Page History
...
See algorithm outline for more details on the algorithm.
...
| Table of Contents |
|---|
...
|
...
|
| Children Display |
|---|
Notes:
- Relies on i2b2 data using the ENACT ontology.
- If your data uses custom code prefixes (instead of ICD10CM: and ICD9CM:), replace the prefix in the code pattern column in DT_LOYALTY_CHARLSON table.
How to
...
Use the Loyalty Cohort Tool
- Create a cohort filter, defining the patients on which to compute loyalty scores. The three columns are:
- patient_num: patient_num from the i2b2 tables
- cohort_name: a name for the cohort. You can optionally compute several cohorts separately, but specifying different values for this.
- index_dt: a date which is a reference point in time at which to compute the loyalty score. It is suggested to select a common recent point in time or to choose each patient's most recent visit date, for example.
- Run the USP_DT_LOYALTYCOHORT stored procedure with the following parameters.
...
2. Execute USP_DT_LOYALTYCOHORT stored procedure
BEGIN USP_DT_LOYALTYCOHORT('TEST', 1, 0, 0, 0); END;
This will create two tables on your db, DT_LOYALTY_RESULT (line level data with variables and score presented for each patient) and DT_LOYALTY_RESULT_SUMMARY (summary table).
Loyalty Cohort Database Tables
DT_LOYALTY_PATHS: This table captures the specific concept paths related to loyalty features, with site-specific codes and optional comments. It is used to map ontology elements to binary variables for computing loyalty scores.
DT_LOYALTY_CHARLSON: This table records Charlson Comorbidity Index categories, their respective weights, and code patterns to evaluate patient comorbidities. It supports the computation of Charlson scores as part of the loyalty evaluation.
DT_LOYALTY_PSCOEFF: Contains field names and corresponding coefficients used for calculating predictive scores, essential for the regression equation used in loyalty score computation. It can be customized with locally-retrained weights (a mechanism for this has been developed but is not yet integrated into the Digital Twin package).
DT_LOYALTY_RESULT_SUMMARY: Summarizes cohort data across various health metrics and tests, including gender-specific denominators, cutoff filters, and detailed descriptions of the cohort's health outcomes and probabilities. It provides an overview to validate cohort characteristics.
DT_LOYALTY_RESULT: Detailed patient-level loyalty score data, capturing demographics, health screenings, and predictive scores for various health outcomes, including recent updates for tracking death dates. Critically, it includes the computed loyalty score.
DT_LOYALTY_RESULT_CHARLSON: Similar to DT_LOYALTY_RESULT but focused on Charlson Comorbidity Index scores and detailed comorbidity categories for patients, including a variety of specific health conditions. It includes the calculated Charlson scores and 10-year survival probabilities.
