The goal of this newsletter is to provide healthcare and biopharma leaders with insights about how their organizations can leverage data science and machine learning. Normally, I discuss how recent research applies to the industry. But these are not normal times. The entire healthcare industry is focused on responding to COVID-19, beginning, of course, with the physicians, nurses, and other healthcare workers on the front lines. Covid-19 is projected to impact us for at least a year. How can data assist healthcare organizations in their response?
The first concern of every healthcare organization is the health and safety of their staff and of the patients that they serve. But you also want to weather the storm, and hopefully emerge from this crisis stronger. What is the role of data in this process?
What you need from your data right now is accurate, reliable, and timely information that provides visibility into a fast-evolving uncertain situation. The first step is to identify the most critical questions that will guide decision makers in your organization through the first phase of this crisis. For example, payers will want to establish how many of the patients they cover are in high-risk categories. Hospitals will want accurate tracking of the status of every admission to forecast ICU utilization, while post-acute providers will need this information to prepare for a surge in patients.
In many healthcare organizations this critical data exists, but is not easily accessible: it is scattered across different databases, involves infrequent data refreshes, or requires significant analytics work. Clearly, organizing the data will accelerate access to the most urgent insights. Indeed, this is typically the first stage of any new data initiative. But how can you execute a complex data cleaning project when your company is operating in crisis mode?
Here, COVID-19 actually provides two crucial components for success: focus and organizational buy-in. First, define a limited set of data elements that can be used to answer as many of the critical questions as possible. Next, work with your analytics team to generate a first version of the critical reports that you identified.
So far, this is just the standard process to create any new report. But you can go farther. These reports will be run daily, perhaps even multiple times a day. Every time you generate them is an opportunity to evaluate the workflow and identify the most time-consuming or error-prone steps. And since this is a top priority for the organization, you should be able to assemble a small team of analysts and engineers to address these bottlenecks. Moreover, with most other projects on hold, you are less likely to be hampered by interfering changes to database structure, a common obstacle for any data organization project.
Many epidemiologists expect that the first wave of COVID-19 will subside by late spring. But the virus will still pose a significant risk, and healthcare organizations will continue to be busy managing the impact of the disease. The steps that you take today to identify, organize, and validate your data will also help you succeed in this second phase. I will expand on this topic in the next post.