Machine Learning System Catches Two-Thirds More Prescription Medication Errors than Existing Clinical Decision Support Systems at Two Major Hospitals
Researchers find a savings of more than one million dollars and prevention of hundreds, if not thousands, of adverse drug events could have been had with machine learning system
Support for artificial intelligence (AI) and machine learning (ML) in healthcare has been mixed among anatomic pathologists and clinical laboratory leaders. Nevertheless, there’s increasing evidence that diagnostic systems based on AI and ML can be as accurate or more accurate at detecting disease than systems without them.
Dark Daily has covered the development of artificial intelligence and machine learning systems and their ability to accurately detect disease in many e-briefings over the years. Now, a recent study conducted at Brigham and Women’s Hospital (BWH) and Massachusetts General Hospital (MGH) suggests machine learning can be more accurate than existing clinical decision support (CDS) systems at detecting prescription medication errors as well.
The researchers published their findings in the Joint Commission Journal on Quality and Patient Safety, titled, “Using a Machine Learning System to Identify and Prevent Medication Prescribing Errors: A Clinical and Cost Analysis Evaluation.”
A Retrospective Study
The study was partially retrospective in that the researchers compiled past alerts generated by the CDS systems at BWH and MGH between 2009-2011 and added them to alerts generated during the active part of the study, which took place from January 1, 2012 to December 31, 2013, for a total of five years’ worth of CDS alerts.
They then sent the same patient-encounter data that generated those CDS alerts to a machine learning platform called MedAware, an AI-enabled software system developed in Ra’anana, Israel.
MedAware was created for the “identification and prevention of prescription errors and adverse drug effects,” notes the study, which goes on to state, “This system identifies medication issues based on machine learning using a set of algorithms with different complexity levels, ranging from statistical analysis to deep learning with neural networks. Different algorithms are used for different types of medication errors. The data elements used by the algorithms include demographics, encounters, lab test results, vital signs, medications, diagnosis, and procedures.”
The researchers then compared the alerts produced by MedAware to the existing CDS alerts from that 5-year period. The results were astonishing.
According to the study:
- “68.2% of the alerts generated were unique to the MedAware system and not generated by the institutions’ CDS alerting system.
- “Clinical outlier alerts were the type least likely to be generated by the institutions’ CDS—99.2% of these alerts were unique to the MedAware system.
- “The largest overlap was with dosage alerts, with only 10.6% unique to the MedAware system.
- “68% of the time-dependent alerts were unique to the MedAware system.”
Perhaps even more important was the results of the cost analysis, which found:
- “The average cost of an adverse event potentially prevented by an alert was $60.67 (range: $5.95–$115.40).
- “The average adverse event cost per type of alert varied from $14.58 (range: $2.99–$26.18) for dosage outliers to $19.14 (range: $1.86–$36.41) for clinical outliers and $66.47 (range: $6.47–$126.47) for time-dependent alerts.”
The researchers concluded that, “Potential savings of $60.67 per alert was mainly derived from the prevention of ADEs [adverse drug events]. The prevention of ADEs could result in savings of $60.63 per alert, representing 99.93% of the total potential savings. Potential savings related to averted calls between pharmacists and clinicians could save an average of $0.047 per alert, representing 0.08% of the total potential savings.
“Extrapolating the results of the analysis to the 747,985 BWH and MGH patients who had at least one outpatient encounter during the two-year study period from 2012 to 2013, the alerts that would have been fired over five years of their clinical care by the machine learning medication errors identification system could have resulted in potential savings of $1,294,457.”
Savings of more than one million dollars plus the prevention of potential patient harm or deaths caused by thousands of adverse drug events is a strong argument for machine learning platforms in diagnostics and prescription drug monitoring.
Researchers Say Current Clinical Decision Support Systems are Limited
Machine learning is not the same as artificial intelligence. ML is a “discipline of AI” which aims for “enhancing accuracy,” while AI’s objective is “increasing probability of success,” explained Tech Differences.
Healthcare needs the help. Prescription medication errors cause patient harm or deaths that cost more than $20 billion annually, states a Joint Commission news release.
CDS alerting systems are widely used to improve patient safety and quality of care. However, the BWH-MGH researchers say the current CDS systems “have a variety of limitations.” According to the study:
- “One limitation is that current CDS systems are rule-based and can thus identify only the medication errors that have been previously identified and programmed into their alerting logic.
- “Further, most have high alerting rates with many false positives, resulting in alert fatigue.”
