Hospitals are sitting on ever-growing mountains of 'big data,' pouring in each day from a variety of settings—clinical, billing, scheduling, and so on. But without movement to aggregate, manage, and analyze the data sets, hospitals could easily find themselves in information overload.
The Daily Briefing's Hanna Jaquith caught up with Ramesh Sairamesh, managing director of corporate strategy and predictive analytics, to discuss how hospitals are harnessing the power of 'big data' to create value for patients and drive operational efficiency.
Q: The buzzword 'big data' is thrown around often these days in health care, but just how 'big' is it?
Sairamesh: Big data can mean a lot of things to different people—are we talking about one hundred or one billion terabytes of data? Basically, it refers to the large and diverse data sets that hospitals are collecting on a day-to-day basis—from tissue cultures, imaging tests, real-time and remote monitoring systems, and even genomic tests. But the sheer amount of clinical data contained in electronic health record (EHR) systems is what has really pushed data collection to another level in hospitals.
Q: So now that hospitals have 'big data,' what are they planning on doing with it?
Sairamesh: Clearly, it seems intuitive that you should be able to use these data somehow- improve care and streamline operations internally. But it is not obvious how to truly leverage the data in meaningful ways or where to start. The value of amassing such large amounts of digital data has only been remotely realized. Unlocking the true potential with patient specific data, from a clinical efficiency and quality perspective, is what everyone is chasing.
Q: In what ways have you seen hospitals achieve strategic goals with big data analytics?
Sairamesh: Many hospitals and health systems have turned to care management or manual surveys to reduce patient readmissions, but this can be costly and fairly ineffective. Big data business intelligence acts like a magnifying glass for hospitals—it offers a way to identify patients at high risk of readmission early on in the process and create targeted interventions that make a difference.
Using Real-Time Analytics to Improve Precision in Care Delivery
Leveraging predictive algorithms and natural language processing
For example, we've worked with hospitals on readmission-prediction technology that combines clinical data feeds such as vitals, lab results, and medications in EHRs with data from billing and information systems to identify patients who are a higher risk of depression, malnutrition, disease progression. Leveraging text analytics to mine psychosocial data from unstructured clinician notes in real time can also provide a better picture of readmission risk, plus reduce the chart review time for clinical staff.
We've found that when given systems that integrate information with workflow, hospitals excel. One Midwest health care system piloting real-time technology had a 60% reduction in chart review time and a 14% reduction in heart failure readmissions on its pilot unit. Another West Coast hospital system saw a 65% reduction in chart review time and a $10,000 average savings per case manager per year.
Q: How can hospitals be sure they are collecting the right quantity and quality of data?
Sairamesh: Hospitals are often sitting on top of great data that they both actively and inadvertently collect because of the digital nature of most technologies used in patient care. There will always be questions about quality, and certainly the quantity can be overwhelming.
While we are only at the tip of the iceberg here, the goal should be to harness these data for improved precision in care.
Q: So it sounds like collecting the data isn't the issue—it's analyzing it. What's the right approach to take?
Sairamesh: By using advanced statistical modeling, artificial intelligence, predictive modeling and natural language processing techniques that have been honed over the years in other industries, we can be much more efficient about identifying critical information embedded in the data that would be impossible to surface manually.
Often, the story told by many disparate pieces of information strung together in unique, analytical ways can generate insight that would not be apparent if we were examining the data more manually and in digestible chunks.
Q: Can you summarize this, for Daily Briefing readers: Where does the industry stand on 'big data,' and what needs to happen next?
Sairamesh: We need to make use of the sophisticated mathematical tools that will help truly unlock the latent value in big data. There is tremendous opportunity to drive greater precision in care at the individual level by using these data to help close gaps in awareness about a patient’s condition or characteristics at any given moment in time. And by taking advantage of these available mathematical techniques, we can not only gain insight on various patient risks, for example, but it can also help illustrate what data we don’t have but should be collecting to be better armed in serving patients, particularly as we think about how to manage wellness in populations of patients.
There's a lot of work to be done in this space. Ultimately, the goal of these technologies should be a smart "feedback loop," not only enabling smarter decisions at the point of care but helping providers understand which interventions are most effective for their patients, depending on their unique clinical and behavioral needs.
Next in the Daily Briefing
Why top hospitals don't have top websites