The Healthcare Analytics Summit is back! Join us live in Salt Lake City, Sept. 13-15.Register Now
Article Summary
When healthcare leaders make data-driven decisions, they often think they see the same thing in the data and assume they’re drawing the same conclusions. However, decision makers often discover later that they were looking at the data differently and didn’t derive the same insights, leading to ineffective and unsustainable choices. Healthcare leaders can manage differing data interpretations by using statistical process control (SPC) methodology to find focus, avoid divergent data interpretations, make better decisions, and monitor change for a sustainable future. By deriving concise insights, SPC separates the signal from the noise, augmenting leaders’ decision-making capabilities.
This article is based on a webinar presented by Jason Jones, PhD, Chief Data Scientist for Health Catalyst, and Adam Ziegel, Product Manager for Vitalware, titled, “Find Focus and Monitor Change with Augmented Intelligence.”
When healthcare leaders make data-driven decisions, they often think they see the same thing when looking at the data, assuming they’re drawing the same conclusions and are ready to take the next steps based on shared insights. However, decision makers often discover months later that they were looking at the data differently from each other and didn’t derive the same insights.
To manage differing data interpretations, healthcare leaders can use statistical process control (SPC), a statistical technique using tools and procedures to monitor process behavior, discover issues in internal systems, and find solutions. Statistical process control methodology can help decision makers find focus, avoid divergent data interpretations, make better decisions, and monitor change for a sustainable future.
What Leaders Need for Better Healthcare Decision Making
Instead of approaching data-science-enabled decision making as artificial intelligence (AI), a tool to replace the human thought process, SPC helpsdata science用户进步到增强智能，一个更动态的数据科学视图。An augmented-intelligence perspective leverages technology’s top capabilities (e.g., math and memory) to enable human decision makers to make the best choices for their patients and organizations.
Via SPC, an augmented-intelligence approach helps leaders make better decisions in two meaningful ways:
#1: Separate the Signal from the Noise
增强智能通过从噪声中分离信号来帮助决策者找到焦点，这意味着从不必要的变化或波动中获取有意义的信息，使领导者能够在汇总数据中找到焦点。For example, if one hospital is performing better than another or if health system leaders set and evaluate animprovementgoal, augmented intelligence via SPC can identify which data is within the normal realm of fluctuation (noise) and which variation denotes superior or lesser performance (signal).
#2: Focus on the Future
从本质上讲，数据是历史性的，这使得领导者很自然地使用数据来回顾过去，而不是展望未来。决策者还必须牢记关注未来，并使用数据来监测变化——预测他们的组织在一年内将处于什么位置，他们是否喜欢这个预测，如果不喜欢，他们可以改变什么。如果决策者对基于数据的预测感到满意，他们必须评估这种表现是否可持续。
Deriving Common Insights and Determining Next Steps Together
To separate the signal from the noise and monitor change, healthcare leaders must derive common insights from their data and determine the next steps together. The SPC methodology supports both goals by using control charts, which help organizations study how a process or metric changes over time.
Separating the Signal from the Noise
例如，为了将信号从噪声中分离出来，卫生系统会比较一家拥有500张以上床位的旗舰医院和另一家拥有不到10张床位的农村医院在一段时间内的病例命中率(图1)。对这些数据运行SPC，卫生系统会识别出正常水平的变异率和趋势违反率，或模式分化率。组织因此决定了他们的表现中可能值得额外关注的异常值。此外，根据数据，最高人民法院后台运行的一个算法，以确定该组织是否看到了变化，如果达到农村医院和旗舰医院之间的水平。
图1的右边显示了过去24个月的实际数据。图上的每一个红点都标志着一个异常值，是组织认为正常变化之外的数据点，值得注意。图上的橙色点表示可能值得调查的趋势违反。
The most noteworthy insight from Figure 1 is an algorithm that runs in the background to determine if the data from the hospitals shows a shift in level. In this case, there are shifts in levels in the rural hospital and flagship hospital. The jumps in levels are the signal—the meaningful information—but what is it actually telling the decision makers? The shifts could be COVID-19 related and not from the encounters themselves, but digging deeper could reveal something separate from COVID-19, such as a coding delay. The role of SPC in such a scenario is to get down to what differentiates these two facilities from the others, giving decision makers a uniform insight.
Focusing on the Future
Another control chart, a forest plot, can help users further derive consistent insights from aggregate data (Figure 2) and focus on the future. With this approach, users can see the trees in the forest and the forest as a whole. In Figure 2, the Y-axis on the left-hand side shows the hospital graphs that Figure 1 considers. The X-axis contains the values for the measure of interest—again, in this case, therevenueintegrity case hit rate from July through September of 2020.
This forest plot shows how the system performs overall—at about a 10-percent revenue integrity case hit rate. There’s low variation (6 percent) with the secondary market hospital with 200 beds and another secondary hospital with about 300 beds and a regional hospital (both also about 6 percent), and so on.
If the individual points form an upward triangle, it means that the hospital is outperforming the system as a whole, which is this vertical line. A downward triangle indicates that a hospital is underperforming the others in the system. A circle means that, statistically, the hospitals are tied.
These insights immediately provide a performance picture of each hospital (the trees) and the system as a whole (the forest). In the example in Figure 2, the flagship hospital from Figure 1 is underperforming the rest of the hospitals in the system.
One limitation of forest plots is that viewers may question whether the triangles differ from each other. In response, an algorithm runs in the background of this analysis to provide letters on the far right indicating a three-way tie for top performance, statistically (letter A). Two hospitals have unique performance levels (B and C), and two more tie (D). Finally, E denotes a negative outlier, the flagship hospital.
通过通过算法添加性能信息，该情节通过简单地展示哪些医院相似，哪些医院不同来辅助决策。从这里，领导者看到的是相同的产出，而不是留下得出不同的见解。
Figure 2 also contains squares that use history to represent where this hospital will be in a year—the future focus. This insight requires separating signal from noise to show decision makers which facility is different from the system instead of which hospitals are clustering together. This example shows signs of improvement among some hospitals, but others, despite being tied in this plot, are moving in different directions. The rural hospital looks to be trending a little bit worse, while the regional 250-bed hospital shows slight improvement signs.
Statistical Process Control Drives Clearer Understanding, More Sustainable Healthcare Decision Making
By separating the signal from the noise and focusing on the future, SPC elevates leadership decision making to a more unified and sustainable process. Rather than replacing human judgment, SPC and related control charts clarify the data, providing succinct interpretations and predictions so that leaders can apply their expertise and skill sets directly to the problem at hand, versus losing precision due to divergent interpretations. With this methodology, SPC helps leaders leverage data science beyond artificial intelligence and on to augmented intelligence.
Additional Reading
你想了解更多关于这个话题吗?Here are some articles we suggest:
- Four Essential Ways Control Charts Guide Healthcare Improvement
- AI-Assisted Decision Making: Healthcare’s Next Frontier
- AI in Healthcare: Finding the Right Answers Faster
- Artificial Intelligence in Healthcare: A Change Management Problem
- Healthcare Data Literacy: A Must-Have for Becoming a Data-Driven Organization
PowerPoint Slides
你想使用或分享这些概念吗?Download the presentation highlighting the key main points.
Click Here to Download the Slides
This site uses cookies
We take pride in providing you with relevant, useful content. May we use cookies to track what you read? We take your privacy very seriously. Please see ourprivacy policy详情和任何问题。