Leveraging EHRs and advancing patient care
October 03, 2016
By Neil Smiley
It was a joke in the early days of electronic health records that EHR stood for “Empty Health Record.” We’ve come a long way in filling in those empty records, but unfortunately, some practitioners still feel the time they spend entering data yields very little in return. Of course, the promise of health care analytics is that this EHR data has the potential to advance care, to help practitioners better predict and manage population risk and aid in prevention. With faster computers, ever-growing data sets and sophisticated algorithms, there should be no limit to our ability to glean insights and drive new innovation, right?
What’s missing?
However, even with the billions of dollars invested in EHRs, meaningful analytics have been slow in coming, and some practitioners are beginning to wonder if the promise of big data has been overhyped. The truth is that the path to success doesn’t lie only in the data, but also in the human insight to make sense of the analysis.
There’s no doubt the potential of health care analytics is huge. However, we are falling short when it comes to bridging the gap between the practitioners using EHRs in daily patient encounters and the data scientists tasked with unlocking these analytical powers. In order to make good on the potential, we must recognize, and effectively capitalize on, the important roles that both practitioners and data scientists play in this collaborative effort. To that end, here are the key challenges thwarting health data analytics, and how practitioners and data scientists can collaborate to advance patient care and improve health outcomes.
• Data can be dirty
Much of the data from which insights could be derived is unstructured, inconsistent and mismatched. The old cliché, “garbage in, garbage out,” certainly applies here. Much of the work associated with data analytics involves cleaning up the data before you can even begin the fun stuff — exploring and arriving at insights. Using technology to harmonize the data can help, but practitioners also play a large role in helping data scientists identify anomalies that might waste time or send analysts down a path toward misleading results.
• Data can be vast
Asking data scientists to comb through terabytes of health care data and surface insights without a practitioner’s guidance is like sending them on a far-flung journey without a map or even a specific destination. There are simply too many blind alleys and rabbit holes along the way. It requires an informed guide to narrow the range of potential paths for exploration.
Knowing the right question to ask is perhaps the most difficult aspect of data analytics. Practitioners can use their domain knowledge to frame a problem or a question that needs to be answered. Data scientists can find correlations in data, but they might be clueless at understanding causation. Working together, data scientists can help practitioners see patterns they might otherwise miss, and practitioners can help data scientists understand the meaning and context behind the data.
• Data can be old
Just 5 percent of patients account for almost half of all health care costs. If only we could find those high-risk patients and focus all of our efforts there. The problem is, using last year’s claims data can be ineffective at predicting this year’s high-cost patients. It won’t be the same patients, but instead a different 5 percent of individuals from the year before. Older data can provide context and supplement predictive models, but good predictions need real-time data that stays up to date as a care episode evolves.
The quality and arrival time of data is dependent on both the work patterns of practitioners on the ground and the effective integration and deployment of telehealth technology. Here, practitioners and data scientists can work together to examine workflows, patient interactions and care episode dynamics, to implement systems that are kept current, so that delays in data arrival don’t get in the way of patient care.
• Data can have gaps
With the move to value-based reimbursement, practitioners must coordinate care across care settings. Data scientists should be deriving insights from the entire care episode, but too often data access is limited to just the care settings within the practitioner’s own organization. These gaps ultimately undermine the accuracy of analytical models developed by data scientists.
The technology required for secure exchange of clinical data exists, but access is often blocked by the absence of a workable legal framework for data sharing. These challenges can only be addressed by practitioners, who must answer questions about what data should be shared, how it will be used, and how to balance care coordination and patient privacy needs with the competitive interest of their own organizations.
Through hard work and collaboration among practitioner organizations, effective data use agreements can be established to enable data scientists to leverage information now locked away in system silos.
About the author: Neil Smiley founded Loopback Analytics in 2009 to deliver an advanced Software-as-a-Service platform health care providers can use to prevent costly readmissions.
It was a joke in the early days of electronic health records that EHR stood for “Empty Health Record.” We’ve come a long way in filling in those empty records, but unfortunately, some practitioners still feel the time they spend entering data yields very little in return. Of course, the promise of health care analytics is that this EHR data has the potential to advance care, to help practitioners better predict and manage population risk and aid in prevention. With faster computers, ever-growing data sets and sophisticated algorithms, there should be no limit to our ability to glean insights and drive new innovation, right?
What’s missing?
However, even with the billions of dollars invested in EHRs, meaningful analytics have been slow in coming, and some practitioners are beginning to wonder if the promise of big data has been overhyped. The truth is that the path to success doesn’t lie only in the data, but also in the human insight to make sense of the analysis.
There’s no doubt the potential of health care analytics is huge. However, we are falling short when it comes to bridging the gap between the practitioners using EHRs in daily patient encounters and the data scientists tasked with unlocking these analytical powers. In order to make good on the potential, we must recognize, and effectively capitalize on, the important roles that both practitioners and data scientists play in this collaborative effort. To that end, here are the key challenges thwarting health data analytics, and how practitioners and data scientists can collaborate to advance patient care and improve health outcomes.
• Data can be dirty
Much of the data from which insights could be derived is unstructured, inconsistent and mismatched. The old cliché, “garbage in, garbage out,” certainly applies here. Much of the work associated with data analytics involves cleaning up the data before you can even begin the fun stuff — exploring and arriving at insights. Using technology to harmonize the data can help, but practitioners also play a large role in helping data scientists identify anomalies that might waste time or send analysts down a path toward misleading results.
• Data can be vast
Asking data scientists to comb through terabytes of health care data and surface insights without a practitioner’s guidance is like sending them on a far-flung journey without a map or even a specific destination. There are simply too many blind alleys and rabbit holes along the way. It requires an informed guide to narrow the range of potential paths for exploration.
Knowing the right question to ask is perhaps the most difficult aspect of data analytics. Practitioners can use their domain knowledge to frame a problem or a question that needs to be answered. Data scientists can find correlations in data, but they might be clueless at understanding causation. Working together, data scientists can help practitioners see patterns they might otherwise miss, and practitioners can help data scientists understand the meaning and context behind the data.
• Data can be old
Just 5 percent of patients account for almost half of all health care costs. If only we could find those high-risk patients and focus all of our efforts there. The problem is, using last year’s claims data can be ineffective at predicting this year’s high-cost patients. It won’t be the same patients, but instead a different 5 percent of individuals from the year before. Older data can provide context and supplement predictive models, but good predictions need real-time data that stays up to date as a care episode evolves.
The quality and arrival time of data is dependent on both the work patterns of practitioners on the ground and the effective integration and deployment of telehealth technology. Here, practitioners and data scientists can work together to examine workflows, patient interactions and care episode dynamics, to implement systems that are kept current, so that delays in data arrival don’t get in the way of patient care.
• Data can have gaps
With the move to value-based reimbursement, practitioners must coordinate care across care settings. Data scientists should be deriving insights from the entire care episode, but too often data access is limited to just the care settings within the practitioner’s own organization. These gaps ultimately undermine the accuracy of analytical models developed by data scientists.
The technology required for secure exchange of clinical data exists, but access is often blocked by the absence of a workable legal framework for data sharing. These challenges can only be addressed by practitioners, who must answer questions about what data should be shared, how it will be used, and how to balance care coordination and patient privacy needs with the competitive interest of their own organizations.
Through hard work and collaboration among practitioner organizations, effective data use agreements can be established to enable data scientists to leverage information now locked away in system silos.
About the author: Neil Smiley founded Loopback Analytics in 2009 to deliver an advanced Software-as-a-Service platform health care providers can use to prevent costly readmissions.