Alert fatigue leads to physician burnout, which is a big problem in large healthcare systems using multiple health information technology (HIT) systems that generate large amounts of alerts, such as: electronic health record (EHR) systems, hospital information systems (HIS), laboratory information systems (LIS), and others.
Commenting on the value of adding machine learning medication alerts software to existing CDS hospital systems, the BWH-MGH researchers wrote, “This kind of approach can complement traditional rule-based decision support, because it is likely to find additional errors that would not be identified by usual rule-based approaches.”
However, they concluded, “The true value of such alerts is highly contingent on whether and how clinicians respond to such alerts and their potential to prevent actual patient harm.”
Future research based on real-time data is needed before machine learning systems will be ready for use in clinical settings, HealthITAnalytics noted.
However, medical laboratory leaders and pathologists will want to keep an eye on developments in machine learning and artificial intelligence that help physicians reduce medication errors and adverse drug events. Implementation of AI-ML systems in healthcare will certainly affect clinical laboratory workflows.
—Donna Marie Pocius
Related Information:
AI and Healthcare: A Giant Opportunity
Machine Learning System Accurately Identifies Medication Errors
Differences Between Machine Learning and Artificial Intelligence
XPRIZE Founder Diamandis Predicts Tech Giants Amazon, Apple, and Google Will Be Doctors of The Future
Strategists agree that big tech is disrupting healthcare, so how will clinical laboratories and anatomic pathology groups serve virtual healthcare customers?
Visionary XPRIZE founder Peter Diamandis, MD, sees big tech as “the doctor of the future.” In an interview with Fast Company promoting his new book, “The Future Is Faster Than You Think,” Diamandis, who is the Executive Chairman of the XPRIZE Foundation, said that the healthcare industry is “phenomenally broken” and that Apple, Amazon, and Google could do “a thousandfold” better job.
Diamandis, who also founded Singularity University, a global learning and innovation community that uses exponential technologies to tackle worldwide challenges, according to its website, said, “We’re going to see Apple and Amazon and Google and all the data-driven companies that are in our homes right now become our healthcare providers.”
If this prediction becomes reality, it will bring significant changes in the traditional ways that consumers and patients have selected providers and access healthcare services. In turn, this will require all clinical laboratories and pathology groups to develop business strategies in response to these developments.
Amazon Arrives in Healthcare Markets
Several widely-publicized business initiatives by Amazon, Google, and Apple substantiate these predictions. According to an Amazon blog, healthcare insurers, providers, and pharmacy benefit managers are already operating HIPAA-eligible Amazon Alexa for:
- Appointments at urgent care facilities,
- Tracking prescriptions,
- Employee wellness incentive management, and
- Care updates following hospital discharge.
For example, the My Children’s Enhanced Recovery After Cardiac Surgery (ERAS Cardiac) program at Boston Children’s Hospital uses Amazon Alexa to share updates on patients’ recovery, the blog noted.
Alexa also enables HIPAA-compliant blood glucose updates as part of the Livongo for Diabetes program. “Our members now have the ability to hear their last blood glucose check by simply asking Alexa,” said Jennifer Schneider, MD, President of Livongo, a digital health company, in a news release.
And Cigna’s “Answers By Cigna” Alexa “skill” gives members who install the option responses to 150 commonly asked health insurance questions, explained a Cigna news release.
Google Strikes Agreements with Health Systems
Meanwhile, Google has agreements with Ascension and Mayo Clinic for the use of Google’s cloud computing capability and more, Business Insider reported.
“Google plans to disrupt healthcare and use data and artificial intelligence,” Toby Cosgrove, Executive Advisor to the Google Cloud team and former Cleveland Clinic President, told B2B information platform PYMNTs.com.
PYMNTs speculated that Google, which recently acquired Fitbit, could be aiming at connecting consumers’ Fitbit fitness watch data with their electronic health records (EHRs).
Apple Works with Insurers, Integrating Health Data
In “UnitedHealthcare Offers Apple Watches to Wellness Program Participants Who Meet Fitness Goals; Clinical Laboratories Can Participate and Increase Revenues,” Dark Daily noted that by “leveraging the popularity of mobile health (mHealth) wearable devices, UnitedHealthcare (UHC) has found a new way to incentivize employees participating in the insurer’s Motion walking program.” UHC offered free Apple Watches to employees willing to meet or exceed certain fitness goals.
The Apple Watch health app also enables people to access medical laboratory test results and vaccination records, and “sync up” information with some hospitals, Business Insider explained.
Virtual Care, a Payer Priority: Survey
Should healthcare providers feel threatened by the tech giants? Not necessarily. However, employers and payers surveyed by the National Business Group on Health (NBGH), an employer advocacy organization, said they want to see more virtual care solutions, a news release stated.
