研究成果：Analysis of financial pressure impacts on the health care industry with an explainable machine learning method: China versus the USA
发表期刊:《Expert Systems with Applications》, 2022.118482
近日，厦门大学医学院、健康医疗大数据国家研究院、管理学院、数据挖掘研究中心团队，联合中国人民大学统计学院、中南大学数学与统计学院、中南大学金属资源战略研究所和湖南大学工商管理学院，在期刊《Expert Systems with Applications》线上刊出题为“Analysis of financial pressure impacts on the health care industry with an explainable machine learning method: China versus the USA”的论文，该刊系人工智能、运筹学与管理科学领域TOP期刊，2022年的影响因子和中科院分区分别为8.665/中科院一区。
该论文通过构建可解释性人工智能模型，从时间和空间角度探究金融压力对中国和美国医疗保健股票市场波动的影响，为市场监管机构和投资者提供一些有价值的建议。该工作得到国家社会科学基金重大项目(20&ZD137)支持。论文内容请查看链接：https://linkinghub.elsevier.com/retrieve/pii/S095741742201569X。
原文摘要：This study analyzes the role of financial pressure in forecasting the volatility of health care stock. The main finding shows that financial pressure helps to improve the volatility forecasting performance of the health care stock in both China and the USA. Empirical analysis further suggests that XGBoost performance outperforms other benchmark models, especially advanced machine learning models. This study also interprets predictions to help financial institutions and investors make correct decisions using Shapley additive explanations. The results illustrate that the prediction contribution of financial pressure is much stronger in China than in the USA. The prediction contribution distribution of the five-dimensional indicator of financial pressure in China is more discrete than in the USA. Different lag periods of financial pressure have an asymmetric predictive contribution to the volatility of the health care stock. The volatility of Chinese health care stock is mainly influenced by the five-dimensional indicator of financial pressure at the medium and late period lag but at the front and medium period lag for the USA. These findings are crucial for policymakers and investors in promoting the sustained health care stock market through financial pressure regulation.Keywords: Health care, Financial pressure, Machine learning, Shapley additive explanations.
厦门大学数据挖掘研究中心
2022年8月31日
