大数据与医学的艺术

Life is short, and the Art long to learn. ––––Chaucer

如果你和我们一样，你也会坚信这样一个概念：医学，其实是一门艺术。当然，医学的确离不开先进的科学技术，比如生物化学和分子生物学。但是，光学习三羧酸循环没法让我们捕捉到患者间的细微差别，光通过显微镜也没法让我们找到激励医生的源泉。相反，医生其实和音乐家、雕塑家、舞蹈家一样，需要通过“实践”来提高技艺，而不是通过学习数据或使用电脑。那么面对如今的“大数据”时代医生能做些什么呢？那些深藏在电子健康档案（EHR）中的二进制编码如何能使这样一门几百年来一直都只是依靠医生的判断或直觉的艺术发生变革呢？本文将尽量解答这些问题，探讨我们是否真的可以通过数据来完善医学这门艺术。

源于数据却高于“数据”

通过正确的信息以及恰当的信息分析工具，我们的确有可能提高我们照护患者的能力。以癌症预防为例，目前癌症的早期检测主要依赖于医生对现行癌症筛查指南的了解程度以及他们对这些指南的依从性。此外，还取决于患者是否愿意每年都去看医生并且听取医生的建议。如果患者不去看医生，或者患者去看了医生但医生没有提到患者有必要做结肠镜检查，那么可能就不会开展这一检查（这可能让患者松了一大口气！）。但是正确的工具和分析方法绝不会让这种情况发生。我们可以通过准确度正在不断提高的技术来识别出风险最高的人群，然后告知医生和患者是否有必要采取行动。有了足够的数据，我们甚至可以观察到癌症遗传趋势，这在以前是绝不可能的，也可以在传统筛查方案检测到癌症之前就已预测到癌症。

“经济的”数据管理

虽然医生可能意识不到，但数据的确可以通过改进费用补偿方式和降低总医疗成本而产生明显的经济效益。这主要是通过两种途径来实现的。第一种途径是用数据来更加准确地反映出患者病情的复杂性。老年保健医疗制度（Medicare）会基于诊断的严重程度来确定患者队列的风险校正评分，然后基于这个评分来确定老人健康保险优惠（Medicare Advantage）计划的补偿额度：评分越高，补偿额度就越高。有时候，这些多出来的钱还会交到主治医生手上。但是医生往往在申请理赔时都没能正确使用ICD-9代码，这让患者的病情看起来没那么复杂，所以得到的补偿就减少了。如今，通过新兴的数据采集工具，可以识别并纠正不正确的编码，尽早发现错失的机会以获得更多的补偿。

第二种途径是应用这些工具来改进医疗成本管理，这会给整个医疗保健体系带来直接效益。通过分析理赔申请并且与保险机构交互，这类软件可以识别出哪些患者属于高度使用者，并且显示出某个社区或卫生系统的医疗成本趋势。这样一来，医疗服务提供者可以针对某类患者或疾病状态制定成本控制策略，形成双赢的局面，既能降低再入院率和控制成本，也能改善患者的生活质量。

最近我们看到了一个很好的例子。一个社区卫生系统通过采用人群管理数据工具识别出了其管辖范围内哪些地理区域有很多居民都没有保险。这个群体的急诊服务使用率特别高，而照护质量指标非常低（比如糖尿病控制、疫苗接种率等等）。基于这些数据有针对性地提供帮助，这个社区卫生系统把低成本、高质量的初级医疗保健机构介绍给了这些居民，使他们的急诊服务使用率降至周围邻居的使用水平以下。与此同时，更好的疾病管理和照护协调也使得该社区的整体健康状况得以提升。最后，数据分析工具还能识别出价格昂贵但可以用仿制药替代的品牌处方药，从而找到更多节约成本的机会。

理性的“数据革命”

