SOCR News MIDAS Biomedical Bootcamp 2021

SOCR News & Events: 2021 MIDAS Data Science for Biomedical Scientists Bootcamp

The Michigan Institute for Data Science (MIDAS) is organizing a week-long Data Science for Biomedical Scientists Bootcamp. This workshop will introduce data science from a biomedical perspective. Bootcamp participants will learn about practical data science applications in biomedical and health case-studies. Modern data science, machine learning, artificial intelligence, and biostatistical methods will be integrated into the training curriculum.

Instructors

Kayvan Najarian

Nambi Nallasamy

Ivo Dinov, University of Michigan, SOCR, MIDAS.

Michael Mathis

Ryan Stidham

Jonathan Gryak

Michael Sjoding

Workshop Logistics

• Dates/Times: Monday through Friday, July 26-30, 2021, 7:00-16:00 US-EDT (daily).
• Registration: Registration Link.
• URL: MIDAS Bootcamp Website.
• Session Format: Two daily sessions (4-hours each).
• Session URL.

Overview

Target Audience: This workshop is open to all biomedical scientists. The curriculum is geared towards junior faculty members who plan to incorporate data science in their scholarly work.

Prerequisite: College level math and statistics.

Main components:

Math and algorithmic foundations for data science

Key concepts of data science

Introduction to Python programming

Machine learning, support vector machine, artificial neural network, deep learning

