D. Gorman is executive chair, Health Workforce New Zealand, and professor of medicine and associate dean, Faculty of Medical and Health Sciences, University of Auckland, Auckland, New Zealand. T.M. Kashner is health science specialist, Office of Academic Affiliations, Department of Veterans Affairs, Washington, DC, and research professor of medicine, Loma Linda University Medical School, Loma Linda, California.
Acad Med. 2018 Aug;93(8):1113-1116. doi: 10.1097/ACM.0000000000002109.
The authors propose that the provision of state-of-the-art, effective, safe, and affordable health care requires medical school graduates not only to be competent practitioners and scientists but also to be policy makers and professional leaders. To meet this challenge in the era of big data and cloud computing, these graduates must be able to understand and critically interpret analyses of large, observational datasets from electronic health records, third-party claims files, surveys, and epidemiologic health datasets.The authors contend that medical students need to be exposed to three components. First, students should be familiar with outcome metrics that not only are scientifically valid but also are robust, useful for the medical community, understandable to patients and relevant to their preferences and health goals, and persuasive to health administrators and policy decision makers. Next, students must interact with an inclusive set of analysts including biostatisticians, mathematical and computational statisticians, econometrists, psychometricians, epidemiologists, informaticians, and qualitative researchers. Last, students should learn in environments in which data analyses are not static with a "one-size-fits-all" solution but, rather, where mathematical and computer scientists provide new, innovative, and effective ways of solving predictable and commonplace data limitations such as missing data; make causal inferences from nonrandomized studies and/or those with selection biases; and estimate effect size when patient outcomes are heterogeneous and surveys have low response rates.
作者提出，提供最先进、有效、安全和负担得起的医疗保健服务，不仅要求医学院毕业生成为有能力的从业者和科学家，还要求他们成为政策制定者和专业领导者。为了在大数据和云计算时代应对这一挑战，这些毕业生必须能够理解和批判性地解读来自电子健康记录、第三方索赔文件、调查和流行病学健康数据集的大型观察性数据集的分析。作者认为，医学生需要接触三个组成部分。首先，学生应该熟悉不仅在科学上有效而且稳健、对医学界有用、患者易懂并与他们的偏好和健康目标相关、对卫生行政人员和政策决策者有说服力的结果指标。其次，学生必须与包括生物统计学家、数理统计学家、计量经济学家、心理计量学家、流行病学家、信息学家和定性研究人员在内的一组包容性分析人员进行互动。最后，学生应该在数据分析不是静态的环境中学习，不存在“一刀切”的解决方案，而是让数学和计算机科学家提供新的、创新的和有效的方法来解决可预测的常见数据限制，如缺失数据；从非随机研究和/或存在选择偏差的研究中进行因果推断；并且在患者结果异质且调查响应率低时估计效应大小。
