Observational Health Data Sciences and Informatics

This is an old revision of the document!

---

ATLAS has embedded the ability to generate prediction models using machine learning methods for precision medicine and disease interception, including:

• Regularized regression
• Random forest
• k-nearest neighbors

The ATLAS prediction feature uses the R package PatientLevelPrediction that builds patient level predictive models using data in Common Data Model format.

Features:

• Takes a cohort and outcome of interest as input.
• Extracts the necessary data from a database in OMOP Common Data Model format.
• Uses a large set of covariates including for example all drugs, diagnoses, procedures, as well as age, comorbidity indexes, etc.
• Large scale regularized regression to fit the predictive models.
• Includes function for evaluating predictive models.
• Supported outcome models are logistic, Poisson, and survival (time to event).

This feature can be accessed by clicking on the Prediction menu item; there are options for inputs in the ATLAS prediction page. Additional details can be found by viewing a video tutorial: https://youtu.be/BEukCbT8UoA. And/or review of the github repository: https://github.com/OHDSI/PatientLevelPrediction