Research results call into question the safety and dependability of using artificial intelligence in medical diagnosis, a development that should be watched by clinical laboratory scientists
ChatGPT, an artificial intelligence (AI) chatbot that returns answers to written prompts, has been tested and found wanting by researchers at the University of Florida College of Medicine (UF Health) who looked into how well it could answer typical patient questions on urology. Not good enough according to the researchers who conducted the study.
AI is quickly becoming a powerful new tool in diagnosis and medical research. Some digital pathologists and radiologists use it for data analysis and to speed up diagnostic modality readings. It’s even been said that AI will improve how physicians treat disease. But with all new discoveries there comes controversy, and that’s certainly the case with AI in healthcare.
Many voices in opposition to AI’s use in clinical medicine claim the technology is too new and cannot be trusted with patients’ health. Now, UF Health’s study seems to have confirmed that belief—at least with ChatGPT.
The study revealed that answers ChatGPT provided “fell short of the standard expected of physicians,” according to a UF Health new release, which called ChatGPT’s answers “flawed.”
The questions posed were considered to be common medical questions that patients would ask during a visit to a urologist.
The researchers believes their study is the first of its kind to focus on AI and the urology specialty and which “highlights the risk of asking AI engines for medical information even as they grow in accuracy and conversational ability,” UF Health noted in the news release.
“I am not discouraging people from using chatbots,” said Russell S. Terry, MD (above), an assistant professor in the UF College of Medicine’s department of urology and the study’s senior author, in a UF Health news release. “But don’t treat what you see as the final answer. Chatbots are not a substitute for a doctor.” Pathologists and clinical laboratory managers will want to monitor how developers improve the performance of chatbots and other applications using artificial intelligence. (Photo copyright: University of Florida.)
UF Health ChatGPT Study Details
UF Health’s study featured 13 of the most queried topics from patients to their urologists during office visits. The researchers asked ChatGPT each question three times “since ChatGPT can formulate different answers to identical queries,” they noted in the news release.
The urological conditions the questions covered included:
The researchers then “evaluated the answers based on guidelines produced by the three leading professional groups for urologists in the United States, Canada, and Europe, including the American Urological Association (URA). Five UF Health urologists independently assessed the appropriateness of the chatbot’s answers using standardized methods,” UF Health noted.
Notable was that many of the results were inaccurate. According to UF Health, only 60% of responses were deemed appropriate from the 39 evaluated responses. Outside of those results, the researchers noted in their Urology paper, “[ChatGPT] misinterprets clinical care guidelines, dismisses important contextual information, conceals its sources, and provides inappropriate references.”
When asked, for the most part ChatGPT was not able to accurately provide the sources it referenced for its answers. Apparently, the chatbot was not programmed to provide such sources, the UF Health news release stated.
“It provided sources that were either completely made up or completely irrelevant,” Terry noted in the new release. “Transparency is important so patients can assess what they’re being told.”
Further, “Only 7 (54%) of 13 topics and 21 (54%) of 39 responses met the BD [Brief DISCERN] cut-off score of ≥16 to denote good-quality content,” the researchers wrote in their paper. BD is a validated healthcare information assessment questionnaire that “provides users with a valid and reliable way of assessing the quality of written information on treatment choices for a health problem,” according to the DISCERN website.
ChatGPT often “omitted key details or incorrectly processed their meaning, as it did by not recognizing the importance of pain from scar tissue in Peyronie’s disease. As a result … the AI provided an improper treatment recommendation,” the UF Health study paper noted.
Is Using ChatGPT for Medical Advice Dangerous to Patients?
Terry noted that the chatbot performed better in some areas over others, such as infertility, overactive bladder, and hypogonadism. However, frequently recurring UTIs in women was one topic of questions for which ChatGPT consistently gave incorrect results.
“One of the more dangerous characteristics of chatbots is that they can answer a patient’s inquiry with all the confidence of a veteran physician, even when completely wrong,” UF Health reported.