在我们为过去一些事情感到惋惜的同时，医疗信息技术的运用却往往需要人为外力的驱使，比如刺激计划或监管要求。目前政府已经开始常规采用刺激性的支付与补偿调整计划来鼓励大家广泛接受EHR。最失败的是“有意义使用规范（Meaningful Use Regulations）”计划成为了政府参与直接患者照护的典型代表。通过5年内每年给遵守规范者提供资金（不遵守规范者则予以罚款），“有意义使用”计划几乎单枪匹马地在医疗保健领域掀起了一场“大数据”革命，它要求医生购买电子健康档案系统并且将其用于人群管理。虽然这看起来是一件好事，但大部分医生都不认为这些系统是“有意义的”。事实上，许多人都在质疑建立电子档案到底有没有用。

无论反对者是否承认，EHR的确是一个新的非常强大的信息网络的支柱，而许多人都相信这个网络有能力使医疗保健服务发生变革。虽然我们不认为“数据革命”会像其他人所说的那样成为灵丹妙药，但我们的确承认正确的工具能让医生从数据中得到有用的信息，从而采取新的“颠覆性”策略来改进患者照护。

艺术是流动的；随着经验的积累，画布、颜料和视角都会逐渐发生变化。莱昂纳多·达·芬奇加深了全世界对于透视画法的理解；巴勃罗·毕加索掀起了一场现代艺术的革命。达·芬奇和毕加索的绘画风格都与他们的前辈截然不同，他们都表达了人类需要以符合其所处年代的方式来理解和描绘这个世界。我们这个年代也是一样，医学的艺术也是如此。科学在发生变化，我们提供医疗服务的角度和方式也应该发生变化。我们记录、检索以及理解健康和疾病的能力绝不可能与过去相同。但是我们对于患者的关注是始终存在的。我们始终有必要基于我们不断变化的对于健康和疾病的认识，感同身受地理解每个患者所处的独一无二的位置。这就是艺术。

Notte博士是美国宾夕法尼亚州阿宾顿纪念医院的家庭医生和临床信息学家。他是EHR Practice Consultants公司的合伙人，该公司主要帮助医生使用电子健康档案。Skolnik博士是阿宾顿纪念医院家庭医学住院医师项目的副总监/费城天普大学的家庭医学和社区医学教授。他是Redi-Reference公司的主编，这是一家制作移动应用程序的软件公司。

【编者按：The life so short, the craft so long to learn.（人生苦短，学海无涯。）——Geoffrey Chaucer（杰弗里·乔叟）】

爱思唯尔版权所有  未经授权请勿转载

By: CHRIS NOTTE, M.D., AND NEIL SKOLNIK, M.D.

Life is short, and the Art long to learn.

–Chaucer

If you are anything like us, you still hold firmly to the notion that medicine, at its core, is an art. Sure, medicine is informed by cutting-edge sciences, such as biochemistry and molecular biology. But one can’t capture the nuances of the patient interaction by studying the Krebs cycle, nor define the motivating forces driving physicians by looking through a microscope. On the contrary, physicians are artists, much like musicians, sculptors, or dancers. And, like any of these artists, it would seem that a physician’s craft should improve through "practice," not by studying data or using a computer. So how do we reconcile this in the era of "Big Data"? How can the "ones and zeros" living deep in the "guts" of our electronic health records (EHRs) promise to revolutionize an art that has relied solely on the judgment – or gut sense – of physicians for centuries? Here we will attempt to answer these questions and ponder whether or not we really can improve the art of medicine with the help of data.

It’s more than the ‘ones and zeros’

With the right information and the proper tools to analyze it, it is possible to conceive of an improvement in our ability to care for patients. Take, for example, the prevention of malignancies. Currently, early detection of cancer relies heavily on a physician’s knowledge of – and compliance with – current cancer screening guidelines. Success also depends on a patient’s willingness to come in for annual visits to receive the instruction. If the patient doesn’t show up for a physical, or if the provider neglects to mention the need for a colonoscopy when the patient does appear, the test may go unordered (much to the patient’s relief, perhaps!). But the right tools and analytics won’t let that happen. Instead, the technology will identify the highest-risk populations with ever-improving accuracy and notify both physician and patient of the need for action. With enough data, we may even be able to make observations in trends of cancer inheritance never before possible and predict cancer long before it might be detected by conventional screening protocols.

