Universitat Internacional de Catalunya
Uciel Chorostecki
Adrià Fernández
Robert Subirana
It introduces students to the fundamental concepts and practical applications of artificial intelligence in biomedical research and healthcare.
Sustainable Development Goals (SDGs): The course on Artificial Intelligence in Biomedicine contributes to the Sustainable Development Goals (SDGs) of the 2030 Agenda, particularly SDGs 3, 9, 10, 12, and 17. It does so by promoting health and well-being, fostering biomedical innovation and infrastructure, supporting social equity through access to scientific data, ensuring responsible consumption of computational resources, and encouraging scientific and social progress through global partnerships in data sharing.
Successful completion of the Introduction to Bioinformatics and Biomolecules Interaction courses
Understand the principles of generative AI and its biomedical applications
Comprehend machine learning and deep learning concepts
Evaluate machine learning models and pipelines
Analyze AI applications in genomics, single-cell analysis, and drug discovery
Discuss ethical implications, interpretability, and governance of AI in medicine
At the end of the course, students will be able to apply and critically evaluate generative AI and machine learning models to solve biomedical challenges, while navigating their ethical and practical implications.
Starting with ChatGPT, Claude, and DeepSeek
Generative AI concepts
Large Language Models including tokens, prompts, and training data
Biomedical use cases like summarization, literature review, and coding assistance
Hands-on exploration interacting with AI models
Critical use covering hallucinations, bias, and responsible prompting
Discussion on generative AI support for biomedical learning and research
Demystifying AI concepts before code
Artificial Intelligence definition
Machine Learning and Deep Learning definitions and differences
Patterns, inference, and data representation
Supervised, unsupervised, and reinforcement learning
Machine learning pipeline from data to evaluation
Biomedical case examples like cancer risk scoring and disease classification
Learning without the math using intuition and tools
Simplified data representation including features, matrices, vectors, and embeddings (4 hours)
Simplified core supervised algorithms including logistic regression, decision trees, random forests, and SVMs
Basic neural network architecture and learning dynamics
Model evaluation covering accuracy, sensitivity, specificity, and AUC (4 hours)
AI in genomics, multi-omics, and single cell
Hit identification via AI chemoinformatics, structural biology, and docking
Lead optimization predicting affinity, toxicity, ADME, and patient stratification
Healthcare foundational models, imaging, wearables, lab robots, and the case of Recursion
AI explainability through linear coefficients, feature importance, Shap, and GNN
AI geopolitics, sustainability, ethics, governance, and open source
Hands-on AI debate regarding patients in hospitals
38 hours combining Lectures (CM) and Case Methods (MC), integrating theoretical presentations with hands-on exercises.
2 hours for the midterm exam
20% Case method sessions
30% Midterm exam
40% Final exam
10% Subjective criteria based on involvement and participation
If the exam consists of multiple-choice questions (test type), according to the degree regulations, for every +1 point for a correct answer, points will be deducted for incorrect answers as follows:
-0.33 points if there are 4 response options (with one correct).
-0.25 points if there are 5 response options (with one correct).
75% Final exam
25% Maintained case method and subjective grade
Minimum grade of 5 on the final exam
General average of 5 or higher
75% minimum attendance
Inappropriate use of electronic devices may result in expulsion
Biomedical Informatics, 4th Edition. Springer (2014). ISBN: 978-1-4471-4473-1