Intro to Using EHR/EMR Data and Using AI to Support Clinical Decision-Making

Resources we’ll use today

1. Google slides
2. colab notebook in python. Here is an overview on how to use colab features.

What we’ll cover

Topics

1. Intro to EHR/EMR data
   • What is in EHR/EMR data?
   • What is it useful for?
   • What does it not have - Limitations?
   • Common issues and tips for using it
2. Examples - Supporting Clinical Decision-making using AI/ML
   • Early event prediction - code blue
   • Early screenings for diagnosis - diabetes
   • Disease progression - kidney disease
   • Patient retention in care
3. Building AI Systems to Support Clinical Decision-Making
   • Scoping (Scoping Guide
   • Data
   • Analytical Formulation
   • Baselines
   • Model Development
   • Model Selection
   • Bias and Fairness
   • Explainability/Interpretability
   • User Interfaces
   • Validation and Testing
   • Deployment
   • Monitoring
4. Hands-on exercises exploring EMR Data using Python and SQL
   • Exploring EHR data
   • Answering descriptive questions using EHR data

Programming Languages

• Python with packages including pandas, sklearn, matplotlib, seaborn
• SQL

Tools

• Google colab for python and sql coding and visualization

Presentation Slides

• Google slides
• Key Takeaways

Colab Notebook for Class and Homework Assignment

• colab notebook in python

Sample Data Used

We’ll be using a sample of 10000 patients from MIMIC III

• Data documentation
• Schema Browser

Code to load, browse, and explore the data

• Code repository to load and explore the data
• github repo for more sample code
• colab notebook exploring the mimic data that you can use code from

Additional projects using mimic data

• MIMIC Extract