| ai_and_the_next_generation_of_physicians | AI and the Next Generation of Physicians | Latest_News | Shared_Content/News/Latest_News/2019/June/ai_and_the_next_generation_of_physicians | <div class="col-md-12">
<div class="col-md-5 pull-right" style="text-align: center;"><img alt="" src="/images/News/WSMA_MayJune-Reports-Cover-Article-Image-645x425px.jpg" class="pull-right" /></div>
<h5>June 21, 2019</h5>
<h2>AI and the Next Generation of Physicians</h2>
<p>
By Pat Curry</p>
<p>Seventeen thousand genes.</p>
<p>
Five driver events of cancer progression: gene expression, mutations, copy number, methylation, and known regulators of gene expression.
</p>
<p>
Fifty-three chemotherapeutic drugs that are FDA-approved and reasonably effective. Every tumor is different. Which drug will work best to fight each one?
</p>
<p>
"Our gut said the gene expression was affecting the drug sensitivity resistance...if we could figure out what those genes were," says Pamela Becker, MD, PhD, professor of medicine in the UW School of Medicine Division of Hematology and a member in the Clinical Research Division of the Fred Hutchinson Cancer Research Center.
</p>
<p>
But there were just too many variables to consider to achieve any degree of accuracy.
</p>
<p>
"To correlate 17,000 things with 50 things doesn't work," Dr. Becker says. "We needed another method; machine learning, also known as AI, is the method that can do that."
</p>
<p>
Dr. Becker is part of a research team at the University of Washington that is applying a combination of machine learning and big data to improve outcomes for cancer patients. The team—from UW Medicine, the Paul G. Allen School of Computer Science and Engineering, and the UW Department of Genome Sciences—developed MERGE, a new algorithm integrating five criteria impacting genes' contribution to cancer development. Using MERGE (short for "Mutation, Expression hubs, known Regulators, Genomic CNV, and mEthylation"), the team narrowed down the group of genes that would be sensitive or resistant to chemotherapeutic drugs from 17,000 to about three dozen.
</p>
<p>
"The higher or lower expression of those genes seems to be related to sensitivity of the entire class of agents," Dr. Becker says. "It reassured us we were on the right track. This was a huge first step to find the genes that seem to be related to drug sensitivity or resistance. That would help us figure out the best drugs for each patient."
</p>
<p>
The MERGE project is just one example of the ways Washington state's medical schools are using AI to advance health care for the next generation of physicians. This type of research is happening in every area of medicine, Dr. Becker says. Patient genetic profiles will be readily available to help customize treatment.
</p>
<p>
"Therapeutic information, drug dosing, this will get more and more sophisticated," she says. "For [medical] students, it's important to know that this is happening. Our tools will be better, more defined, and more able to recognize all these differences between people."
</p>
<h3>A tool for bringing specialties to rural areas</h3>
<p>
Robert Sorrells, PhD, associate dean for pre-clinical education and associate professor of anatomy at Pacific Northwest University of Health Sciences, was a cognitive science specialist for the University of Georgia's Artificial Intelligence Group before joining the faculty at PNWU. With a research focus on neural and computer modeling of knowledge representation and learning, Dr. Sorrells has written on artificial intelligence and the future of health care.
</p>
<p>
For PNWU, whose mission is to increase access to health care in rural areas, AI offers some intriguing possibilities for reaching rural areas that are underserved by specialties such as radiology and pathology.
</p>
<p>
"AI isn't going to replace those specialties," Sorrells says. "They aren't there in those rural areas. AI can be an interface between the physician and patient to deliver health care. When we can't get into rural areas, we can train the students to use the interface. Those are huge initiatives we want to equip our students for. With electronic medical records, algorithms can see patterns the primary care physician doesn't have the time or the bandwidth to put together. We want to train our students to use them before they're even available to the nascent population."
</p>
<p>
Today's medical students, who were born in the digital age and are extremely comfortable with technology, seem open and accepting of the new tools they will have at their disposal. Electronic stethoscopes can find a heart murmur that can't be heard by the human ear and provide a preliminary diagnosis.
</p>
<p>
Ultrasounds that connect to smartphones or tablets will be the "next huge thing in primary care," he says. "You can get this information without having to send a patient off to the city," he says. "When these tools hit the market, our students will already know how to use them."
