Model selection in Artificial Intelligence for Medical Microbiology: workshop presentation and decision cheat sheet

This repository contains the presentation slides and a complementary decision cheat sheet developed for the ESCMID Artificial Intelligence and Machine Learning in Medical Microbiology Diagnostics Workshop 2025, held in Zurich, Switzerland. The materials were created and delivered by Alejandro Guerrero-López, PhD, Visiting Professor of Machine Learning at Universidad Carlos III de Madrid. The presentation slides provide a comprehensive, step-by-step guide designed specifically for medical microbiologists and infectious disease specialists who aim to integrate AI and machine learning techniques into their diagnostic workflows. The slides cover critical concepts including: How to frame a clear microbiological research question to guide AI model selection Understanding different types of microbiological data (genomic sequences, images, mass spectrometry, tabular clinical data) Evaluating key data characteristics such as sample size, dimensionality, and label availability A practical decision tree for choosing appropriate machine learning models based on data properties and research goals Overview of classical machine learning and deep learning model categories, with guidance on when and how to apply them Emphasis on interpretability and explainability methods relevant to clinical microbiology The decision cheat sheet (PDF) complements the slides by providing a concise, easy-to-follow workflow that researchers and clinicians can use when planning AI projects or evaluating model options: How to analyze your data and task requirements before selecting a model Clear pathways to decide between classical ML and deep learning approaches Practical advice on model evaluation, interpretation, and the importance of baseline comparisons