Skip to main page content
U.S. flag

An official website of the United States government

Dot gov

The .gov means it’s official.
Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re on a federal government site.

Https

The site is secure.
The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
. 2025 Feb 6;12(2):61.
doi: 10.3390/jcdd12020061.

Clinical Decision Support for Patient Cases with Asymptomatic Carotid Artery Stenosis Using AI Models and Electronic Medical Records

Mackenzie Madison  1 , Xiao Luo  1   2 , Jackson Silvey  2 , Robert Brenner  1 , Kartik Gannamaneni  1 , Alan P Sawchuk  1
Affiliations

Clinical Decision Support for Patient Cases with Asymptomatic Carotid Artery Stenosis Using AI Models and Electronic Medical Records

Mackenzie Madison et al. J Cardiovasc Dev Dis. .

Abstract

An artificial intelligence (AI) analysis of electronic medical records (EMRs) was performed to analyze the differences between patients with carotid stenosis who developed symptomatic disease and those who remained asymptomatic. The EMRs of 872 patients who underwent a carotid endarterectomy between 2009 and 2022 were analyzed with AI. This included 408 patients who had carotid intervention for symptomatic carotid disease and 464 patients for asymptomatic, >70% stenosis. By analyzing the EMRs, the Support Vector Machine achieved the highest sensitivity at 0.626 for predicting which of these patients would go on to develop a stroke or TIA. Random Forest had the highest specificity at 0.906. The risk for stroke in patients with carotid stenosis was a balance between optimum medical treatment and the underlying disease processes. Risk factors for developing symptomatic carotid disease included elevated glucose, chronic kidney disease, hyperlipidemia, and current or recent smoking, while protective factors included cardiovascular agents, antihypertensives, and beta blockers. An AI review of EMRs can help determine which patients with carotid stenosis are more likely to develop a stroke to assist with decision making as to whether to proceed with intervention or to demonstrate and encourage reduced stroke risk with risk factor modification.

Keywords: TIA; artificial intelligence; carotid stenosis; predicting stroke; stroke.

PubMed Disclaimer

Conflict of interest statement

The authors declare no conflicts of interest.

Figures

Figure 1
Figure 1
Model comparison based on AUROC curve. The dashed line shows the results of predictions based on random chance.
Figure 2
Figure 2
SHAP interpretation of Random Forest. Red indicates a factor making a stroke more likely, while blue indicates a factor making a stroke less likely.
Figure 3
Figure 3
SHAP waterfall plots for Case 1—73-year-old white male. Red factors make a stroke more likely while blue factors make a stroke less likely.
Figure 4
Figure 4
SHAP waterfall plots for Case 2—77-year-old white male. Red factors make a stroke more likely while blue factors make a stroke less likely.
Figure 5
Figure 5
SHAP waterfall plots for Case 3—65-year-old white female. Red factors make a stroke more likely while blue factors make a stroke less likely.
See this image and copyright information in PMC

Similar articles

See all similar articles

References

    1. AbuRahma A. An analysis of the recommendations of the 2022 Society for Vascular Surgery clinical practice guidelines for patients with asymptomatic carotid stenosis. J. Vasc. Surg. 2024;79:1235–1239. doi: 10.1016/j.jvs.2023.12.041. - DOI - PubMed
    1. Naylor A.R. Why is the management of asymptomatic carotid disease so controversial? Surgeon. 2015;13:34–43. doi: 10.1016/j.surge.2014.08.004. - DOI - PubMed
    1. Paraskevas K.I., Mikhailidis D.P., Veith F.J. Are Symptomatic Patients Appropriate Candidates for Carotid Artery Stenting? No (at Least Not at Present) Vascular. 2010;18:185–188. doi: 10.2310/6670.2010.00027. - DOI - PubMed
    1. Luo X., Ding H., Broyles A., Warden S.J., Moorthi R.N., Imel E.A. Using machine learning to detect sarcopenia from electronic health records. Digit. Health. 2023;9 doi: 10.1177/20552076231197098. - DOI - PMC - PubMed
    1. Teoh D. Towards stroke prediction using electronic health records. BMC Med. Inform. Decis. Mak. 2018;18:127. doi: 10.1186/s12911-018-0702-y. - DOI - PMC - PubMed

LinkOut - more resources