</p>
<p>
Will AI change the coursework at medical school? Somewhat, Dr. Sorrells says, but only to the extent that it changes the tools.
</p>
<p>
"We still teach science," he says. "A doctor needs to be able to think through her science. When there isn't an iPhone or a stethoscope available, what kind of data do I need? We won't ever replace having very knowledgeable doctors who can reason through what can't be dealt with algorithmically."
</p>
<p>
One role of medical schools, he says, is to help address not only the technical advances provided by AI, but the ethical and cultural ones that physicians face every day.
</p>
<p>
"We need to train AI about cultural distinctions and teach them to respect and embrace those differences," he says.
</p>
<h3>An incubator for innovation</h3>
<p>The state's newest medical school, Washington State University's Elson S. Floyd College of Medicine, has taken the approach of creating a technology incubator, which is in the planning stages. The goal of the incubator will be to develop tools, technology, processes, and innovations to address health care problems in rural and underserved care and population health. It stands to reason that many of these initiatives will involve some aspect of AI.
</p>
<p>
The medical school also does a "wonderful job of providing an environment for students and faculty
to explore applications of emergent technology in all aspects of medicine," says Chris Martin, director of the medical school's simulation-based training.
</p>
<p>
Martin, whose department focuses on the use of virtual reality for medical training, notes that many of the devices his students use—Bluetooth-enabled blood pressure cuffs and glucometers, for example— download the results to a computer.
</p>
<p>
"We are finding is we have plenty of data," he says. "The vast amounts of data are difficult for a single person or a team to synthesize to understand what's going on with a patient or a disease process."
</p>
<p>
That, he says, is where AI can be extremely helpful.
</p>
<p>
"It won't replace the role of a physician, but will be able to augment their practice to provide better outcomes," he says. "It's just another example of how technology can better a practice rather than alter the trajectory of a profession. AI is great, but it still needs a teacher."
</p>
<p>
<em>This article was featured in the May/June 2019 issue of WSMA Reports, WSMA's print newsletter. WSMA Reports is a benefit of membership. Non-members may <a href="https://wsma.org/WSMA/News_Publications/Publications/WSMA/News_Publications/Publications/Publications.aspx?hkey=8f14076c-10b8-48ab-9595-8c256836e393">purchase a subscription</a>.</em>
</p>
<p>
<em>Pat Curry is WSMA Reports' senior editor.</em>
</p>
</div> | 6/21/2019 12:00:00 AM | 1/1/0001 12:00:00 AM |
| algorithms_as_partners_not_practitioners | Algorithms as Partners, Not Practitioners | Latest_News | Shared_Content/News/Latest_News/2019/June/algorithms_as_partners_not_practitioners | <div class="col-md-12">
<div class="col-md-5 pull-right" style="text-align: center;"><img alt="" src="/images/News/WSMA_MayJune-Reports-Cover-Article-Image-645x425px.jpg" class="pull-right" /></div>
<h5>June 21, 2019</h5>
<h2>Algorithms as Partners, Not Practitioners</h2>
<p><em>Members only, sign-in required</em></p>
<p>
By John Gallagher
</p>
<p>
Physicians can't escape the buzz around artificial intelligence in medicine, or the concerns that it raises: visions of robots in white coats practicing medicine by algorithm. If that's not bad enough, many suspect that AI might actually add to their workload instead of relieving it.
</p>
<p>
It's true that AI has the potential to change the way physicians practice. But that change can be for the better.
</p>
<p>
</p>
<p>
Properly done, AI could help with differential diagnoses, pick up on patterns that might otherwise escape detection, and identify roadblocks to better care. In fact, it is already doing these very things. To expand this achievement even further, the emphasis needs to be on how humans and computers can work
together to do better than either can do alone.
</p>
<p>
"You have to realize that it's a tool, just as a stethoscope is a tool," says Michael Anderson,
MD, chief medical officer at CHI Franciscan and a WSMA member. "There will never be a substitute for hands-on care of patients."
</p>
<p>
AI might be able to identify patterns in data to provide decision support for clinicians, says Lara Mangravite, president of Sage Bionetworks, a Seattle-based nonprofit focused on open practices for large-scale collaborative biomedical research. That would provide clinicians with more time to focus on "the human part of the job."
