Southern University of Science and Technology
This course aims to provide students with a systematic understanding of the fundamental concepts, major tasks, and evolving paradigms of artificial intelligence and machine learning, with a particular emphasis on their innovative applications in modern medicine and public health. The course begins with a brief introduction to classical models, such as linear and logistic regression, discussing their underlying principles and limitations, and then focuses on the latest advances in deep learning and generative AI for medical applications. Students will learn the theoretical foundations of deep learning and explore core architectures including neural networks, convolutional neural networks (CNNs), and Transformers, understanding their use in medical image analysis, clinical text understanding, disease risk prediction, and decision support. The course will emphasize large language models (LLMs), foundation models, multimodal analysis techniques, and generative AI in healthcare, exploring their mechanisms and innovative potential in tasks such as automated clinical trial matching, pathology image analysis, cardiovascular risk prediction, automated report generation, and medical question answering. In addition, the course will introduce reinforcement learning for intelligent decision-making and adaptive treatment strategies, as well as transfer learning for handling small-sample medical data. It will also address critical issues such as model interpretability, fairness, and privacy protection. Through a combination of theoretical instruction and hands-on programming practice (using PyTorch, scikit-learn, etc.), students will develop the ability to understand and apply state-of-the-art AI techniques to real-world medical challenges and critically analyze and innovate at the intersection of AI and medicine.
Prerequisite: None
Semesters Taught: Spring 2026.
Course Information: Syllabus
This course focuses on multimodal medical data analysis and artificial intelligence, providing a systematic introduction to machine learning methods—particularly deep learning and generative AI—and their applications in medical research. Guided by real-world medical problems, the course integrates practical data sources such as electronic health records (EHRs) and populationbased cohort data, and presents a unified analytical framework covering data processing, feature representation, model development, and result interpretation for decision support. The course is organized into three modules: (1) Foundations (Sections 1–5): Introduces the background and motivation of medical AI, key characteristics of medical problem formulation, and fundamental methods in traditional machine learning and deep learning. (2) Single-modality Data Analysis (Sections 6–10): Covers structured data, time-series data, clinical text, and medical imaging, with emphasis on data standardization, feature engineering, representation learning, and typical modeling approaches. (3) Multimodal Modeling (Sections 11–15): Focuses on multimodal data fusion strategies and representative models, with discussions of key challenges such as missing modalities and distribution shifts in real-world settings.
Prerequisite: None
Semesters: Fall 2026.