“In only one of the evaluated responses did the AI note it ‘cannot give medical advice’ … The chatbot recommended consulting with a doctor or medical adviser in only 62% of its responses,” UF Health noted.
For their part, ChatGPT’s developers “tell users the chatbot can provide bad information and warn users after logging in that ChatGPT ‘is not intended to give advice,’” UF Health added.
Future of Chatbots in Healthcare
In UF Health’s Urology paper, the researchers state, “Chatbot models hold great promise, but users should be cautious when interpreting healthcare-related advice from existing AI models. Additional training and modifications are needed before these AI models will be ready for reliable use by patients and providers.”
UF Health conducted its study in February 2023. Thus, the news release points out, results could be different now due to ChatGPT updates. Nevertheless, Terry urges users to get second opinions from their doctors.
“It’s always a good thing when patients take ownership of their healthcare and do research to get information on their own,” he said in the news release. “But just as when you use Google, don’t accept anything at face value without checking with your healthcare provider.”
That’s always good advice. Still, UF Health notes that “While this and other chatbots warn users that the programs are a work in progress, physicians believe some people will undoubtedly still rely on them.” Time will tell whether trusting AI for medical advice turns out well for those patients.
The study reported above is a useful warning to clinical laboratory managers and pathologists that current technologies used in ChatGPT, and similar AI-powered solutions, have not yet achieved the accuracy and reliability of trained medical diagnosticians when answering common questions about different health conditions asked by patients.
It’s not just radiology. Gen Z residents will be matching in pathology and other specialties, and that means clinical laboratories should be ready to adapt their recruiting and training to Gen Z’s unique characteristics
It’s a big event in medical schools across the nation when it is time for residency programs to match residency candidates with first-year and second-year post-graduate training positions. But this year has a special twist because—for example in radiology—this is the first class of Generation Z (Gen Z) residency candidates to be matched with radiology residency programs.
In their abstract, the authors wrote, “This year, the radiology community will experience the beginning of a generational change by matching its first class of Generation Z residents. To best welcome and embrace the changing face of the radiology workforce, this Viewpoint highlights the values that this next generation will bring, how radiologists can improve the way they teach the next generation, and the positive impact that Generation Z will have on the specialty and the way radiologists care for patients.”
Members of Gen Z are now entering the workforce in large numbers. To recruit high-quality candidates from this generation, healthcare employers—including clinical laboratories and pathology practices—may have to adapt the way they interact with and train these individuals.
Gen Z is generally described as individuals who were born between 1995 and 2012. Also known as “Zoomers,” the demographic comprises approximately 25% of the current population of the United States. They are extremely diverse, tend to be very socially conscious, and can easily adapt to rapid changes in communications and education, according to the AJR paper.
Although the paper deals with radiology, this type of information can also be valuable to clinical laboratories as Gen Z pathologists are poised to enter clinical practice in growing numbers. This marks the beginning of the professional laboratory careers of Zoomers, while Millennials move up into higher levels of lab management, the oldest Gen Xers near retirement age, and Baby Boomers retire out of the profession.
“Gen Z employees bring unique values, expectations, and perspectives to their jobs,” said Paul McDonald (above), Senior Executive Director at staffing firm Robert Half in a news release. “They’ve grown up in economically turbulent times, and many of their characteristics and motivations reflect that.” Thus, clinical laboratories may have to develop methods for recruiting and training Gen Z staff that match the unique characteristics of Gen Z candidates. (Photo copyright: LinkedIn.)
Zoomers Like Digital and Artificial Intelligence Technology
One of the most unique aspects of Gen Z is that they have never lived in a world without the Internet and have little memory of life without smartphones. Zoomers grew up totally immersed in digital technology and tend to be comfortable using digital tools in their everyday life and in the workplace. They lean towards being very open to artificial intelligence (AI) and how it can assist humans in analysis and diagnostic methods.
