Every health system has a business intelligence team. Its purpose is often broad and vague, such as “increasing income and patient safety while complying with all regulations and standards.”
When the health system faces particularly difficult challenges, such as new risk-sharing payer contracts that require careful risk reduction, and other possible examples, these BI teams will be called.
In the business world, BI is usually done around products: how many times have they been downloaded, how many times have users clicked the red button, and so on. However, in healthcare, BI is usually only done at the highest level: the number of lives saved, the dollar cost.
The medical BI team takes time to segment the patient population and then track the results of these patient segments. However, the problem is that this top-down approach tends to ignore the complexity of intermediate intervention. There are often too many confounding factors: Which interventions are actually effective?
Linking patients to interventions and ultimately to results requires the BI team to spend more time on details. This interrelated understanding needs to be as rigorous as the artificial intelligence program needs to be successful, safe, and effective. And this kind of understanding can only be achieved through careful design.
At Penn Medicine, we created an integrated product team around our AI application. These product teams are made up of data scientists, doctors, and software engineers, to name a few.
This year we also began to include a BI analyst to help us make better design decisions so that it is feasible to connect patients with interventions and results, and can be easily reported through dashboards and reports.
In our first joint project, we deployed a machine learning application to better find incomplete patient records. Our pilot showed immediate improvement, but these benefits began to diminish after deployment.
Due to the engineering level of entering our BI dashboard, we have an in-depth understanding of what happened in these intermediate steps, and can quickly zero the problem and make corrections.
Business intelligence in healthcare is about making the right decisions. Data science in healthcare aims to provide insights that help make better decisions. A health system that uses the natural consistency of these two disciplines may see better results faster.
Mike Draugelis is the chief data scientist at Penn Medicine.