Not yet a revolution, but machine learning systems can still make a big difference
May 12, 2017 AT 3:29 PM
For a while, now, proponents of technology innovations such as machine learning and artificial intelligence systems have been forecasting breakthroughs that will improve diagnostic accuracy, tailor treatments, and perhaps even replace work performed by some categories of clinicians.
But we’re not quite there yet.
Writing recently at the Health Affairs blog, a team of stakeholders from Booz Allen Hamilton, an analytics and management consulting company, along with an ER physician, note that, for all the expectation, machine learning has not yet arrived as the game changer many have been anticipating.
“While we believe machine learning holds great promise, it is far from clear how it will transform health and health care in the short to mid-term,” they write. “Today, policy makers and industry executives face decisions about when and how to invest in machine learning to optimize organizational effectiveness and efficiency without wasting capital funds on premature or nonvalue-adding technologies.”
With that caveat on the table, they go on to argue that “while many machine-learning solutions are not yet mature and sophisticated enough to support complex clinical decisions, machine learning can be effectively deployed today to reduce more routine, time-consuming, and resource-intensive tasks, allowing freed-up personnel to be redeployed to support higher-end work.”
For example, current limitations notwithstanding, providers can still use machine learning “to eliminate routine, ‘mundane,’ but resource-intensive processes,” such as reviewing patient histories prior to a visit. “This task can be very time consuming and cumbersome for clinicians, especially with large amounts of patient data, much of which may be contained in unstructured notes. Consequently, during patient visits, clinicians may only rely on a partial patient history, such as the most recent visit, or they resort to having the patient recount his or her history, which can be unreliable. Machine learning in conjunction with natural language processing can be used to go through a patient’s entire medical history in the EHR, instantly looking for hundreds to thousands of different crucial facts.”
After scanning a patient’s entire history, machine-learning systems can provide recommendations on what is important based on the patient’s presenting symptoms. “Clinicians can then be left to evaluate the outputs and make the best diagnosis, treatment, and care decisions.”
As for how best to take advantage of current machine learning options, the writers suggest healthcare organizations should, first, avoid using just one vendor, and second, opt for an infrastructure that is “based on modular and open architecture principles that make it easy to add or update components and integrate machine-learning solutions as plug-in functions to the EHR.”
Moreover, “to build, test, and deploy machine-learning algorithms, organizations need internal policies and mechanisms to transfer health data in and out of legacy systems securely,” thus enabling them to take advantage of industry innovations, and they must also be sure to articulate the goals or endpoints or any diagnostic or treatment process. “Without these standardized endpoints,” they argue, “it is not possible to ‘train’ machine-learning algorithms to sufficiently explain the variability in outcomes that can be translated into better tailoring of diagnostic or treatment processes.”
Looking ahead, the writers don’t doubt that “the speed of innovations in machine learning will continue to accelerate, and health care will be a key industry experiencing ‘disruption.’” A bona fide revolution, however, is still yet to come. Nonetheless, they say, “by taking a practical approach to evaluating and adopting machine learning, health systems can improve patient care today, while preparing for future innovations.”