“This group of professionals has grown up with technology available to them around the clock and is accustomed to constant learning,” said Paul McDonald, Senior Executive Director at staffing firm Robert Half in a news release. “Companies with a solid understanding of this generation’s values and preferences will be well prepared to create work environments that attract a new generation of employees and maximize their potential.”
According to the AJR paper, Zoomers learn best by doing, so employers should concentrate on interactive learning opportunities, such as simulations, virtual reality, and case-based methods for teaching the aspects of the job. They are likely to expect digital and blended resources as well as traditional approaches to learning their new job responsibilities.
The paper goes on to state that Gen Z members value diversity, equity, inclusivity, sustainability, civic engagement, and organizational transparency. Their social consciousness and diversity may yield a greater range of perspectives and problem-solving approaches which may bolster their sensitivity to patient-centered care.
“The oldest in Gen Z have already seen a recession and a war on terrorism. They’ve seen politics at its worst. And now they’ve seen a global pandemic and are about to see recession again,” said David Stillman, founder of GenGuru, a boutique management consulting firm that provides insights on how best to connect with Baby Boomers, Generation X, Millennials, and Gen Z, in an interview with the Society for Human Resource Management (SHRM). “They are survivors,” he added.
According to the SHRM, “Stillman says Millennials, who preceded Generation Z, were coddled by their parents. He maintains that Generation Z’s parents were more truthful, telling their offspring, ‘You’re going to have a really tough time out there, you have to work super hard,’ which he says created ‘the most competitive generation in the workforce since the Baby Boomers.’”
Gen Z Wants More than a Paycheck, They Want Purpose
The American Journal of Roentgenology paper also states that Gen Z members grew up in a rapidly changing world and tend to be resilient, adaptable, and flexible. They have experienced and witnessed many stressors and navigate these issues by focusing on mental health and a meaningful work-life balance. With respect to a profession, they are searching for more than just a paycheck, and they want a purposeful career where they feel a sense of belonging.
Increase information sharing and transparency to help alleviate fear and anxiety.
Incentivize them by showing them clear paths to career progression.
Make sure they know how their individual contributions matter to the organization.
Motivate them by giving them room for autonomy and experimentation.
Provide specific and constructive feedback.
Harness community and in-person interactions to boost professional collaborations.
Prioritize wellness and mental health.
“Be prepared to spend time with them face to face,” McDonald stated. “They want to be mentored and coached. If you coach them, you’re going to retain them.”
Preparing to Attract Gen Z to Clinical Laboratories
As Generation Z comes of age, more of them will be working in the medical professions. Clinical laboratories and anatomic pathology groups would be well advised to prepare their businesses by adjusting leadership, adapting recruiting efforts, and shifting marketing to attract Zoomers and remain relevant and successful in the future.
Although sweeping statements about individual generations may be limiting, understanding their unique insights, values, and backgrounds can be helpful in the workplace. With a large amount of Gen Z workers now transitioning from college into careers, it will be beneficial for clinical laboratory managers to recognize their unique characteristics to recruit and maintain talented workers more effectively.
Google designed the suite to ease radiologists’ workload and enable easy and secure sharing of critical medical imaging; technology may eventually be adapted to pathologists’ workflow
Clinical laboratory and pathology group leaders know that Google is doing extensive research and development in the field of cancer diagnostics. For several years, the Silicon Valley giant has been focused on digital imaging and the use of artificial intelligence (AI) algorithms and machine learning to detect cancer.
Now, Google Cloud has announced it is launching a new medical imaging suite for radiologists that is aimed at making healthcare data for the diagnosis and care of cancer patients more accessible. The new suite “promises to make medical imaging data more interoperable and useful by leveraging artificial intelligence,” according to MedCity News.
In a press release, medical technology company Hologic, and healthcare provider Hackensack Meridian Health in New Jersey, announced they were the first customers to use Google Cloud’s new suite of medical imaging products.
