Certificate in Clinical Informatics and Data Science: 3 Digital Badges

Branimir Ljubic, Ph.D., M.D. – Dr. Ljubic will be the primary instructor for the core technical elements (computer science methods, machine learning) of the program.  He is widely published on the use of machine learning and other computer methods in clinical applications.  He holds a Ph.D. in Computer and Information Science from Temple University and an M.D. from Medical School, University of Belgrade. He is the Senior Scientist for Clinical, Medical Informatics and Advanced Research Computing in the Rutgers Office of Advanced Research Computing (OARC). The OARC is the university’s centralized resource for research computing, including machine learning and clinical informatics. As part of the Office of Information Technology (OIT), OARC provides Rutgers researchers with essential computing, networking, storage, and data-handling capabilities, and offers students valuable exposure, training, and education in advanced research computing (HPC computing), clinical and medical informatics, machine learning methods, data analysis, and programming languages. In addition to these core services, OARC manages critical resources like the Clinical Research Data Warehouse (CRDW), REDCap for secure data collection and management, and the Research Data Storage (RDS) system, ensuring comprehensive support for the university’s research community.

Noëlle Foster, Ph.D. – Dr. Foster specializes in the collection and preparation of secondary data for use in research, and will teach the compliance and data preparation sections.  She also serves as the overall badge coordinator and administrator, and is your first point of contact for general questions.  She is the Principal Clinical Informatics Analyst for the RWJMS Psychiatry Department.  She holds an MS in Information Technology (Database Systems) and a Ph.D. in Science and Technology Studies, both from Rensselaer Polytechnic Institute.

This is the first of three digital badges that comprise the Certificate in Clinical Informatics and Data Science.  Participants will be assessed on core technical skills, data science workflow, and compliance and other legal considerations in which clinical data is collected and used.

Modules:

• Introduction and Getting Started
• Compliance Considerations and Secondary Data Use
• Basics of Data Cleanup and Preparation
• Clinical Informatics and Advanced Research Computing
• Introduction to Machine Learning
• Linear Models: Linear Regression, SVM, Logistic Regression
• Traditional Learning Models: k-Nearest Neighbor
• Traditional Learning Models: Decision Trees
• Ensemble Methods: Random Forest
• Traditional Learning Models: Perceptron

In this second badge, we will be dealing with more sophisticated complex methods for knowledge discovery. 

Modules:

• Deep Machine Learning Methods: Artificial Neural Networks
• Deep Machine Learning Methods: Convolutional Neural Networks
• Deep Machine Learning Methods: Recurrent Neural Networks
• Deep Machine Learning Methods: Generative Adversarial Networks
• Data Mining Introduction
• Data Mining Methods
• Unsupervised Learning: Clustering

This badge will focus on specialized methods for unstructured data.  We will also touch on emerging social considerations of machine learning such as the use of artificial intelligence in clinical sites and its potential to ameliorate or reinforce social disadvantages.  This badge will include the final project. 

Modules:

• Probabilistic Models: Bayes
• Social Networks: Basic Theory
• Social Networks: Random and Power Law Networks
• Natural Language Processing
• Emerging considerations in clinical data
• Image Processing
• Final Project

For the final project, each student will be expected to design and execute a machine learning project on a subject of their choosing.  The student will be responsible for sourcing the data, including any compliance and IRB considerations.  In the event the student wishes to use data from an existing project, it will be the student’s responsibility to add the named program staff to their IRB protocol to allow data sharing in advance of project start. 

There are open de-identified data sets that can be used for this project.

The student will turn in two products for successful project completion:

• Project proposal (3 pages):
  1. A description of the proposed project, including what is the question you want this project to answer
  2. High-level description of the selected dataset
  3. The method selected and a justification for why this is the best solution to answer your proposed question
  4. What, if any transformations will be made to the data to facilitate use in the chosen method
  5. A work plan, including estimated level of effort for each task. This can be in the form of a GANTT or PERT chart.
• A final presentation (10 minutes) and written report (3 pages)
  1. Presentation to be recorded and uploaded to Canvas and should be written as a discussion of the results and the answer to your question, as if presenting to a supervisor. It should include any associated tables or visualizations in the presentation slides.
  2. Three page report to include a discussion of the process itself, along with any unexpected problems along the way, how realistic your work plan ended up being, and what are your lessons learned for the next project

It is expected that each student will work on their own individual project.  In the event that we have members of the same team participating on the same time who wish to collaborate on an existing project, that may be possible provided each student owns different pieces of the project and can justify which parts are whose. 

Projects will be graded pass/fail.