SQL-Based Risk Models

SQL-based Risk Models define risk scores using custom SQL logic. They are a good fit when:

• Risk calculation requires complex logic or custom data aggregations
• Risk cannot be expressed using Populations defined in Relevant
• Technical flexibility is more important than interpretability

Note: Unlike risk factor–based Risk Models, SQL-based Risk Models do not provide a breakdown of how a patient’s risk score was calculated.

How SQL-based models calculate risk

A SQL-based Risk Model consists of a single SELECT statement that returns 1 row per patient. It should select two columns:

• patient_id: this should match the primary key from the patients table in the rdm schema.
• score: a numeric value representing the patient’s risk. There are no constraints on how you calculate the score