This course will focus on the use of logistic regression for clinical risk predictions. Students will use the statistical package R to create risk models using an existing dataset of non-diabetes patients undergoing hemoglobin A1c testing (Wells, 2018). The dataset was extracted from Wake Forest EHR and is in a format ready for analyses with minimal cleaning. Students will need to have access to a laptop to each session to be used to conduct the analyses with R.
Although simple R skills will be taught in the class, it would be helpful if students have some familiarity with R before the course. There are a variety of free introductory R courses available online including Learn R Programming.
Thursdays starting June 2, 2022 through August 25, 2022, 7:30 - 9:00 pm
Example discussion topics include:
- Overview of medical decision making and potential uses for risk models
- Using EHR data for research
- Creation and validation of risk models, using logistic regression
- Deciding when it may be appropriate to use newer machine learning methods (e.g. random forest or deep learning)
- Variable selection
- Missing data
- Clinical implementation of prediction tools
Contact Laya Mohan firstname.lastname@example.org with any questions!