Turning unruly data into valuable insights: the holy grail or achievable?

Let’s define "Data Strategy” as the method for effectively using objective facts to make better decisions. In reality, it’s less about the data than it is about the insights derived from that data. This obstacle has significantly challenged healthcare organizations. We’ve all seen the studies showing data-driven organizations grow faster, are more profitable, and have a faster time to market. Why does gaining insights from data seem to take so long? As one executive said to his analyst, “if the data you end up publishing goes against my gut, I’ll probably just go with my gut.”

There are reasons why this executive’s view is not uncommon. The pace for obtaining clean data, new data sources, and insights from the data takes a substantial amount of time

1. Create a doable strategy.
   
   
   1. Healthcare organizations have more data than ever before, and many are challenged to leverage it.
      
      
2. Insights: Develop a list of key business questions and insights that can be formed or answered by the data.
   
   
   1.  Preliminary questions: Focus on the key questions to understand your objectives.
      1. Simplify your approach by focusing on answering three questions:
         1.  What is our organization’s current level of performance?
         2.  Why is our organization at that level?
         3.  What can our organization do to improve?
      2. Keep your focus on the progression:
         1. What measures of success are most important for performance?
         2. Identify the strategic questions for each metric.
            For example:
            Is your organization looking for growth, market expansion, or greater profitability? How much focus needs to be on quality?
      3. Insights answered by the data: Focus on your performance against your goals.

1. Review your organization’s performance based on the insights illustrated in your data.

1. How is your organization performing against your goals?
2. Why are you performing at that level?
3. If you are not performing at the level you expected, why not?
   
   

3. Validation: Assess your data to ensure informed, actionable insights.
   
   
   1. Clean and normalized data will immensely help improve your analysis.
      1. Fortunately, automation can help normalize your data and transition staff time from cleaning the data to mining the data for insights.
         1. New standards such as FHIR (Fast Healthcare Interoperability Resources), tools like AI (Artificial Intelligence) and machine learning also contribute to progress.
      2. Assess the data, compare it against your key goals (above) to determine what’s needed.

1. Do interconnected data dependencies exist? Include data that is achievable for integration with the frequencies you need.
1. Inputs will not always be readily available, therefore it is important to derive or develop good proxies.
   1. For example, Social Determinants of Health is difficult to obtain, geographic characteristics of populations are being used as proxies for an individual’s level of clinical risk based on their socio-demographics. Other examples, include open and closed claims sources, disease registries, and financial performance data.
      
      

4. Activation: Create your roadmap.
   
   
   1.  Align your organization to the priority of time, resource investments required to execute your data strategy.
   2. Prioritize use cases of decisions and associated metrics.
   3. Determine the data needed both inside and outside your organization.
   4. Determine the unified data environment to support analysts and data scientists.
      
      
5. Evolve your data organization.
   
   
   1. Healthcare organizations are increasingly getting better at organizing and hiring new skill sets required to advance their data strategy needs. Healthcare organizations are appointing Chief Data Officers responsible for access to quality data who in turn are hiring data stewards responsible for understanding the data issues related to the source system (primarily transaction systems).
   2. Operational changes to transaction systems create downstream data as it is incorporated into analytic applications, data stewards are responsible for identifying such issues and ensuring these systems don’t compromise the integrity of the data.
      
      

Wrangling data and creating a culture that relies on insights to make decisions is difficult, leading many organizations failing to move their organization past a culture of “reporting” with the ability to gain insights to inform decision making. Organizations that follow the process with the proper investment and focus on the right goals will be rewarded. You can achieve a “doable data strategy.”