Hypertension is a major risk factor for cardiovascular diseases (CVD). Identifying the persons at these high risks plays an important role because it would save time and money before using any complex, invasive and/or expensive diagnostic methods. This task can be partly dealt with the help of advanced machine-learning methods. Specifically, a prediction model can be developed based on some indicator factors of the people at high risks which are easily-obtained, non-invasive and low-cost. This paper presents the first work towards predicting hypertension risks based on of the people in the Northwest region of Vietnam where hypertension rate is increasing. 2.509 samples were collected and classified into two levels of hypertension or no hypertension. We investigated and compared the performance of robust machine learning methods including single classifiers such as Naïve Bayes, MLP, Decision Tree, kNN and SVM; and ensemble classifiers such as bagging, boosting and voting methods, to generate mathematical models to predict the risk of hypertension disease. The experimental results showed that the best random forest model yielded 72.39% in the F1 score. This result was quite promising and can be applied in Vietnamese hospital nowadays.

高血压是心血管疾病（CVD）的主要危险因素。识别处于这些高风险中的人起着重要作用，因为它可以在使用任何复杂、侵入性和/或昂贵的诊断方法之前节省时间和金钱。这项任务可以部分借助先进的机器学习方法来完成。具体来说，可以根据高危人群的一些容易获得、无创、成本低的指标因素，建立预测模型。本文介绍了基于越南西北部地区高血压发病率不断增加的人群预测高血压风险的第一项工作。共采集样本2.509份，分为高血压或无高血压两个级别。我们调查并比较了鲁棒机器学习方法的性能，包括单一分类器，如朴素贝叶斯、MLP、决策树、kNN 和 SVM；和集成分类器，如 bagging、boosting 和投票方法，以生成数学模型来预测高血压疾病的风险。实验结果表明，最佳随机森林模型的 F1 得分为 72.39%。这个结果很有希望，现在可以在越南医院应用。