“Hackensack Meridian Health has begun using it to detect metastasis in prostate cancer patients earlier, and Hologic is using it to strengthen its diagnostic platform that screens women for cervical cancer,” MedCity News reported.
“Google pioneered the use of AI and computer vision in Google Photos, Google Image Search, and Google Lens, and now we’re making our imaging expertise, tools, and technologies available for healthcare and life sciences enterprises,” said Alissa Hsu Lynch (above), Global Lead of Google Cloud’s MedTech Strategy and Solutions, in a press release. “Our Medical Imaging Suite shows what’s possible when tech and healthcare companies come together.” Clinical laboratory companies may find Google’s Medical Imaging Suite worth investigating. (Photo copyright: Influencive.)
.
Easing the Burden on Radiologists
Clinical laboratory leaders and pathologists know that laboratory data drives most healthcare decision-making. And medical images make up 90% of all healthcare data, noted an article in Proceedings of the IEEE (Institute of Electrical and Electronics Engineers).
More importantly, medical images are growing in size and complexity. So, radiologists and medical researchers need a way to quickly interpret them and keep up with the increased workload, Google Cloud noted.
“The size and complexity of these images is huge, and, often, images stay sitting in data siloes across an organization,” said Alissa Hsu Lynch, Global Lead, MedTech Strategy and Solutions at Google, told MedCity News. “In order to make imaging data useful for AI, we have to address interoperability and standardization. This suite is designed to help healthcare organizations accelerate the development of AI so that they can enable faster, more accurate diagnosis and ease the burden for radiologists,” she added.
According to the press release, Google Cloud’s Medical Imaging Suite features include:
Imaging Storage: Easy and secure data exchange using the international DICOM (digital imaging and communications in medicine) standard for imaging. A fully managed, highly scalable, enterprise-grade development environment that includes automated DICOM de-identification. Seamless cloud data management via a cloud-native enterprise imaging PACS (picture archiving and communication system) in clinical use by radiologists.
Imaging Lab: AI-assisted annotation tools that help automate the highly manual and repetitive task of labeling medical images, and Google Cloud native integration with any DICOMweb viewer.
Imaging Datasets and Dashboards: Ability to view and search petabytes of imaging data to perform advanced analytics and create training datasets with zero operational overhead.
Imaging AI Pipelines: Accelerated development of AI pipelines to build scalable machine learning models, with 80% fewer lines of code required for custom modeling.
Imaging Deployment: Flexible options for cloud, on-prem (on-premises software), or edge deployment to allow organizations to meet diverse sovereignty, data security, and privacy requirements—while providing centralized management and policy enforcement with Google Distributed Cloud.
First Customers Deploy Suite
Hackensack Meridian Health hopes Google’s imaging suite will, eventually, enable the healthcare provider to predict factors affecting variance in prostate cancer outcomes.
“We are working toward building AI capabilities that will support image-based clinical diagnosis across a range of imaging and be an integral part of our clinical workflow,” said Sameer Sethi, Senior Vice President and Chief Data and Analytics Officer at Hackensack, in a news release.
The New Jersey healthcare network said in a statement that its work with Google Cloud includes use of AI and machine learning to enable notification of newborn congenital disorders and to predict sepsis risk in real-time.
Hologic, a medical technology company focused on women’s health, said its collaboration integrates Google Cloud AI with the company’s Genius Digital Diagnostics System.
“By complementing our expertise in diagnostics and AI with Google Cloud’s expertise in AI, we’re evolving our market-leading technologies to improve laboratory performance, healthcare provider decision making, and patient care,” said Michael Quick, Vice President of Research and Development and Innovation at Hologic, in the press release.
Hologic says its Genius Digital Diagnostics System combines AI with volumetric medical imaging to find pre-cancerous lesions and cancer cells. From a Pap test digital image, the system narrows “tens of thousands of cells down to an AI-generated gallery of the most diagnostically relevant,” according to the company website.
