BME 487: AI in BME
Description
The goal of this course is to introduce and apply AI tools to problems in BME.
AI and machine learning algorithms have had a major impact on biomedical science in the past decade. AI algorithms can learn patterns from biomedical data sets to provide actionable insights on disease diagnosis or treatment. This course will focus on practical applications of AI in BME with hands-on tutorials.
This course will provide an overview of a wide range of AI and machine-learning tools (clustering, regression, decision trees, random forests and neural networks), biomedical data sets (imaging, omics and data-mining) and diseases (cancer, infectious-, cardiovascular- and neurological-).
Software used - MATLAB/Python/R. Basic familiarity with MATLAB or Python programming is required. However, no extensive programming experience is necessary beyond those covered in ENG 101. Matlab, R and Python code and documentation will be provided for all lectures.
Source: XKCD
Who is this for?
This course is intended for BME undergraduates and graduate students interested in learning AI. EECS and other students interested in biomedical applications of AI can also register if they satisfy the Prerequisites.
The course counts towards biochemical, biomechanical and bioelectrical undergraduate concentrations, and the Bioelectrics and Neural Engineering, Biotechnology, and Systems Biology graduate concentrations.
This course is for 3 credit hours and will be on Mon/Wed from 1:30-3:00.
Students need to bring their own laptops with MATLAB installed. MATLAB is available for free from CAEN.
Prerequisites
Biology 172 or 174, Intro to Biology*
Math 116, Calculus*
Engineering 101, Intro to Computing*
Statistics & Linear algebra (preferable, but not required)
or Graduate standing
* - Can be waived with permission from the instructor (csriram@umich.edu)
Course outline
Introduction to AI & Machine-learning (ML)
Matlab programming boot camp
Exploratory data analysis
Plotting and data visualization
Handling big-data sets
Hypothesis testing
Unsupervised learning and clustering
Classifying cancer sub-types using clustering
Regression (Linear, Logistic, Lasso, Stepwise, PLS)
Predicting cardiovascular risk using regression
Model validation
Decision trees
Predicting diabetic retinopathy using decision trees
Random Forests (RF)
Identifying synergistic drug combinations using RF
Neural Networks (NN)/ Deep Learning
Diagnosing breast cancer from biopsy images using NN & Transfer Learning
Interpreting AI
Guest lectures on applications of AI in medical imaging, neural engineering, systems biology, microbiome and data mining.
Guest lectures (2020, 2021): Kelly Arnold, Jeff Fessler, Cindy Chestek (UM)
Guest lectures (2022): Cindy Chestek (UM), Lana Garmire (UM), Jaeyun Sung (Mayo Clinic), Arjun Krishnan (MSU)
Grading
Assignments (50%). There will be five assignments. The assignments will involve both programming and biomedical modules.
Design project (50%). The final project will be done as a team of students. Each team is required to give a short presentation on their project along with a written report. Both the presentation and the report should provide a background of the biomedical problem, a description of the AI methods used, effective visualization of the results, and a discussion on possible limitations and future directions of the project.
Past projects by students include:
Analysis of COVID-19 transmission in South Korea
Diagnosing Breast Cancer using imaging
Classifying cell types from Spatial Transcriptomics datasets
Diagnosis of Parkinson’s using voice measurements
Interrogating drivers of vaginal microbiome stability and dysbiosis
Analysis of lung tumor single cell RNA sequencing data
Urinary Tract Infection prediction from health record data
Predicting survival time of Hepatocellular Carcinoma patients using clinical markers
Additional reading
Review articles on AI and healthcare
Machine Learning in Medicine | NEJM
Artificial intelligence in healthcare | Nature Biomedical Engineering
A guide to machine learning for biologists | Nature Reviews Mol. Cellular Biology
Avoiding common pitfalls in machine learning omic data science |Nature Materials
Transforming healthcare with AI: The impact on the workforce and organizations | McKinsey
Latest news, tutorials, and articles on AI
Text books & courses
An Introduction to Statistical Learning (using R) by Gareth James, Daniela Witten, Trevor Hastie & Robert Tibshirani (free ebook)
MATLAB Machine Learning by Stephanie Thomas & Michael Paluszek.
MATLAB for Machine Learning by Giuseppe Ciaburro
Deep Learning by Ian Goodfellow, Yoshua Bengio and Aaron Courville (free ebook)
Data Science training for Biomedical Scientists through MIDAS UMich
Meta-resource on MATLAB Machine Learning
Practicing Machine Learning and Coding