“One of the challenges employers face in managing their healthcare costs is that healthcare is delivered locally, and change is not scalable. It’s a market-by-market effort,” said Brian Marcotte, President and CEO of the NBGH, in the news release. “Employers are turning to market-specific solutions to drive meaningful changes in the healthcare delivery system.
“Virtual care solutions bring healthcare to the consumer rather than the consumer to healthcare,” Marcotte continue. “They continue to gain momentum as employers seek different ways to deliver cost effective, quality healthcare while improving access and the consumer experience.”
More than 50% of employers said their top initiative for 2020 is implementing more virtual care solutions, according to NBGH’s “2020 Large Employers Health Care Strategy and Plan Design Survey.”
AI Will Affect Clinical Laboratories and Pathology Groups
Diamandis is not the only visionary predicting big tech will continue to disrupt healthcare. During a presentation at last year’s Executive War College Conference on Laboratory and Pathology Management in New Orleans, Ted Schwab, a Los Angeles-area healthcare strategist and entrepreneur, said artificial intelligence (AI) will have a growing role in the healthcare industry.
Schwab’s perspectives on healthcare’s transformation are featured in an article in The Dark Report, Dark Daily’s sister publication, titled, “Strategist Explains Key Trends in Healthcare’s Transformation.”
“If you use Google in the United States to check symptoms, you’ll get five-million to 11-million hits,” Schwab told The Dark Report. “Clearly, there’s plenty of talk about symptom checkers, and if you go online now, you’ll find 350 different electronic applications that will give you medical advice—meaning you’ll get a diagnosis over the internet. These applications are winding their way somewhere through the regulatory process.
“The FDA just released a report saying it plans to regulate internet doctors, not telehealth doctors and not virtual doctors,” he continued. “Instead, they’re going to regulate machines. This news is significant because, today, within an hour of receiving emergency care, 45% of Americans have googled their condition, so the cat is out of the bag as it pertains to us going online for our medical care.”
Be Proactive, Not Reactive, Health Leaders Say
Healthcare leaders need to work on improving access to primary care, instead of becoming defensive or reactive to tech companies, several healthcare CEOs told Becker’s Hospital Review.
Clinical laboratory leaders are advised to keep an eye on these virtual healthcare trends and be open to assisting doctors engaged in telehealth services and online diagnostic activities.
—Donna Marie Pocius
Related Information:
2020 Executive War College on Lab and Pathology Management – April 28-29
Amazon and Apple Will Be Our Doctors in the Future, Says Tech Guru Peter Diamandis
Introducing New Alexa Healthcare Skills
Livongo for Diabetes Program Releases HIPAA-Compliant Amazon Alexa Skill
“Answers by Cigna” Skill for Amazon Alexa Simplifies, Personalizes Healthcare Information
2020 Predictions for Amazon, Haven, Google, Apple
Health Strategies of Google, Amazon, Apple, and Microsoft
How Big Tech Is Disrupting Big Healthcare
Strategist Explains Key Trends in Healthcare’s Transformation
Smartphone Apps Enable Healthcare Consumers to Receive Primary Care without Traditional Office Visits, But How Will They Provide Needed Medical Laboratory Samples?
These virtual office visits use artificial intelligence and text messaging to allow real physicians to diagnose patients, write prescriptions, and order clinical laboratory tests
Clinical laboratories may soon be receiving test orders from physicians who never see their patients in person, instead evaluating and diagnosing them through a smartphone app. In response to major changes in the primary care industry—mostly driven by consumer demand—mobile app developers are introducing new methods for delivering primary care involving smartphones and artificial intelligence (AI).
Medical laboratories and pathology groups should prepare for consumers who expect their healthcare to be delivered in ways that don’t require a visit to a traditional medical office. One question is how patients using virtual primary care services will provide the specimens required for clinical laboratory tests that their primary care providers want performed?
Two companies on the forefront of such advances are 98point6 and K Health, and they provide a glimpse of primary care’s future. The two companies have developed smartphone apps that incorporate AI and the ability to interact with real physicians via text messaging.
Virtual Primary Care 24/7 Nationwide
Dark Daily has repeatedly reported that primary care in America is undergoing major changes driven by many factors including increasingly busy schedules, the popularity of rapid retail and urgent care clinics, consumer use of smartphones and the Internet to self-diagnose, and decreasing numbers of new doctors choosing primary care as a career path.
