Data and analytics have become increasingly critical to the operation of any successful healthcare organization. And with the advent of healthcare imperatives such as value-based care and population health management, analytics technology has become more important than ever.
Here, four experts in healthcare analytics technology offer their advice and suggestions for healthcare CIOs implementing analytics in their provider organization. These are a variety of best practices for analytics implementation in healthcare.
Stakeholders, and required data
Implementing an analytics system first requires outcomes defined by multiple stakeholders that second drives alignment on what data elements are required, said Bradley Hunter, a research director at KLAS Research.
"Collaboration is a cornerstone to helping drive outcomes," Hunter said. "In the absence of collaboration amongst key stakeholders in the organization there will be no alignment on which outcomes should be focused on, and unnecessary effort will be expended."
The best practice is to bring all key stakeholders together to collaborate and debate on which outcomes will be key to the success of the organization, Hunter said. Once these are decided upon, then a strategy for achieving outcomes can be put in place – then one can go and get the data, he added.
The goal of getting the data is to tell the right story to the right people at the right time so they can make the best decisions, said Ryan Pretnik, a research director at KLAS.
"Good data drives great decisions," Pretnik stated. "This is why alignment on outcomes is so key – it drives what data needs to be brought in front of the decision makers. Once the outcomes are defined, the needed data becomes apparent."
This allows for the implementation of the data platform to be much more straightforward, he explained, adding that there are a myriad of available data sources, and knowing which data elements are needed before implementation helps to streamline the process.
"Those looking to implement an analytics tool should involve their end users early on in the process," Pretnik said. "This will help organizations choose a solution that is easy for end users to navigate and understand. Involving end users early in the process ensures high adoption which leads to more consistent outcomes."
A shared vision, and AI
In the past decade, new incentives and value-based programs that reward payers and providers for proactively managing the health of members have increased their collaboration and created an even greater need for data and analytics to get the right care to the right patient at the right time.
"CIOs at provider organizations seeking to implement analytics technology face a huge variety of choices and competing priorities, not to mention the work that goes into setting up and using analytics," said Mark Morsch, vice president of technology at health IT and analytics vendor Optum. "Collaborate with a shared vision. To be successful in a business setting, investments in advanced analytics and AI must have a defined objective and align to your overall technology strategy."
Further, healthcare is so specialized that it’s important to build a team of multi-disciplinary professionals who have advanced analytics technology paired with healthcare expertise one needs to drive clinical and financial performance, Morsch added.
"The right partner has this combination of industry experience and technology and business savvy to help CIOs make strategic investments that can build over time," he said.
"An effective approach can be to focus on processes in either area – clinical or administrative – that have a well-defined business need and available data."
Mark Morsch, Optum
Another best practice is to start with applications of data and analytics that can show an immediate impact by freeing up time and cost, Morsch said.
"These applications can span both clinical and administrative processes within a health system’s operations," he said. "An effective approach can be to focus on processes in either area – clinical or administrative – that have a well-defined business need and available data. One of the most interesting recent developments is more applied uses of artificial intelligence, like natural language processing for revenue cycle management or deep learning models that support disease prediction."
Applying natural language processing can help make sense of the vast amount of unstructured information in EHR clinical notes to gain new insights about cost, quality, and access and opportunities, he added.
"In a more practical fashion, natural language processing can help transform revenue cycle operations with intelligent assistance to coding and clinical documentation specialists to find the key diagnostic and treatment information in a medical record," he said.
"Combining the information captured using natural language processing with machine learning is also emerging as effective technique to recognize indicators of undiagnosed conditions and facilitate connections between providers to identify gaps in care," he added.
These use-cases must be backed by a business case and defined problem where technology can augment the human – patient, provider or administrator – experience, he added.
Analytics – and any IT infrastructure – needs to be pursuant to a business case: That business drives any technology first and foremost, and that requires a stakeholder analysis of existing conditions so one can best understand how the change management will affect each of the stakeholders, said Dr. Alan Pitt, chief medical officer at CloudMedx, a big data health analytics company.
"That business purpose is usually defined in a pre-state and post-state," Pitt said. "Process, people and technology doing something. Better patient care, shorter length of stay, filling gaps in care, a myriad of reasons for justifying people, process and technology. The user of higher analytics and AI will change the process and technology for the business purpose."
A second best practice would be that technology vendors and CIOs should work hand in glove with a clinical sponsor for all new initiatives, he stated.
"They should not be done in isolation," he said. "As it specifically pertains to analytics and AI, current problems that could be addressed include the overwhelming amount of documentation and reporting that providers are required to do leading to burnout. And that analytics and AI offer the opportunity to change our relationship to the sea of data that we have to deal with on a daily basis as providers."