Hologic plans to work with Google Cloud on storage and “to improve diagnostic accuracy for those cancer images,” Hsu Lynch told MedCity News.
Medical image storage and sharing technologies like Google Cloud’s Medical Imaging Suite provide an opportunity for radiologists, researchers, and others to share critical image studies with anatomic pathologists and physicians providing care to cancer patients.
One key observation is that the primary function of this service that Google has begun to deploy is to aid in radiology workflow and productivity, and to improve the accuracy of cancer diagnoses by radiologists. Meanwhile, Google continues to employ pathologists within its medical imaging research and development teams.
Assuming that the first radiologists find the Google suite of tools effective in support of patient care, it may not be too long before Google moves to introduce an imaging suite of tools designed to aid the workflow of surgical pathologists as well.
New study analyzes the dramatic decline in the utilization of imaging diagnostics between 2008 and 2014 and suggests that reductions in imaging use could be the result of changes in federal policy, increased deductibles, and cost-cutting focuses
Anatomic pathologists have experienced sustained cuts to reimbursements for both technical component and professional component services during the past eight to 10 years. But what has not happened to pathology is a 33% decline in the volume of biopsies referred to diagnosis. Yet that is what some studies say has happened to imaging reimbursement since 2006.
Using Medicare data for Part B imaging procedures covering the years 2001 to 2014, researchers at a major university identified that, beginning in 2006, the total reimbursement for imaging procedures declined at a steady rate throughout the following eight years covered by the study. It is unclear what implications the finding of this study of imaging utilization might predict for the utilization of advance anatomic pathology services.
Routine Use of Imaging in Diagnostics is Slowing Down
The researchers calculated utilization rates for “advanced” imaging modalities and component relative value unit (RVU) rates for all imaging modalities. They determined that trends in imaging rates and RVU rates rose between 2000 and 2008, but then sharply declined from 2008 to 2014. The researchers theorized that the reduction might have been due to changes in federal policy, increasing deductibles, and focus on cost-cutting by hospitals and healthcare providers.
Levin, along with Thomas Jefferson University associates Vijay M. Rao, MD, FACR, current Chair of Radiology, and Laurence Parker, PhD, Associate Professor of Radiology; and University of Wisconsin-Madison statistics Professor Charles D. Palit, PhD, argue that the decrease in imaging orders might reduce diagnostic costs, but also could negatively impact surgical pathologists, radiologists, medical researchers, and patients themselves.
In a Modern Healthcare article, Levin states that the reduction in utilization of imaging and radiology could be a slippery slope leading to decreased access to life-saving diagnostic tools that could leave patients “not getting the scans they probably need.”
What’s Fueling the Multi-Year Decline in Utilization of Imaging and Radiology?
Using data acquired from Medicare part B databases, the authors reported that total reimbursements for NDI peaked at $11.9 billion in 2006, but saw a steep decline of 33% to just over $8 billion in 2015. They attribute some of this decline as a result of the Deficit Reduction Act of 2005, which went into effect in 2007, as well as other cuts to NDI reimbursement funding. Reimbursement to radiologists, according to Levin et al, dropped by more than 19.5%, and reimbursement to cardiologists dropped nearly 45% between 2006 and 2015.
Surgical pathologists may see parallels in the total reimbursement for imaging during the years 2002-2015 compared to pathology technical component and professional component reimbursement during those same years. Taken from the Thomas Jefferson University study, the graphic above shows “total Part B payments for non-invasive diagnostic imaging to all physicians under the Medicare Physician Fee Schedule, 2002 to 2015. Vertical axis shows billions of dollars. The abrupt decline in 2007 was due to the Deficit Reduction Act. The declines in 2009, 2010, and 2011 were due largely to code bundling in, respectively, transthoracic echocardiography, radionuclide myocardial perfusion imaging, and CT of the abdomen and pelvis.” (Caption and image copyright: Thomas Jefferson University.)
