As healthcare IT continues to evolve, tech-savvy organizations are turning to automated clinical interpretation as a way to speed information to physicians and other healthcare providers. This technology reviews defined data points and makes recommendations based on predefined parameters, prompting clinicians to look further into a situation and make an assessment of the patient’s condition.
Automated clinical interpretation is used throughout healthcare, underpinning medical alarms, alerts, protocols and so on. For example, this kind of technology highlights mammogram abnormalities, inputting symbols and colors around possible anomalies so radiologists can immediately spot areas of concern and closely review them. The software can also prioritize different images based on identified irregularities, which can efficiently direct workflow.
The benefits of automated clinical interpretation
There are substantial advantages to technology that uses data analytics to sift through volumes of information and pinpoint possible problems. As healthcare becomes more data rich, automated clinical interpretation offers organizations a way to navigate the volume and differentiate valuable, sometimes critical, information from noise.
Solutions that provide automated clinical interpretation are becoming especially valuable as organizations delve further into population health management. These solutions allow healthcare organizations to leverage data at a granular level, looking across populations in minutes, not weeks, to identify at-risk individuals and transition them into the right care setting. The technology can also zero in on disease outbreaks or evolving trends, such as an emerging influenza epidemic or the increased localized incidence of diabetes among young children.
Home health monitoring is another developing area for automated clinical interpretation. For example, a call center can watch over the health of a patient by reviewing data generated from a machine, such as a cardiac monitor. If staff in the call center receive an alert about an abnormal patient reading, they can respond, bringing immediate care to a patient who needs it. Not only does this technology elevate home health delivery, it allows patients to receive certain types of treatment in their home that have traditionally required a hospital stay for 24/7 observation.
Andrew Underhill, Chief Technologist, Systems Made Simple |
Risks to consider
While automated clinical interpretation is beneficial for catching problems early and managing population health, it is not without risk. In fact, there are some real dangers associated with this technology, and it can be a little disconcerting that key medical decisions could be made based on rules that IT software professionals create. The reality is that if organizations don’t fully understand the hazards involved in using automated clinical interpretation or take steps to mitigate those hazards, then these helpful tools also have the potential to do serious harm.
For instance, as clinicians increase their use of automated clinical interpretation, they may become desensitized and fail to recognize risks if they are not explicitly called out. Consider the example of a radiologist who begins relying on mammogram interpretation software to highlight concerning areas. Although the radiologist still reviews the films, he may focus more on those areas highlighted and not look at the rest of the mammogram as closely. If the software fails to underscore a potential issue, the radiologist could overlook the error, putting the patient in danger. Similarly, if the software prioritizes an image as low risk, the radiologist may take a more cursory look at the image, assuming the technology is correct, and inadvertently miss a budding problem.
There is also the chance for false positives with this technology, which can cause patients undo concern, drive up care costs and even lead to provider apathy over time. For instance, when automated clinical interpretation is used in home health monitoring, software that is not properly calibrated can sound alarms too often, causing the patient distress and increasing the costs of care if the provider responds to the alarm only to find out the patient is fine. Over time, if the problem is not resolved, monitoring organizations can start to dismiss alarms and alerts, and accidently miss a critical situation.
Allaying risks
To truly benefit from solutions that encompass automated clinical interpretation, organizations should be aware of the risks and gain a complete understanding of their consequences. Depending on the type of data, risk mitigation can vary; these include natural language processing, mobile app and medical device machine-generated data, and image processing. With this in mind, organizations should take precautions, such as the following, to lessen the chance for problems:
Use fully vetted companies and software. There are a lot of entrants into the automated clinical interpretation market, and not all are based on best practice. Home health monitoring software in particular covers a wide range in terms of safety and accuracy. When selecting technology to support clinical decision making, be sure to look for companies registered with the Food and Drug Administration (FDA) that offer approved devices and software, as these have gone through extensive testing and can enable the most reliable care.
Establish a governance and oversight process. This should address how the organization monitors solution performance, catches issues and makes improvements. For example, if there needs to be a change in protocol regarding an automated clinical interpretation tool, the governance structure should indicate who is qualified to authorize the change, such as a physician actively practicing in the related area of medicine. The governance process should also address training requirements for the technology, including timeframes, staff qualifications, content and so on.
Get involved in creating sound protocols. Many emerging solutions employ self-defined decision-making protocols based on independent opinion made in isolation. This can lead to significant risk if the protocols are off base. Organizations can limit this risk by getting directly involved in forming protocols. An effective way to do this is to join an industry forum that focuses on collective protocol development, taking into account shared best practice and established evidence.
Conduct trials. Before using any automated clinical interpretation tool, consider conducting a trial to determine whether the software functions as expected. While there have been many clinical trials for medical equipment, such as home monitoring devices, there have been few trials testing the software and decision algorithms that tie the equipment to the clinician’s decision. Taking time to test these protocols can avert potential shortfalls that could lead to patient harm.
Understand the care pathway. A patient’s care plan may change as a result of a clinical interpretation or event. However, an alert initiated by an automated system is normally one piece of a greater picture. Organizations should fully recognize the criticality and potential deficiencies associated with automated interpretations and appreciate where any alerts fall in the overall response to the patient. In other words, how critical is the alert and what other factors will be considered before making any alterations to the care plan?
Adopt a continuous improvement process. Automated clinical interpretation technology should not be installed and then operated without review. Organizations should perform regular discovery activities to assess whether the automated clinical interpretation protocols need amending.
Establish a resilient infrastructure. In certain situations, such as with home health monitoring software, the infrastructure that girds the solution is under the control of multiple entities, including telecommunications companies, Internet service providers, hospitals and medical equipment corporations. When implementing this type of technology, organizations should consider using a packaged service where one organization has ownership of problem resolution and can guarantee service consistency and quality equivalent to what is involved in supporting care in a hospital environment. So, for instance, if a patient is being monitored in the home, the solution and equipment are supported by a wireless system, Internet provider and telephone system that integrate well and deliver a high level of reliable service to facilitate continuous communication with the call center.
Empower users to question the unusual. While being proactive can minimize the likelihood of error, these systems can still fail. End users should be aware of their role in preventing error and empowered to question anything that seems out of place. For example, if a home monitoring system is not sending regular alerts or readings on a high-risk patient, the call center should be responsible for responding, perhaps making a phone call to the patient, to double check everything is functioning properly.
Being proactive yields benefits
The benefits of automated clinical interpretation are significant, and the use of this kind of technology will only grow in the future. Organizations that appreciate the dangers involved with automated clinical interpretation and work to allay risks will position themselves best to leverage this technology to its full potential.