Related Experiment Video
Updated: Jun 15, 2025
12:18
A Machine Learning Approach to Design an Efficient Selective Screening of Mild Cognitive Impairment
Published on: January 11, 2020
7.5K
Development and multinational validation of an algorithmic strategy for high Lp(a) screening
Arya Aminorroaya1, Lovedeep S Dhingra1, Evangelos K Oikonomou1
1Section of Cardiovascular Medicine, Department of Internal Medicine, Yale School of Medicine, New Haven, CT, USA.
Nature Cardiovascular Research
|August 28, 2024
Summary
Elevated lipoprotein (a) (Lp(a)) increases cardiovascular risk, but testing is rare. A new machine learning model, ARISE, effectively identifies individuals with high Lp(a) levels, improving screening efficiency.
Area of Science:
- Cardiovascular Medicine
- Artificial Intelligence
- Genetics and Genomics
Background:
- Elevated lipoprotein (a) (Lp(a)) is a significant, heritable risk factor for premature atherosclerotic cardiovascular disease.
- Current Lp(a) testing rates are low (<0.5%), hindering the clinical application of emerging targeted therapies.
- There is a critical need for improved screening strategies to identify individuals at risk due to elevated Lp(a).
Purpose of the Study:
- To develop and validate a machine learning model for targeted screening of elevated Lp(a) (≥150 nmol/L).
- To assess the model's ability to reduce the number of individuals needing testing to identify those with elevated Lp(a).
- To evaluate the model's performance across diverse, large-scale cohort studies.
Main Methods:
- Development of a machine learning model, ARISE (Algorithmic Risk Inspection for Screening Elevated Lp(a)), using data from the UK Biobank (N=456,815).
- External validation of the ARISE model in three independent cohort studies: ARIC (N=14,484), CARDIA (N=4,124), and MESA (N=4,672).
- Analysis of model performance based on reduction in the number needed to test (NNT) to identify individuals with elevated Lp(a).
Main Results:
- The ARISE model demonstrated consistent performance across internal and external validation cohorts.
- ARISE reduced the number needed to test for elevated Lp(a) by up to 67.3%, depending on the probability threshold used.
- The model effectively utilizes commonly available clinical features for screening purposes.
Conclusions:
- The ARISE machine learning model offers a promising tool for optimizing screening for elevated Lp(a).
- Deployment of ARISE in electronic health records could significantly enhance the yield of Lp(a) testing in real-world clinical settings.
- Improved screening facilitated by ARISE may facilitate earlier intervention and better management of cardiovascular risk associated with elevated Lp(a).
