SHAP-Assisted Machine Learning Model to Predict Stroke Location in Atrial Fibrillation-Related Ischemia

Abstract

The anatomical distribution of ischemic stroke in patients with atrial fibrillation (AF) is highly heterogeneous, yet its biological determinants remain poorly defined. This study aimed to identify clinical and molecular predictors of stroke topography and to develop explainable machine learning models to forecast lesion location. We retrospectively analyzed 500 patients with AF-related ischemic stroke, collecting clinical data and blood-based biomarkers including C-reactive protein (CRP), interleukin-6 (IL-6), B-type natriuretic peptide (BNP), D-dimer, and fibrinogen. Patients were stratified by stroke territory (anterior vs. posterior circulation) and hemispheric side (left vs. right). Two XGBoost models were built to predict each spatial pattern. Model performance was assessed using ROC analysis, and SHapley Additive exPlanations (SHAP) were applied for feature interpretation. Posterior circulation infarcts were associated with higher CRP and IL-6 levels (p < 0.0001), while BNP was elevated in left-sided strokes, and IL-6 in right-sided strokes. The circulation model achieved an AUC of 0.71, and the hemispheric model an AUC of 0.76. SHAP highlighted CRP and IL-6 as top predictors for posterior infarcts, and BNP for left-sided infarcts. These findings suggest that inflammatory and cardiac biomarkers carry spatially specific predictive value in AF-related stroke and support the use of interpretable models for personalized risk stratification.

Keywords

atrial fibrillation ischemic stroke stroke topography explainable machine learning SHAP biomarkers XGBoost variable importance CRP IL-6 BNP

Get full access to this article

View all access options for this article.

References

Bivard

Levi

Krishnamurthy

, et al. (2020). Perfusion CT in acute stroke: A comprehensive analysis of infarct and penumbra. Radiology, 267(2), 543–550. https://doi.org/10.1148/radiol.12120971

Caplan

L. R.

Hennerici

(1992). Impaired emboli clearance as a link between hypoperfusion and ischemia. Archives of Neurology, 78(3), 245–252. https://doi.org/10.1001/archneur.1992.00530260016007

Chappell

F. M.

Wardlaw

J. M.

(2020). Imaging outcomes in randomized trials of stroke. International Journal of Stroke, 15(5), 482–491. https://doi.org/10.1177/1747493020937389

Chen

Liu

Zhang

(2021). Global burden of atrial fibrillation and its complications: A contemporary update. Journal of Cardiology Research, 58(2), 145–153.

Diener

H. C.

Sacco

R. L.

Yusuf

, et al. (2020). Rivaroxaban versus aspirin in secondary prevention. New England Journal of Medicine, 382(2), 213–221. https://doi.org/10.1056/NEJMoa1911425

Elkind

M. S. V.

(2020). Inflammatory mechanisms of stroke. Stroke, 51(10), e343–e348. https://doi.org/10.1161/STROKEAHA.120.030027

Hart

R. G.

Diener

H. C.

Coutts

S. B.

, et al. (2023). Embolic strokes of undetermined source. Lancet Neurology, 22(4), 320–332. https://doi.org/10.1016/S1474-4422(23)00067-4

Hindricks

Potpara

Dagres

, et al. (2020). 2020 ESC guidelines for the diagnosis and management of atrial fibrillation developed in collaboration with the European association for cardio-thoracic surgery (EACTS). European Heart Journal, 42(5), 373–498. https://doi.org/10.1093/eurheartj/ehaa612

Huang

Feng

Wang

(2020). CHA₂DS₂-VASc score and beyond: Predicting ischemic stroke in atrial fibrillation. Thrombosis and Haemostasis, 120(5), 759–765. https://doi.org/10.1055/s-0040-1713086

10.

Inoue

Ohara

Shimizu

, et al. (2022). Left atrial appendage morphology and thromboembolic risk. Circulation Journal, 86(1), 123–130. https://doi.org/10.1253/circj.CJ-21-0626

11.

Kim

Park

(2020). Pathophysiological insights into cardioembolic strokes: New evidence and therapeutic targets. Stroke Insights, 31(4), 221–229.

