Introduction to Machine Learning for Healthcare

Overview

This course integrates computing, medicine, epidemiology, and health informatics to study the life cycle of artificial intelligence solutions. It covers clinical problem definition, data quality and interoperability, modeling, validation, interpretation, clinical evaluation, implementation, monitoring, and governance, with critical discussion of bias, equity, privacy, safety, and regulation.

Objective

Enable students to understand, develop, critically evaluate, and communicate machine learning solutions for healthcare problems, considering technical performance, clinical validity, risks, implementation feasibility, ethical and regulatory aspects, and impact on patient care.

Syllabus

1. Foundations of AI and machine learning in healthcare: historical evolution; prediction, diagnosis, prognosis, triage, recommendation, and clinical decision-making.
2. Clinical data, quality, and interoperability: EHRs, images, signals, clinical notes, omics data, social determinants, missingness, leakage, governance, HL7 FHIR, OMOP, LOINC, and SNOMED CT.
3. Study design and model life cycle: population, outcome, time windows, partitioning, validation, calibration, ROC/PR curves, decision curves, and subgroups.
4. Supervised and unsupervised learning: classification, regression, trees, ensembles, SVMs, neural networks, clustering, dimensionality reduction, and phenotyping.
5. Deep learning for biomedical data: convolutional networks and transformers for images, signals, and time series; transfer learning and evaluation by patient and institution.
6. Biomedical NLP, language models, and multimodal models: information extraction, embeddings, transformers, summarization, RAG, hallucination, safety, privacy, and responsible use.
7. Survival, causal inference, and real-world evidence: competing risks, censoring, target trials, confounding, DAGs, propensity scores, and limitations of observational data.
8. Clinical decision support and implementation: workflow integration, usability, alert fatigue, human–AI interaction, patient safety, and prospective evaluation.
9. Explainability, equity, privacy, and security: SHAP/LIME, saliency, algorithmic bias, Brazil’s LGPD, anonymization, robustness, and clinically useful explanations.
10. Regulation, scientific reporting, and monitoring: SaMD, ANVISA RDC 657/2022, FDA, TRIPOD+AI, CONSORT-AI, SPIRIT-AI, DECIDE-AI, drift, MLOps, and continuous governance.

Methodology

Classes integrated with seminars, analysis of real-world case studies, and development of a practical mini-project using datasets such as MIMIC-IV, PhysioNet, public images, or synthetic data when ethical restrictions apply.

Assessment

Assessment considers participation in classes and discussions, practical assignments and reports, as well as the final project and its presentation.

References

• SHORTLIFFE, Edward H.; CIMINO, James J. (eds.). Biomedical Informatics: Computer Applications in Health Care and Biomedicine. 5. ed. Cham: Springer, 2021.
• BENSON, Tim; GRIEVE, Grahame. Principles of Health Interoperability: FHIR, HL7 and SNOMED CT. 4. ed. Cham: Springer, 2021.
• JAMES, Gareth; WITTEN, Daniela; HASTIE, Trevor; TIBSHIRANI, Robert; TAYLOR, Jonathan. An Introduction to Statistical Learning: with Applications in Python. Cham: Springer, 2023.
• RAJKOMAR, Alvin; DEAN, Jeffrey; KOHANE, Isaac. Machine Learning in Medicine. New England Journal of Medicine, v. 380, n. 14, p. 1347-1358, 2019.
• HERNÁN, Miguel A.; ROBINS, James M. Causal Inference: What If. Boca Raton: Chapman & Hall/CRC, 2020.
• COLLINS, Gary S. et al. TRIPOD+AI statement: updated guidance for reporting clinical prediction models that use regression or machine learning methods. BMJ, v. 385, e078378, 2024.
• WORLD HEALTH ORGANIZATION. Ethics and governance of artificial intelligence for health: WHO guidance. Geneva: WHO, 2021.
• AGÊNCIA NACIONAL DE VIGILÂNCIA SANITÁRIA. Resolução da Diretoria Colegiada — RDC nº 657, de 24 de março de 2022. Brasília: ANVISA, 2022.
• AGÊNCIA NACIONAL DE VIGILÂNCIA SANITÁRIA. Software como Dispositivo Médico: Perguntas e Respostas — RDC 657/2022. Brasília: ANVISA, 2022.
• FOOD AND DRUG ADMINISTRATION. Marketing Submission Recommendations for a Predetermined Change Control Plan for Artificial Intelligence-Enabled Device Software Functions. Silver Spring: FDA, 2025.
• LIU, Xiaoxuan et al. Reporting guidelines for clinical trial reports for interventions involving artificial intelligence: the CONSORT-AI extension. Nature Medicine, v. 26, p. 1364-1374, 2020.
• CRUZ RIVERA, Samantha et al. Guidelines for clinical trial protocols for interventions involving artificial intelligence: the SPIRIT-AI extension. Nature Medicine, v. 26, p. 1351-1363, 2020.
• VASEY, Baptiste et al. Reporting guideline for the early-stage clinical evaluation of decision support systems driven by artificial intelligence: DECIDE-AI. Nature Medicine, v. 28, p. 924-933, 2022.
• WORLD HEALTH ORGANIZATION. Ethics and governance of artificial intelligence for health: guidance on large multi-modal models. Geneva: WHO, 2024.
• TOPOL, Eric J. Deep Medicine: How Artificial Intelligence Can Make Healthcare Human Again. New York: Basic Books, 2019.
• WANG, Ping; LI, Yan; REDDY, Chandan K. Machine learning for survival analysis: a survey. ACM Computing Surveys, v. 51, n. 6, p. 1-36, 2019.
• BEAM, Andrew L.; KOHANE, Isaac S. Big Data and Machine Learning in Health Care. JAMA, v. 319, n. 13, p. 1317-1318, 2018.
• AGRESTI, Alan. An Introduction to Categorical Data Analysis. 3. ed. Hoboken: Wiley, 2019.
• PANESAR, Arjun. Machine Learning and AI for Healthcare: Big Data for Improved Health Outcomes. Berkeley: Apress, 2019.
• BERNER, Eta S. Clinical Decision Support Systems: Theory and Practice. 2. ed. New York: Springer, 2007.