Writing in Stat, two physicians who had just completed internal medicine residencies, explained their own decisions to leave primary care. In their article, titled, “We were inspired to become primary care physicians. Now we’re reconsidering a field in crisis,” Richard Joseph, MD, and Sohan Japa, MD, cited factors that include long hours, low compensation in comparison with specialty care, and deficiencies in primary care training. At the time of their writing they were senior residents in primary care-internal medicine at Brigham and Women’s Hospital in Boston.
They also pointed to a decline in office visits to primary care doctors. “Patients are increasingly choosing urgent care centers, smartphone apps, telemedicine, and workplace and retail clinics that are often staffed by nurse practitioners and physician assistants for their immediate health needs,” they wrote.
One solution to declining populations of primary care physicians is a smartphone app created by Seattle-based 98point6. The service involves “providing virtual text-based primary care across the entire country, 24/7 of everyday,” explained Brad Younggren, MD, an emergency physician and Chief Medical Officer at 98point6, in a YouTube interview. “It’s text-based delivery of care overlaid with an AI platform on top of it.”
The service launched on May 1, 2018, in 10 states and is now available nationwide, according to press releases. 98point6 offers the service through individual subscriptions or through deals with employers, health plans, health systems, and other provider organizations. The personal plan costs $20 for the first year and $120 for the second, plus $1 per “visit.”
- Subscribers use text messaging to interact with an “automated assistant” that incorporates artificial intelligence. While messaging, they can describe symptoms or ask questions about medical topics.
“After the automated assistant has gathered as many questions as it deems necessary, it hands [the information] off to a physician,” Younggren said. In most cases, all communication is via text messaging. However, the doctor may ask the subscriber to send a photo or participate in a video meeting.
- The doctor then makes a diagnosis and treatment plan. Prescriptions can be sent to a local pharmacy and the subscriber can be referred to a clinical laboratory for tests. LabCorp or Quest Diagnostics are preferred providers, but subscribers can choose to have orders sent to independent labs as well, states the company’s website.
Younggren claims the company’s physicians can resolve more than 90% of the cases they encounter. If, however, they can’t resolve a case, they can refer the patient to a local physician. And because most of 98point6’s interactions with subscribers are text-based, that messaging serves as reference documentation for other doctors, he said.
The 98point6 physicians are full-time employees and work with the company’s technologists to improve the AI’s capabilities, Younggren said. The company claims its doctors can diagnose and treat more than 400 conditions, including: allergies, asthma, skin problems, coughs, flu, diabetes, high blood pressure, and infections. For medical emergencies, subscribers are advised to seek emergency help locally.
98point6 also can function as a front end for interacting with patients in health systems that have their own primary-care doctors, Younggren said. The company’s health system clients “don’t actually have a good digital primary care front end to deliver care,” he said. “So, we can essentially give them that, and then we can also get some detailed understanding of how to coordinate care within the health system to drive patients to the care that they need.” For example, this can include directing the patient to an appropriate sub-specialist.
Leveraging Patient Data to Answer Health Questions
K Health in New York City offers a similar service based on its own AI-enabled smartphone app. The app incorporates data gleaned from the records of more than two million anonymous patients in Israel over the past 20 years, explained company co-founder Ran Shaul, co-founder and Chief Product Officer, in a blog post.
The software asks users about their “chief complaint” and then compares the answers with data from similar cases. “We call this group your ‘People Like Me’ cohort,” Shaul wrote. “It shows you how doctors diagnosed those people and all the ways they were treated.”
The K Health app is free, but for a fee ranging from $14 for a one-time visit to $39 for an annual subscription, users can text with doctors, the company’s website states.
Unlike 98point6, K Health’s doctors are employed by “affiliated physician-owned professional corporations,” the company says, not K Health itself.
“The doctor you chat with will discuss a recommended treatment plan that may include a physical exam, lab tests, or radiology scans,” states K Health’s website. “They may send you directly for some of these tests, but others will require you to visit a local doctor.”
These are just the latest examples of new technologies and services devised to help patients receive primary care. How a patient who uses a smartphone app gets the necessary clinical laboratory tests performed is a question yet to be answered.
Clinical laboratory leaders will want to watch this shift in the delivery of primary care and look for opportunities to serve consumers who are getting primary care from nontraditional sources.
—Stephen Beale
Related Information:
Bringing Primary Care to Smartphones
We Were Inspired to Become Primary Care Physicians. Now We’re Reconsidering A Field in Crisis
How K Delivers Free Personalized Healthcare Information
Robbie Cape Wants Everyone to Have Access to Affordable Primary Care
23andMe Invites Customers to Add Health and Drug Data to Stored Genetic Test Results, Encroaching on Markets Where Both Apple and Clinical Laboratories Generate Revenue
Combining consumers’ health data, including clinical laboratory test results, to genetic data for predispositions to chronic diseases could be key to developing targeted drugs and precision medicine treatments
Genetic testing company 23andMe is beta testing a method for combining customers’ private health data—including clinical laboratory test results and prescription drug usage—with their genetic data to create the largest database of its kind.
