Predicting 10-Year and Lifetime Stroke Risk in Chinese Population

Xing, Xue-zhong; Yang, Xueli; Liu, Fangchao; Li, Jianxin; Chen, Jichun; Liu, Xiaoqing; Cao, Jie; Shen, Chong; Yu, Ling; Lu, Fanghong; Wu, Xianping; Zhao, Liancheng; Liu, Ying; Hu, Dongsheng; Lu, Xiangfeng; Gu, Dongfeng

doi:10.1161/strokeaha.119.025553

Cited by 35 publications

(44 citation statements)

References 29 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…With the acceleration of population aging, China has faced the biggest burden of stroke in recent years [4,5]. Stroke is the leading cause of death in China [6]. Predictive studies have found that the estimated lifetime stroke risk for individuals aged 25 years or more in China is expected to reach very high levels, and this will be a major concern for the prevention and management of stroke in the future [7].…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Stroke Prediction with Machine Learning Methods among Older Chinese

Fang

2020

IJERPH

View full text Add to dashboard Cite

Timely stroke diagnosis and intervention are necessary considering its high prevalence. Previous studies have mainly focused on stroke prediction with balanced data. Thus, this study aimed to develop machine learning models for predicting stroke with imbalanced data in an elderly population in China. Data were obtained from a prospective cohort that included 1131 participants (56 stroke patients and 1075 non-stroke participants) in 2012 and 2014, respectively. Data balancing techniques including random over-sampling (ROS), random under-sampling (RUS), and synthetic minority over-sampling technique (SMOTE) were used to process the imbalanced data in this study. Machine learning methods such as regularized logistic regression (RLR), support vector machine (SVM), and random forest (RF) were used to predict stroke with demographic, lifestyle, and clinical variables. Accuracy, sensitivity, specificity, and areas under the receiver operating characteristic curves (AUCs) were used for performance comparison. The top five variables for stroke prediction were selected for each machine learning method based on the SMOTE-balanced data set. The total prevalence of stroke was high in 2014 (4.95%), with men experiencing much higher prevalence than women (6.76% vs. 3.25%). The three machine learning methods performed poorly in the imbalanced data set with extremely low sensitivity (approximately 0.00) and AUC (approximately 0.50). After using data balancing techniques, the sensitivity and AUC considerably improved with moderate accuracy and specificity, and the maximum values for sensitivity and AUC reached 0.78 (95% CI, 0.73–0.83) for RF and 0.72 (95% CI, 0.71–0.73) for RLR. Using AUCs for RLR, SVM, and RF in the imbalanced data set as references, a significant improvement was observed in the AUCs of all three machine learning methods (p < 0.05) in the balanced data sets. Considering RLR in each data set as a reference, only RF in the imbalanced data set and SVM in the ROS-balanced data set were superior to RLR in terms of AUC. Sex, hypertension, and uric acid were common predictors in all three machine learning methods. Blood glucose level was included in both RLR and RF. Drinking, age and high-sensitivity C-reactive protein level, and low-density lipoprotein cholesterol level were also included in RLR, SVM, and RF, respectively. Our study suggests that machine learning methods with data balancing techniques are effective tools for stroke prediction with imbalanced data.

show abstract

Section: Introductionmentioning

confidence: 99%

“…There are several stroke prediction tools in China. However, traditional methods, such as the Cox proportional hazard model [6,8], are unable to effectively explore the complex non-linear relationships in data. Machine learning methods can learn complex structures by incorporating numerous variables with high dimensional data [9].…”

Section: Introductionmentioning

confidence: 99%

Stroke Prediction with Machine Learning Methods among Older Chinese

Fang

2020

IJERPH

View full text Add to dashboard Cite

show abstract

“…The China-PAR stroke risk prediction model is the most recent tool to identify high-risk stroke adults in China (9). The validity of this risk model among samples of different age ranges has not been studied.…”

Section: Discussionmentioning

confidence: 99%

“…We calculated the stroke risk for each BLSA participant using the published models of the R-FSRS (15) and China-PAR 10year stroke risk models (9). Supplementary Table 1 presents the predictor variables used in the R-FSRS and China-PAR models.…”

