Using AI To Personalize Healthcare And Improve Patient Safety

Using AI To Personalize Healthcare And Improve Patient Safety

What’s the current state of AI in healthcare? “Deployment is King” answer the Mayo Clinic’s Dr. John Halamka and Paul Cerrato. There have been very few working implementations so far of the numerous AI solutions for healthcare developed by startups and established companies. “For a solution to work within a health system, it needs to conform to specific governance protocols, integrate with its IT systems, and win the support of clinicians and administrators who will be using it on a daily basis,” write Halamka and Cerrato.

An example of such tightly integrated, working AI solution was announced last month by Ballad Health, the leading healthcare provider in the Appalachian Highlands. For a little over a year, Ballad Health has worked to deploy MedAware’s AI-based medication safety monitoring platform, embedding it within the workflow of its Epic electronic medical record or EMR system. For AI health startups, “the biggest challenge is finding a partner that would be patient enough to walk us through the integration with a big EMR, specifically Epic,” says Gidi Stein, MedAware’s co-founder and Chief Scientific Officer.

MedAware helps Ballad’s pharmacists identify evolving risks of medication-related harm during a patient’s hospital stay, providing them with visibility to personalized, patient-specific medication-related risks. This time-sensitive view allows Ballad pharmacists to identify at-risk patients by acuity level quickly, better prioritize patient care, and drive increased efficiencies. As a result, Ballad Health expects that thousands of patients will be safer from adverse drug events.

Medical errors may be responsible for more than a quarter of a million deaths annually in the U.S. alone. About 70% of medication errors that result in adverse drug events are prescription errors. Witnessing such errors while working as a physician in a leading Israeli hospital led Stein to launch MedAware as a data-driven platform for patient safety.

The widely-used decision-support systems that help physicians identify potential risks and adverse drug interactions, follow a rule-based approach, established on the basis of what is known about a specific medication as it applies to the general population. Stein derived his idea for MedAware’s solution from the shift from rule-based to data-driven systems for fraud identification he observed in the financial services industry: From alerting banks and consumers every time the system has identified a general anomaly to sounding the alarm only when there was a deviation from the personal data trail and the regular behavior of the specific customer.

“A systematic failure requires a systematic solution,” says Stein. The existing drug safety systems fail on a number of counts and typically involve various elements of the healthcare system. As many of the alerts they generate are false alarms, they create “alert fatigue” among physicians, pharmacists and nurses that tend to ignore them. At the same time, says Stein, “there are many other types of risks that are undetected by these rule-based systems.” In addition, they neglect to take care of what Stein calls “the temporal dimension,” paying attention to new risks associated with lab tests and vital signs and other data that arrive after the medication has been prescribed and dispensed.

In contrast, as a 24/7 monitoring system and a patient-specific, adaptable tool, MedAware’s solution is “reducing the number of alerts by more than 90% and improving their clinical accuracy and their actionability,” says Stein. The MedAware platforms not only monitors in real-time the present condition of specific patients, it is also aware of their past medical history and is capable of pointing to future potential risks.

MedAware’s predictive abilities were confirmed recently in a Harvard Medical School study. It validated MedAware’s algorithm for the identification of patients at risk of Opioid Use Disorder or OUD using more than 649,000 outpatient records. The MedAware solution achieved an accuracy of more than 93%, as compared with two expert opinions.

Many patients develop OUD after getting prescription opioids, resulting in over 70,000 deaths in the U.S. annually. MedAware can help physicians make better decisions when they prescribe opioids and identify patients in the early stages of addiction. “The patients that we identify are early enough in the process that there’s a good chance that we can save them from themselves,” says Stein. Moreover, the MedAware platform can be used as a population health tool, screening large populations to identify those that may need further evaluation for suspected OUD.

Another potential use of the MedAware platform—adding its AI algorithms to medical devices—has been demonstrated by its collaboration with Baxter International. Smart infusion pumps use Dose Error Reduction Systems (DERS) to help prevent medication errors by checking programmed doses against preset limits specific to a drug. Developing meaningful DERS limits across all drugs and care areas within a hospital’s drug library, and deploying required changes through thousands of pumps throughout the hospital, is challenging and requires detailed analysis and significant resources to maintain.

Analyzing 3,823,367 infusions performed over a 10-month period, a Baxter study found 44,819 pump programming entries that were outliers to common programming patterns. Approximately 25% of these outliers were identified through MedAware’s algorithms but did not trigger DERS and clinicians did not receive an alert. The study has demonstrated the potential to improve patient safety by using AI to build and maintain smart infusion drug libraries that dynamically review infusions and signal possible infusion errors, according to the Baxter investigators. “This is a completely different application based on the same core technology that we have developed,” says Stein.

MedAware serves as a great example of three significant developments: A. The success of AI solutions for healthcare is highly dependent on integration with the existing infrastructure and work practices of the healthcare system. B. MedAware approach to applying AI to patient safety demonstrates the overall shift, in healthcare and other domains, from rule-based to data-driven solutions which enable personalization. C. Modern AI, i.e., machine learning models analyzing large quantities of data, is heralding a transition from mass services to customization targeted at a narrow set of customers or even individuals.

link

Leave a Reply

Your email address will not be published. Required fields are marked *