Prevention & lipidsCohort study
Flag, Identify, Network, Deliver (FIND) Lp(a) Machine Learning Model
- Light Gradient-Boosting Machine machine learning model (MLM) was trained (90% of data) and tested (10% of data)
- deidentified medical records
- Family Heart Database
Authors' conclusion
One-half of adults with ASCVD flagged by FIND Lp(a) MLM had confirmed elevated Lp(a) ≥125 nmol/L. A 2.2-fold screening enrichment was observed (≥125 nmol/L model); enrichment was greater with ≥150 and ≥200 nmol/L submodels. Early performance indicators support the central role of FIND Lp(a) MLM within the ongoing FIND Lp(a) program.
Study context
- Population
- Adults with ASCVD in the Family Heart Databaseadults with ASCVD, test data set n = 34,499 · test data set 34,499
- Participants
- 344,987 adults
- Exposure
- Percentage of adults with confirmed Lp ≥125 nmol/L were compared
- test data setn = 34,499
- cohort flagged by modeln = 1,553
Main finding
As reported in the abstract
Confirmed Lp(a) ≥125 nmol/L
- screening enrichment due to the model
- 2.2-fold
Observational data: associations, not effects.
Further findings
Screening enrichment, submodels
- ≥150 nmol/L submodel
- 2.3 fold
- ≥200 nmol/L submodel
- 2.7 fold
Feature group importance (Tree Shapley Additive exPlanations)
- Medication use was the most important feature group followed by diagnoses and lipid labs
Original publication
FIND Lp(a) Machine Learning Model. JACC Adv. Published 11 August 2026. doi:10.1016/j.jacadv.2026.103114
Summarised from the published abstract; every number is checked against it