</p>
<h3>Harnessing big data</h3>
<p>
Measured by dollars alone, AI is a force to be reckoned with in health care. According to a December 2018 report from ReportLinker, investment in health care AI is expected to grow at an astonishing compounded growth rate of 50 percent between now and 2025, reaching more than $36 billion.
</p>
<p>
As with any emerging technology, AI has been the subject of sweeping rhetoric about its potential, both as a disruptor and a panacea. "Data science is really cool and a really hot topic," says Sean Mooney, PhD, chief research information officer of UW Medicine. "It is transforming fields."
</p>
<p>
Mooney acknowledges that as a phrase, AI is subject to shifting meanings. "Isn't it funny how the big- picture words are the hardest to define?" he notes. "I think of AI as learning something from data and using that to inform things." [For a glossary of terms, see facing page.]
</p>
<p>
Indeed, AI can help sort through reams of data to discover information that might otherwise be overlooked. WSMA President Thomas Schaaf, MD, a geriatrician and hospice care physician,
says that to him, the value of AI-type systems is the ability to see patterns that could potentially be missed by a human being in analyzing large data sets.
</p>
<p>
"It also brings the ability to suggest alternative diagnosis and treatments based on sufficient data about the patient and their case," he says.
</p>
<p>
Medicine offers no shortage of data from which to draw. But what AI relies upon is good data. That may be rarer than many think.
</p>
<p>
"We like to say, we have EHRs, we have big data, end of story," says David Haynor, MD, a neuroradiologist at UW Medicine, "but it's nowhere near that simple."
</p>
<p>
Indeed, getting data that is reliable and comparable is a challenge uniquely suited to the capabilities of AI systems. Haynor notes that a doctor may treat 1,000 patients for sore throat. "By the time he narrows it down, he's only got 152 patients with a reasonable amount of data: one lab test, a physical exam, and a follow up," he says. "That's not big data. That's tiny data."
</p>
<p>
Creating an algorithm for AI isn't necessarily difficult. Mooney recalls a colleague at Notre Dame who designed a predictive analysis on anonymized claims data using the kind of algorithm Netflix employs to suggest movies you might like.
</p>
<p>
Getting algorithms to be specific and sensitive is another matter. The patterns an algorithm picks out may be legitimate, but not important. Schaaf notes that one analysis determined that "the most reliable marker for a skin cancer lesion is the presence of a ruler next to the lesion."
</p>
<p>
That's why data scientists consider the input of physicians pivotal in creating AI that truly advances patient care.
</p>
<p>
"Data scientists may be using AI to solve problems that clinicians don't need solved; that's a problem that I think might become a real issue," Mangrative says. "Just because clinicians don't know how to manage a 50-terabyte file doesn't mean they aren't exquisitely talented about those things they do know about. We need to take both sets of expertise into account as we design systems."
</p>
<h3>Teaching AI to learn</h3>
<p>
Not surprisingly, imaging is a field ripe for AI, and radiology is one specialty where AI has already made significant inroads. Routine imaging, such as mammograms, creates a treasure trove of data that lends itself to computer analysis.
</p>
<p>
A recent study of the performance of AI algorithms in interpreting mammogram results showed that AI is fulfilling its promise—to a degree. "The results were as good or better as the average radiologist's," Haynor says, "but not as good as the best radiologist's."
</p>
<p>
However, algorithms can learn, just as people do. As algorithms have more data to work with, they can become more adept at flagging what truly matters with ever greater accuracy. This machine learning, as it is called, is key to advancing AI and creating systems that provide physicians with information they need to make decisions.
</p>
<p>
For example, Haynor cites the recommended chest CTs that smokers should get every few years. AI could correlate certain features, such as the size of a nodule, with past scans and determine if it warrants the radiologist's further attention.
</p>
<p>
"The ideal model would be something that functions as a radiologist's assistant," he says, "a tireless guy who looks at every study, working in the middle of the night and on Sundays, telling you, 'Look at this particular slice and tell me if that is of concern."
</p>
<h3>Involving physicians in the process</h3>
<p>
Even primary care is beginning to feel the impact of AI. Providence St. Joseph Health has created an AI-powered chatbot called Grace to make sure patients can navigate more smoothly through the system. Sunitra Mishra, MD, chief executive of the Express Care program at Providence, says her clinical team worked closely with the tech team to create the logic to determine when patients should be seen at one venue versus another.
