Healthcare Analytics: Regression in R

Introduction

• Welcome to the course
• What you should know

1. Designing Your Research

• Scientific method review
• Using a cross-sectional approach
• Reviewing existing literature for ideas
• Dealing with scientific plausibility
• Selecting a linear regression hypothesis
• Selecting a logistic regression hypothesis
• Installing necessary packages

2. Preparing for Linear Regression

• Plots for checking assumptions in linear regression
• Interpreting diagnostic plots
• Categorization and transformation
• Indexes
• Quartiles
• Ranking
• Regression review
• Preparing to report results
• Challenge: Create a categorical variable for quartiles
• Solution: Create a categorical variable for quartiles

3. Beginning Linear Regression Modeling

• Choices of modeling approaches
• Overview of modeling process
• Linear regression output
• Models 1 and 2
• Model metadata

4. Final Linear Regression Modeling

• Beginning Model 3
• Making a working Model 3
• Finalizing Model 3
• Looking at the final model
• Fishing and interaction
• Other strategies for improving model fit
• Defending the final model
• Presenting the final model

5. Preparing for Logistic Regression

• Analogies to linear regression process
• Parameter estimates in logistic regression
• Odds ratio interpretation
• Basic logistic code
• Forward stepwise regression: First two rounds
• Forward stepwise regression: Round 3
• Challenge: Fit a logistic regression model
• Solution: Fit a logistic regression model

6. Developing the Logistic Regression Model

• Running Model 1
• Adding odds ratios to models
• Model metadata
• Forward stepwise: Round 2
• Forward stepwise: Round 3
• Using AIC to assess model fit
• When to compare nested models
• How to compare nested models
• Models 1 and 2 presentation
• Model 3 presentation
• Interpreting the final model

Conclusion

• Review of metadata
• Review of the process
• Next steps