Example of biomedical research projects with data science

Incorporating data science in biomedical grant proposals

Program Schedule

Day	Time	Instructor	Session Topic	Content
Monday	7:00 - 8:30 AM	Kayvan Najarian	Session 1: Welcome and introduction to the program	A review of the program and logistics
				Why data science, artificial intelligence, and machine learning?
	8:30 - 8:45 AM	Break
	8:45 - 10:15 AM	Ivo Dinov	Session 2: Math foundations I – Brief introduction to mathematical foundations of machine learning	Math notation and fundamentals
				Linear Algebra and Matrix Computing
				Optimization theory
				Differential Equations
				Calculus of Differentiation & Integration
	10:15 - 10:30 AM	Break
	10:30 AM - 12:00 PM	Ivo Dinov	Session 3: Math foundations II – Brief introduction to mathematical foundations of machine learning	Dimensionality
				Principal Component Analysis (PCA)
				High-dimensional Visualization (hands-on demos)
	12:00 - 1:00 PM	Lunch Break
	1:00 - 2:30 PM	Kayvan Najarian	Session 4: Clustering vs Classification; k-means; k-Nearest Neighbors	Supervised & Unsupervised methods
				k-means/Spectral/Hierarchical clustering (unsupervised)
				k-NN (supervised), Naïve Bayes classification
	2:30 - 2:45 PM	Break
	2:45 - 4:15 PM	TBA	Session 5: Introduction to Python programming	Basics of Python programming
Tuesday	7:00 - 8:30 AM	Ivo Dinov	Session 6: Linear regression, logistic regression	Simple linear regression, logit modeling
				Ordinary least squares estimation
				Example scenarios
				Controlled feature selection (knockoff)
	8:30 - 8:45 AM	Break
	8:45 - 10:15 AM	Kayvan Najarian	Session 7: Simple classification methods and feature analysis	Naïve Bayes classification, Feature selection and reduction
	10:15 - 10:30 AM	Break
	10:30 AM - 12:00 PM	Kayvan Najarian	Session 8: Model validation and assessment	Metrics for assessment of model performance, n-fold cross validation
	12:00 - 1:00 PM	Lunch Break
	1:00 - 2:30 PM	Michael Mathis	Session 9: Using machine learning for clinical and health applications I
	2:30 - 2:45 PM	Break
	2:45 - 4:15 PM	TBA	Session 10: Python programming for linear regression, logistic regression; ridge regression and Naïve Bayes	Python for applying simple machine learning methods to a clinical decision-making problem
Wednesday	7:00 - 8:30 AM	Kayvan Najarian	Session 11: Artificial neural networks I	Fundamentals of artificial neural networks and their advantages/limitations
	8:30 - 8:45 AM	Break
	8:45 - 10:15 AM	Kayvan Najarian	Session 12: Regression trees	Classification and regression tree (CART)
	10:15 - 10:30 AM	Break
	10:30 AM - 12:00 PM	Kayvan Najarian	Session 13: Random Forest	Ensemble use of regression trees for random forest and other boosting methods
	12:00 - 1:00 PM	Lunch Break
	1:00 - 2:30 PM	Ryan Stidham	Session 14: Using machine learning for clinical and health applications II
	2:30 - 2:45 PM	Break
	2:45 - 4:15 PM	TBA	Session 15: Python programming for neural networks, regression trees and random forest	Python for applying CART, random forest, and neural networks to a clinical decision-making problem
Thursday	7:00 - 8:30 AM	Kayvan Najarian	Session 16: Support vector machines	Using Kernel methods for support vector machines (SVM)
	8:30 - 8:45 AM	Break
	8:45 - 10:15 AM	Jonathan Gryak	Session 17: Deep Learning I	Deep Learning overview, appropriate uses of deep learning, convolutional neural networks, U-Net
	10:15 - 10:30 AM	Break
	10:45 AM - 12:00 PM	Jonathan Gryak	Session 18: Deep Learning II	LSTM, Autoencoders
	12:00 - 1:00 PM	Lunch Break
	1:00 - 2:30 PM	TBA	Session 19: Python programming for support vector machine	Python for applying SVM to a clinical decision-making problem
	2:30 - 2:45 PM	Break
	2:45-4:15 PM	TBA	Session 20: Python programming for deep learning	Python for applying deep learning models to a clinical decision-making problem
	7:00-8:30 AM
Friday		Kayvan Najarian	Session 21: Strategies to add a data science flavor to health-related projects and grant proposals	Some general tips on how to integrate data Scientific ideas in primarily clinical/biomedical grant proposals
	8:30-8:45 AM	Break
	8:45 - 10:15 AM	Michael Sjoding	Session 22: Using machine learning for clinical and health applications III
	10:15-10:30 AM	Break
	10:30 AM - 12:00 PM	Nambi Nallasamy	Session 23: Using machine learning for clinical and health applications IV
	12:00-1:00 PM	Lunch Break
	1:00 - 2:30 PM	Michael Mathis	Session 24: Guidelines on using machine learning for clinical applications
	2:30 - 2:45 PM	Break
	2:45 - 4:15 PM	Ivo Dinov, Jonathan Gryak, Michael Mathis, Kayvan Najarian Nambi Nallasamy, and Michael Sjoding	Session 25: Wrap-up	Q&A; plans for follow-up sessions during the coming year

Capstone Project

Interactive-learning (open-ended) project using a large Autism data tensor (n=1,098; k=2,145). Use the RMD source, the example HTML output, and the provided data to experiment with some of the DSPA techniques. Think of ways to augment these data (e.g., expand the time range and increase the feature richness).

Additional Resources

• Sessions 2, 3, 6 Roadmap/Summary.
• DSPA Wikipedia.
• DSPA Appendices: Bayesian Simulation, Modeling and Inference » Information-Theoretic Foundation of Statistical Learning » Surface, Shape, and Manifold Representation and Visualization » Power Analysis in Experimental Design » Database SQL/NoSQL Queries & Google BigQuery » Image Convolution, Filtering, & Fourier Transform » Causality, Transfer Entropy, & Mechanistic Effects » Agent-based Reinforcement Learning.
• DSPA Springer Page & SpringerLink (PDF Download).
• dspa.predictive.space & DSPA MOOC Canvas Site.

---

---

---

Translate this page:

(default)	Deutsch	Español	Français	Italiano	Português	日本語	България	الامارات العربية المتحدة	Suomi	इस भाषा में	Norge
한국어	中文	繁体中文	Русский	Nederlands	Ελληνικά	Hrvatska	Česká republika	Danmark	Polska	România	Sverige