According to Levin and Rao, the Choosing Wisely initiative was intended “to reduce the use of tests and treatments that were felt to be overused or often unnecessary.” Imaging examinations were included in the list of tests that were deemed to be “of limited value” in many situations. Levin and Rao suggested that there might have been a need to curtail testing pushed by payers, policymakers, and physicians at the time, but that the Choosing Wisely initiative could have added to a decline in imaging testing spurred on by the confusion physicians felt when attempting to access unclear scenarios and recommendations for the 124 imaging tests listed.
Imaging Decline Could Have Unintended Consequences for Providers and Patients
In a Radiology Business article, Levin outlined some of the unintended consequences facing healthcare due to the reduction in imaging utilization. He states that “private imaging facilities are starting to close down” and “MRI and other advanced imaging exams are beginning to shift into hospital outpatient facilities.” He predicts that the shift from private facilities to hospital facilities could cause imaging costs to increase for customers and healthcare providers.
Levin suggests that Medicare could “raise the fees a little and make the private offices a little more viable.” The profit margins, Levin argues, “are so low right now that you basically can’t run a business.” Medicare as a program might be seeing huge savings, Levin notes in several articles, but physicians, laboratories, and patients are feeling the pinch as a result.
In an interview with Physicians Practice, Rao echoed Levin’s concerns. “Policy makers lack understanding of the value of imaging and spectrum of the services provided by radiologists,” he declared. “On an institutional level, under the new payment models, radiology is transitioning to a cost center and radiologists often don’t have a seat at the table.”
Rao points out that this devaluing of radiologists’ work affects not only healthcare facilities, but patients themselves. Radiologists provide “major contributions to patient care by making accurate diagnoses, and doing minimally invasive treatments given many technological advances leading to appropriate management and improved outcomes,” he argues. How long before Pathology follows a similar track?
Balancing Cost and Quality in Testing Without Sacrificing Patient Needs
The fear seems to be that the push to lower costs by eliminating unnecessary imaging is inhibiting radiologists and diagnosticians from providing necessary imaging for patients. And that delaying diagnoses affects the ability of healthcare providers to provide adequate and timely patient care. Rao suggests, however, that physicians’ use of medical imaging could simply be evolving.
“There were other factors that also helped limit the rapid growth, such as greater attention by physicians to practice guidelines, concerns about radiation exposure to patients, and the Great Recession of 2007 to 2009,” Rao noted in a Thomas Jefferson University news release. “However, we expect that additional changes, such as the advent of lung cancer and other screening programs, and the use of computerized clinical decision support, will continue to promote and support appropriate use of imaging technology.”
The drive to reduce healthcare expenditures should not be dismissed. We may soon see parallels in the rise and fall of imaging utilization for genetic testing, surgical pathology, and other new and expensive clinical laboratory technologies as policymakers attempt to balance increased spending against the clinical value of these diagnostic tools.
Big data offers new opportunities for healthcare providers, clinical laboratories, and pathology groups, and this new alliance hopes to accelerate big data capabilities
Big data has the potential to deliver unprecedented insight into optimizing the patient care experience and managing outcomes for healthcare providers. That is particularly true for clinical laboratories, and pathology groups. Yet, with the sheer amount of data generated by today’s ever-expanding menus of diagnostic procedures, communicating this data between systems and analyzing data at high-levels still presents challenges.
To help healthcare organizations jumpstart their Big Data programs, key stakeholders are joining forces. One such alliance involves Siemens Healthineers and IBM Watson Health. In an October 2016 press release, the two organizations announced a five-year global strategic alliance aimed at helping healthcare professionals optimize value-based care that leverages increasingly complex data collected for use in precision medicine.
What should intrigue pathologists and medical laboratory managers about this new alliance is the fact that Siemens Healthineers owns two of the world’s largest businesses in radiology/imaging and in vitro (IVD). Thus, it can be expected that the alliance will be looking to identify ways to combine radiology data with clinical laboratory data that produce knowledge that can be applied to clinical care. (more…)