Section: Prediction Modelsmentioning

confidence: 99%

“…Therefore, applying these models to the Chinese population might not be appropriate. In 2019, Xing et al (9) established a new stroke prediction model aimed at predicting stroke risk among Chinese individuals included in the Prediction for Atherosclerotic Cardiovascular Disease Risk in China (China-PAR) project. The China-PAR stroke risk models were developed based on a middle-aged cohort (40-79 years, mean age 48.6 years) and included factors such as age and age-related diseases.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Validation of 10-Year Stroke Prediction Scores in a Community-Based Cohort of Chinese Older Adults

et al. 2020

View full text Add to dashboard Cite

A stroke prediction model based on the Prediction for Atherosclerotic Cardiovascular Disease Risk in China (China-PAR) project was developed. We compared its predictive ability with the revised Framingham Stroke Risk Score (R-FSRS) for 5-year stroke incidence in a community cohort of Chinese adults, namely the Beijing Longitudinal Study of Aging (BLSA). Calibration, discrimination, and recalibration were used to compare the predictive ability between the two prediction models. Category-less net reclassification improvement (NRI) and integrated discrimination improvement (IDI) values were also assessed. During a mean follow-up duration of 5.1 years, 106 incidents of fatal or non-fatal strokes occurred among 1,203 participants aged 55-84 years. The R-FSRS applied to our cohort underestimated the 5-year risk for stroke in men and women. China-PAR performed better than the R-FSRS in terms of calibration (men, R-FSRS: χ 2-value 144.2 [P < 0.001], China-PAR: 10.4 [P = 0.238]; women, R-FSRS: 280.1 [P < 0.001], China-PAR: 12.5 [P = 0.129]). In terms of discrimination, R-FSRS and China-PAR models performed modestly in our cohort (C-statistic 0.603 [95% CI: 0.560-0.644] for men using China-PAR and 0.568 [95% CI: 0.524-0.610] using the R-FSRS; the corresponding numbers for women were 0.602 [95% CI: 0.564-0.639] and 0.575 [95% CI: 0.537-0.613). The recalibrated China-PAR model significantly improved the discrimination in C statistics and produced higher category-less NRI and IDI for stroke incidence than the R-FSRS. Although China-PAR fairly estimated stroke risk in our cohort, it did not sufficiently identify adults at high risk of stroke. Caution would be exercised by practitioners in applying the original China-PAR to Chinese older adults. Further studies are needed to develop an adequate prediction model based on the recalibrated China-PAR or to find new risk markers which could upgrade this model.

show abstract

Ischemic stroke prediction using machine learning in elderly Chinese population: The Rugao Longitudinal Ageing Study

Chang,

Zhang,

Shi

et al. 2023

Brain and Behavior

View full text Add to dashboard Cite

ObjectiveCompared logistic regression (LR) with machine learning (ML) models, to predict the risk of ischemic stroke in an elderly population in China.MethodsWe applied 2208 records from the Rugao Longitudinal Ageing Study (RLAS) for ischemic stroke risk prediction assessment. Input variables included 103 phenotypes. For 3‐year ischemic stroke risk prediction, we compared the discrimination and calibration of LR model and ML methods, where ML methods include Random Forest (RF), Gaussian kernel Support Vector Machines (SVM), Multilayer perceptron (MLP), K‐Nearest Neighbors Algorithm (KNN), and Gradient Boosting Decision Tree (GBDT) to develop an ischemic stroke risk prediction model.ResultsAge, pulse, waist circumference, education level, β2‐microglobulin, homocysteine, cystatin C, folate, free triiodothyronine, platelet distribution width, QT interval, and QTc interval were significant induced predictors of ischemic stroke. For ischemic stroke prediction, the ML approach was able to tap more biochemical and ECG‐related multidimensional phenotypic indicators compared to the LR model, which placed more importance on general demographic indicators. Compared to the LR model, SVM provided the best discrimination and calibration (C‐index: 0.79 vs. 0.71, 11.27% improvement in model utility), with the best performance in both validation and test data.ConclusionIn a comparison of LR with five ML models, the accuracy of ischemic stroke prediction was higher by combining ML with multiple phenotypes. Combined with other studies based on elderly populations in China, ML techniques, especially SVM, have shown good long‐term predictive performance, inspiring the potential value of ML use in clinical practice.

show abstract

Predicting 10-Year and Lifetime Stroke Risk in Chinese Population

Cited by 35 publications

References 29 publications

Stroke Prediction with Machine Learning Methods among Older Chinese

Stroke Prediction with Machine Learning Methods among Older Chinese

Validation of 10-Year Stroke Prediction Scores in a Community-Based Cohort of Chinese Older Adults

Ischemic stroke prediction using machine learning in elderly Chinese population: The Rugao Longitudinal Ageing Study

Contact Info

Product

Resources

About