BME1560H Artificial Intelligence for Biomedical Engineering

Streams

Molecular Engineering

Sessions

Fall

Description

Biomedical engineering is the application of engineering principles to design, develop and evaluate medical systems for healthcare purposes. Majority of biomedical systems generate a huge amount of multimodal data from a variety of sensors and devices. This unique data can help develop better predictive systems, improve health outcomes and the design of such systems. The last decade has seen an unprecedented success of artificial intelligence (AI) in many applications, including biomedical applications. Data from wearable devices, ambient sensors, video cameras, and speech have been used to detect falls, diseases (e.g., dementia, sleep apnea, cancer), predict important health events (e.g., epileptic seizures), and so on. Machine learning, and more specifically deep learning approaches are at the forefront of developing predictive models that can be used in clinical settings.

This course provides an opportunity to graduate students with both the breadth and depth in the area of machine learning and deep learning applied to biomedical applications. The course firstly introduces basic machine learning algorithms, including supervised learning (e.g., logistic regression, naïve bayes, decision trees, etc.) and unsupervised learning (k-means, hierarchical clustering). It will be followed by describing other fundamental concepts, such as classifier evaluation and statistical testing to compare classifiers. The next part is the study of different deep learning models (in seminar style) for various biomedical applications that deals with multiple types of data, including and not limited to biosignals, physiological data, environmental data, speech, text, images, and videos. Different types of supervised and unsupervised frameworks for sequential and non-sequential data will be discussed, including Feed-forward neural network, Convolution neural networks, Autoencoders, Long Short-Term Memory, Temporal Convolution Network. In the last part, advanced deep learning architectures applied to biomedical application will be discussed, including Generative Adversarial Network and Contrastive Learning. The course will comprise of programming assignment (2), paper critiques and a group project (team of 1-3 persons).

Prerequisites

Knowledge of programming (e.g., python) is recommended

Components

Lecture

Restrictions

Students who have taken MIE1517 may not take this course