12.

Lee

J. H.

Moon

Choi

(2021). Anatomical patterns of atrial fibrillation-related stroke: Lateral and posterior dominance. Neurovascular Journal, 44(3), 198–205.

13.

(2020). Inflammatory markers in stroke: Roles of IL-6, CRP, and D-dimer. Brain & Blood Vessels, 35(6), 314–320.

14.

Liu

Wang

Yuan

(2023). Explainable machine learning in stroke prediction using SHAP and clinical biomarkers. AI in Healthcare, 17(1), 1–12. https://doi.org/10.1016/j.aiih.2023.100187

15.

Lundberg

S. M.

Erion

Chen

, et al. (2020). From local explanations to global understanding with explainable AI for trees. Nature Machine Intelligence, 2, 56–67.

16.

Lundberg

S. M.

Lee

S. I.

(2020). A unified approach to interpreting model predictions. Advances in Neural Information Processing Systems, 33, 4765–4774.

17.

Montaner

Delgado

Perea-Gainza

, et al. (2021). BNP and cardioembolic stroke subtypes. Stroke, 52(9), 3056–3063. https://doi.org/10.1161/STROKEAHA.121.033086

18.

Nguyen

Patel

Singh

(2021). Machine learning in cerebrovascular prediction: Models, data, and impact. Journal of Neural Computation, 29(1), 10–20.

19.

Ouyang

Zhang

(2023). Spatial variability of embolic stroke in AF: A radiological review. Frontiers in Neurology, 14, 1102456. https://doi.org/10.3389/fneur.2023.1102456

20.

Rao

Chen

(2022). Gradient boosting algorithms in medical AI: Strengths, weaknesses, and practical tips. Computational Medicine Reports, 3(2), 45–56.

21.

Ribeiro

M. T.

Singh

Guestrin

(2020). “Why should I trust you?”: Explaining predictions of any classifier. KDD Proceedings, 1135–1144. https://doi.org/10.1145/2939672.2939778

22.

Singh

Mahajan

Verma

(2020). Stroke topography and clinical outcome: The missing link. Journal of Stroke and Cerebrovascular Diseases, 29(8), 104875. https://doi.org/10.1016/j.jstrokecerebrovasdis.2020.104875

23.

Tan

Zhang

Wang

(2023). Biomarker-guided risk assessment in atrial fibrillation patients with stroke. Clinical Biomarkers, 41(2), 122–129.

24.

Vila

Castillo

Dávalos

Chamorro

(2000). Proinflammatory cytokines and early neurological worsening in ischemic stroke. Stroke, 51(4), 1075–1080. https://doi.org/10.1161/01.STR.31.10.2325

25.

Wang

Zhao

(2021). Prognostic utility of inflammatory markers in acute ischemic stroke with AF. Journal of Translational Stroke, 16(1), 25–33.

26.

Wintermark

Albers

G. W.

Broderick

J. P.

, et al. (2020). Imaging recommendations for acute stroke and transient ischemic attack patients: A consensus statement from the AHA/ASA. Stroke, 51(7), e224–e243. https://doi.org/10.1161/STR.0000000000000210

27.

Yaghi

Bernell

Lanzino

, et al. (2021). Clinical and imaging features of cardioembolic stroke. Journal of Stroke and Cerebrovascular Diseases, 30(4), 995–1002. https://doi.org/10.1016/j.jstrokecerebrovasdis.2020.105685

28.

Zhang

Liu

(2021). Posterior circulation strokes: Clinical challenges and solutions. Stroke Research and Treatment, 2021, 8891217. https://doi.org/10.1155/2021/8891217

29.

Zhao

Chen

(2023). Spatial prediction of embolic stroke using biomarker-integrated machine learning. Cerebrovascular Science, 18(3), 144–151.

30.

Zhou

Lin

Tang

(2022). Cardioembolism in atrial fibrillation: New imaging perspectives. Neuroimaging Clinics, 32(2), 199–210. https://doi.org/10.1016/j.nic.2022.01.002