</p>
<p>
As a primary care provider, Dr. Mishra says she can see the opportunities AI offers. It could potentially reduce some of the administrative burdens for physicians, giving them more time for patient care and reducing the risk of burnout.
</p>
<p>
Of course, to do so, AI needs to be integrated into the workflow in such a way that it does not make more work. "I'm like all doctors," Dr. Haynor says. "If it takes me more time, and it doesn't make my work easier, why should I do it?"
</p>
<p>
In addition, AI will have to earn the confidence of physicians through rigorous study of how it performs. "It's really important to invest in research around the performance of data science methods in practice and whether they achieve the outcomes we want them to achieve," Mooney says. If AI is able to fulfill all of these goals, it will prove extraordinarily useful in the practice of medicine.
</p>
<p>
What it won't do is supersede the physician-patient relationship.
</p>
<p>
"Whether a patient improves or not depends on whether patients trust their physician and feel listened to,"
says WSMA member Jessica Schlicher, MD, MBA, medical director for care transformation for CHI Franciscan. "Caring for our people with compassion, coupling the experience with clinical expertise—that's where the power of the physician is. I don't ever see a world where AI can replace that."
</p>
<p>
<em>This article was featured in the May/June 2019 issue of WSMA Reports, WSMA's print newsletter. WSMA Reports is a benefit of membership. Non-members may <a href="https://wsma.org/WSMA/News_Publications/Publications/WSMA/News_Publications/Publications/Publications.aspx?hkey=8f14076c-10b8-48ab-9595-8c256836e393">purchase a subscription</a>.</em>
</p>
<p>
<em>John Gallagher is a Washington-based freelance writer who specializes in covering health care and health care policy topics.</em>
</p>
</div> | 6/21/2019 12:00:00 AM | 1/1/0001 12:00:00 AM |
| weekly_rounds_june_17_2019_hard_won_change_is_worth_the_effort | Weekly Rounds: June 17, 2019 - Hard-won Change Is Worth the Effort | Latest_News | Shared_Content/News/Weekly_Rounds/2019/weekly_rounds_june_17_2019_hard_won_change_is_worth_the_effort | <div class="col-md-12">
<div class="col-md-5 pull-right" style="text-align: center;"><img alt="Weekly Rounds logo" class="pull-right" src="/images/Logos/Weekly_Rounds_Article_Image_645x425.jpg" /></div>
<h5>June&nbsp;17, 2019</h5>
<h2>Hard-won Change Is Worth the Effort</h2>
<p>
Jennifer Hanscom, Executive Director/CEO
</p>
<p>
Last week, I was in Chicago attending the AMA annual meeting. After all these years it's still a thrill to witness the incoming president's swearing in, but this year's oath of office felt like the most momentous one ever. In fact, it was an historic moment as the organization's first female African-American president put her hand on Bibles belonging to her great-grandfather and aunt and repeated the time-honored words of the oath.
</p>
<p>
As Patrice Harris, MD, stepped up to that podium (<a href="https://www.facebook.com/AmericanMedicalAssociation/videos/853283701704193/?notif_id=1559682244507710&amp;notif_t=live_video_schedule_broadcaster">view it here</a> at timestamp 48:45), she did so as only the fifth female president in the organization's 174 years. Notably, as she gave her presidential address, she did so 100 years after the first female physician was seated in the AMA House of Delegates. Also? Dr. Harris was preceded by another female president, Dr. Barbara McAneny, and she will be followed by Dr. Susan Bailey, who was elected AMA president-elect.
</p>
<p>
Seeing these brilliant women leading in medicine today was a powerful reminder to me that change takes time—sometimes longer than we wish, but progress comes nonetheless. And as the AMA House of Delegates debated numerous issues—including health system reform, hospital consolidation, and gender inequity in medicine—we saw that democratic progress in action.