Such information—stored securely but accessible to 23andMe for sale to pharmaceutical companies for drug research and to diagnostics developers—would place 23andMe in a market position even Apple Health cannot claim.
Additionally, given the importance of clinical lab test data—which makes up more than 70% of a patient’s medical records—it’s reasonable to assume that innovative medical laboratories might consider 23andMe’s move a competitive threat to their own efforts to capitalize on combining lab test results with patients’ medical histories, drug profiles, and demographic data.
23andMe plans to use third-party medical network Human API to collect and manage the data. Involvement in the beta test is voluntary and currently only some of the genetic company’s customers are being invited to participate, CNBC reported.
Apple Healthcare, 23andMe, and Predicting Disease
The announcement did not go unnoticed by Apple, which has its own stake in the health data market. Apple Healthcare’s product line includes:
- Mobile device apps for using at point-of-care in hospitals;
- iPhone apps that let customers store and share their medical and pharmaceutical histories and be in contact with providers;
- ResearchKit, which lets researchers build specialized apps for their medical research;
- CareKit, which lets developers build specialized monitoring apps for patients with chronic conditions; and
- Apple Watch, which doubles as a medical device for heart monitoring.
What Apple does not have is genetic data, which is an issue.
An Apple Insider post notes, “As structured, 23andMe’s system has advantages over Apple’s system including not just genetic data, but insights into risks for chronic disease.”
This is significant. The ability to predict a person’s predisposition to specific chronic diseases, such as cancer, is at the heart of Precision Medicine. Should this capability become not only viable and reliable but affordable as well, 23andMe could have a sizeable advantage in that aspect of the health data market.
Genetic Test Results Combined with Clinical Laboratory Test Results
23andMe is hopeful that after people receive their genetic test results, they will then elect to add their clinical laboratory results, medical histories, and prescription drug information to their accounts as well. 23andMe claims its goal is to provide customers with easy, integrated access to health data that is typically scattered across multiple systems, and to assist with medical research.
“It’s a clever move,” Ruby Gadelrab, former Vice President of Commercial Marketing at 23andMe who now provides consulting services to health tech companies, told CNBC. “For consumers, health data is fragmented, and this is a step towards helping them aggregate more of it.”
CNBC also reported that Gadelrab said such a database “might help 23andMe provide people with information about their risks for complex, chronic ailments like diabetes, where it’s helpful for scientists to access a data-set that incorporates information about individual health habits, medications, family history and more.”
Of course, it bears saying that the revenue generated from cornering the market on combined medical, pharmaceutical, and genetic data from upwards of 10-million customers would be a sizable boon to the genetic test company.
CNBC reported that “the company confirmed that it’s a beta program that will be gradually rolled out to all users but declined to comment further on its plans. The service is still being piloted, said a person familiar with the matter, and the product could change depending on how it’s received.”
Will 23andMe Have to Take on Apple?
23andMe already earns a large portion of its revenue through research collaborations with pharmaceutical companies, and it hopes to leverage those collaborations to produce new drug therapies, CNBC reported.
This new venture, however, brings 23andMe into competition with Apple on providing a centralized location from where consumers can access and share their health data. But it also adds something that Apple does not have—genetic data that can provide insight into consumers’ predispositions to certain diseases, which also can aid in the development of precision medicine treatments for those diseases.
Whether Apple Healthcare perceives 23andMe’s encroachment on the health data market as a threat remains to be seen.
Nevertheless, this is another example of a prominent company attempting to capitalize on marketable customer information. Adding medical information to its collected genetic data could position 23andMe to generate significant revenue by selling the merged data to pharmaceutical companies and diagnostics developers, while also helping patients easily access and share their data with healthcare providers.
It’s a smart move, and those clinical laboratory executives developing ways to produce revenue from their lab organization’s patient lab test data will want to watch closely as 23andMe navigates this new market.
—JP Schlingman
Related Information:
23andMe is Moving into Apple’s Territory with a Pilot to Pull in Medical Data, Not Just DNA
23andMe Venturing onto Apple’s Turf with Health Data Collection
23andMe Already Has Millions of People’s DNA. Now It Wants Their Health Data Too.
23andMe Wants to Collect Users’ Medical Data, Stepping into Apple’s Territory