It pays to care about the data

Physicians may not realize it, but data can have a significant financial advantage, by improving reimbursement and decreasing the overall cost of care. This can be achieved in two ways. The first way is by using the data to paint a more accurate picture of patient complexity. Medicare assigns a risk-adjusted score to patient cohorts based on the severity of their diagnoses and reimburses Medicare Advantage plans based on that score: the higher the score, the better the reimbursement. Occasionally, those additional dollars are passed along to the treating physicians. But all too often physicians do not use ICD-9 codes properly on their claims, making their patients appear less complicated and thereby receiving lower reimbursement. Through emerging data collection tools, improper coding can be identified and corrected, and missed opportunities can be discovered early enough to capture additional funds.

The second way these tools can be used leads to direct benefit to the health care system in general, through improved medical cost management. By interfacing with insurers and analyzing claims, the software can identify patients who are high utilizers and can show trends in medical costs across a community or health system. This allows providers to target certain patients or disease states around which to build cost-containment strategies and create win-win scenarios that decrease hospital readmissions, limit cost, and improve patient quality of life.

We recently learned of a great example of this. Using a population management data tool, a community health system was able to identify a geographic area in their region with a large uninsured population. This group had a disproportionately high utilization of emergency medical services and very low care quality markers (such as diabetes control, vaccination rates, etc.). Through targeted outreach based on these data, the system was able to direct individuals into low-cost, high-quality primary care sites and reduce emergency service utilization to levels below the surrounding neighborhoods. Simultaneously, the health of the community improved through better disease management and care coordination. Finally, data analytics tools uncovered additional opportunities for savings by identifying expensive brand-name drug prescriptions that could be replaced with generic drug alternatives.

A reluctant revolution

As we have lamented on previous occasions, the adoption of health care information technology is often driven by artificial external forces, such as stimulus programs or regulatory requirements. The government has routinely used incentive payments and reimbursement adjustments in order to spur widespread acceptance of EHRs. Most infamously, the Meaningful Use Regulations program has become the poster child for government involvement in direct patient care. Through the use of annual payments over a 5-year period (combined with the threat of penalties for lack of compliance), Meaningful Use has almost single-handedly enabled the Big Data revolution in health care by requiring physicians to purchase electronic health records systems and use them for population management. While seemingly a good thing, most physicians would hardly regard these systems as "meaningful." In fact, many question if there is any value in having an electronic record at all.

Whether or not their detractors admit it, EHRs do form the backbone of a new and very powerful information network – one which many believe has the power to revolutionize health care. While we certainly do not view the "data revolution" as the panacea others have claimed it to be, we do recognize that the right tools are emerging to enable physicians to learn from data and implement new, novel, and "disruptive" strategies to improve patient care.

Art is not static; the canvas, the paints, and the viewpoints change over time as experience evolves. Leonardo da Vinci furthered the world’s understanding of perspective. Pablo Picasso led a revolution in modern art. Each was different from his predecessors, and each expressed a human need to understand and portray the world in a manner consistent with his age. The same is true of our age and the art of medicine. The science has changed, as has the viewpoint and perspective from which we provide care. Our ability to record, retrieve, and understand health and disease will never be the same. But the attention to the patient is ever present. The necessity of interpreting the shifting world of health and disease to provide an empathic understanding of each patient’s individual and unique place in the world will never go away. Therein lies the Art.

Dr. Notte is a family physician and clinical informaticist for Abington (Pa.) Memorial Hospital. He is a partner in EHR Practice Consultants, a firm that aids physicians in adopting electronic health records. Dr. Skolnik is associate director of the family medicine residency program at Abington Memorial Hospital and professor of family and community medicine at Temple University, Philadelphia. He is editor in chief of Redi-Reference Inc., a software company that creates mobile apps.