</p>
<p>
The House narrowly voted to continue its opposition to a single-payer health system, adopting new policy instead to push for universal coverage by supporting improvements to the Affordable Care Act, including:
</p>
<ul>
<li>Eliminating the subsidy "cliff," thereby expanding eligibility for premium tax credits beyond 40 percent of the federal poverty level.</li>
<li>Increasing the generosity of premium tax credits.</li>
<li>Expanding eligibility of cost-sharing reductions.</li>
</ul>
<p>
Of particular interest to Washington state (in light of the WSMA's recent work to eliminate the personal and philosophical exemption to the MMR vaccine), the House passed a resolution to support state policies that would allow minors to override their parents' objections to vaccinations. In a separate resolution, the delegates directed the AMA to develop model legislation for mature minor consent to vaccinations.
</p>
<p>
Your WSMA delegates took several issues to the AMA including:
</p>
<ul>
<li>The concept of offering state employee health plans to every state resident and making the federal employee health benefits program health insurance plans available to everyone (referred for further study).</li>
<li>Alignment of federal privacy law and regulations with HIPAA for the purposes of treatment, payment, and health operations, while ensuring protections are in place against the use of "Part 2" substance use disorder records in criminal proceedings (amended to reference applicable state laws).</li>
<li>Universal access for essential public health services (alternative language adopted).</li>
<li>An AMA study and report on the conditions under which our country could successfully eliminate the manufacture, distribution, and sale of combustible cigarettes and other combustible tobacco products (referred for decision).</li>
</ul>
<p>
Here's a shout-out and special thanks to WSMA delegates and alternates to the AMA, for their passion and concern for the future of the profession. The delegation includes:
</p>
<ul>
<li>Rod Trytko, MD, MBA, MPH, chair of the delegation</li>
<li>Matthew Grierson, MD, vice chair</li>
<li>Peter Dunbar, MD</li>
<li>Nariman Heshmati, MD</li>
<li>Erin Harnish, MD</li>
<li>Shane Macaulay, MD</li>
<li>Beth Peterson, MD</li>
<li>Sheila Rege, MD</li>
</ul>
<p>
Policy-making matters not only to the profession, but to patients. And as Dr. Harris noted in her keynote address, "There is strength in our collective voice." I couldn't agree more. If you're interested in proposing policy or serving as a delegate at the 2019 Annual Meeting of the WSMA House of Delegates meeting (Oct. 12-13 at the Hilton Seattle Airport &amp; Conference Center), here's some helpful information you'll need to know.
</p>
<p>
Resolutions, one of the key policy drivers for the association, are considered by the House of Delegates at their annual meeting each fall. Learn more about <a href="https://wsma.org/WSMA/Events/Annual_Meeting/How_to_Write_a_Resolution/WSMA/Events/Annual_Meeting/How_to_Write_a_Resolution/How_to_Write_a_Resolution.aspx?hkey=a22cd977-aaed-4445-9ee2-6fc4716a7136">how to write a resolution</a> and take note of these dates:
</p>
<ul>
<li>Aug. 16: Deadline to submit your resolution for publication in the delegate handbook, the compilation of resolutions and other business to be considered by the House of Delegates. </li>
<li>Aug. 30: Delegate handbook will be available for download from the WSMA website. </li>
<li>Sept. 13: Final deadline to submit resolutions. To be considered, resolutions received after this date require consent of two-thirds of the House at the opening session.</li>
</ul>
<p>
You can also circulate your ideas and engage in conversation before the meeting by posting your thoughts on WSMA's password-protected, members-only <a href="https://wsma.org/WSMA/Membership/Discussion_Forums/Reference_Committees_and_General_Discussion/WSMA/Membership/Discussion_Forums/Reference_Committees_and_General_Forum.aspx?hkey=d674a5ae-4fb5-48d6-b969-16aab9b63647">online discussion forum</a>.
</p>
<p>
The House of Delegates is composed of WSMA members who represent, and are designated by, their respective county or state specialty society, as well as representatives of WSMA's special sections and board of trustees. If you would like to serve as a delegate at the 2019 WSMA Annual Meeting, contact your local county or state specialty society.
</p>
<p>
Getting involved at this level means bringing influence and insight to the policy-making process of your state medical association. It means being a leader in your profession and making a difference in the lives and health of your patients. As Dr. Harris said: "We don't run away from problems ... physicians run towards them!"
</p>
<p>
Won't you join me in running toward the problems, and help me affect change for the better in our profession in Washington?
</p>
</div> | 6/17/2019 12:00:00 AM | 1/1/0001 12:00:00 AM |