Based on an article by Myrna Traylor from the.
The healthcare sector is on the cusp of a technological revolution, primarily driven by advancements in artificial intelligence (AI). From automating routine tasks to assisting in complex diagnostics, AI is poised to reshape the clinical landscape.
While the current state-of-the-art technology is not encroaching significantly on the roles and responsibilities of medical assistants, phlebotomists and medical laboratory professionals, change is happening rapidly as more AI-based products enter the workplace.
Common AI Tools
AI is a broad term describing a lot of new technology in the healthcare space. There are many different applications of AI, and different features can be categorized by the way they work. For instance, generative AI, which got popular because of the buzz around ChatGPT and other large language models, can synthesize data and present easy-to-understand responses to questions. This technology can be used to convert complex medical text into simpler, eighth-grade reading level communications.
“Machine learning” and “deep learning” describe technology that scours large data sets to apply algorithms or certain parameters to create a desired result. This is particularly valuable for reviewing radiology images—these models have proven to be very accurate in discerning dangerous lesions and performing other clinically relevant tasks that may be difficult or tedious for human reviewers.
An offshoot of deep learning is “natural language processing,” or NLP. Once a program has been trained on medical terminology, clinical documentation and published research, it can help clinicians transcribe notes from patient visits and generate reports. When this information is incorporated into a patient’s electronic health record (EHR), it can speed up this administrative task and give practitioners more time to spend with patients.
Each of these formats is being used or developed for use in medical settings, from the research bench to the bedside.
How Professionals Are Using AI
AI adoption in healthcare is accelerating, with clinics integrating various AI-powered tools to enhance patient care and operational efficiency. Hospitals like Rush University Medical Center and Children’s Hospital Los Angeles are already leveraging AI to streamline workflows and improve patient outcomes. These institutions are early adopters, setting benchmarks for others to follow.
There are several main areas in which AI and/or its companion tools are being used, such as:
- Diagnostics: AI algorithms analyze medical images, such as X-rays and MRIs, with high accuracy, aiding radiologists in early disease detection.
- Predictive analytics: By analyzing patient data, AI can predict disease outbreaks and patient deterioration, or predict staffing needs for emergency departments, allowing for proactive interventions.
- Personalized treatment: AI algorithms tailor treatment plans based on individual patient profiles, improving outcomes.
- Administration: Automating administrative tasks like scheduling and billing frees up healthcare staff to focus on patient care.
Michele H. Johnson, MD, professor of radiology and biomedical imaging, neurosurgery and surgery, and director of interventional neuroradiology at Yale School of Medicine, has had experience with several AI companies who are providing interpretive and noninterpretive solutions for radiologists. In the former case, she describes how AI has assisted in triage by reviewing radiological images and prioritizing those which harbor critical findings, facilitating the care needs of patients in the emergency department. “The AI algorithm could look at an X-ray and screen it for certain things like blood clots, stroke or fracture,” says Johnson. “If it identifies that in the images, then it can put that [patient] higher up in the work list, so the radiologist can get to something that AI identified as potentially problematic faster.”
The most important thing AI is helping with right now is workflow, Johnson says. “For example, if you know statistically what times of day the surges occur in the emergency room, then you can make sure that you have the necessary personnel there.”
For the Record
Another AI application that is getting a lot of attention is NLP tools, which can assist practitioners with exam notes, EHR (EMR) reviews and automatic reporting. Johnson co-authored a 2022 paper that described how “NLP-based applications can mine the EMR to synthesize and provide relevant patient history and lab data in a single screen at the time of dictation. Automated identification of actionable findings, in concert with radiology information systems, can automatically report incidental findings, assist with physician notification and track the completion status of appropriate follow-up.”
There are now hundreds of such products being developed and trialed in clinics and hospitals. One service, Nabla, is using ambient voice recognition to generate clinical notes from clinician-patient conversations. “Nabla was designed to cater to health systems, medium-sized clinics and small practices,” says Kenza Bouzoubaa, brand and communications manager at Nabla. The program transcribes notes into the EHR, where the clinician can double-check them for accuracy.
Nabla supports more than 50 specialties, from internal medicine to orthopedics to oncology. It also has the ability to train to recognize accents or nonstandard English. “Our transcription model was specifically trained to recognize 50-plus different accents because we made sure the data set we used to train our models included audio recordings featuring various accents,” Bouzoubaa says.
The Big Picture
AI’s integration is bringing numerous benefits to healthcare professionals, but there are also educational, monitoring and verification challenges that need to be addressed.
Possible Benefits
- Efficiency: Automating routine tasks reduces people’s workload, allowing staff to concentrate on more critical activities.
- Accuracy: AI can minimize human errors in diagnostics and treatment, ensuring higher accuracy in patient care.
- Timesaving: AI-powered tools expedite processes like patient data analysis and record-keeping, saving valuable time for medical staff.
Limitations and Challenges
- Bias: AI systems can exhibit biases based on the data they are trained on, leading to disparities in care. Such disparities already exist in patient care, and AI, if not properly managed or designed, could make them worse.
- Cost: The initial investment in AI technology can be high, posing a barrier for smaller clinics.
- Training: Healthcare professionals need training to effectively use AI tools, requiring time and resources.
- Ethical issues: The ethical implications of AI, including patient privacy and data security, are critical considerations that need addressing.
Potential Future Uses
- Robotic surgery: AI can assist in precision surgeries, enhancing outcomes and reducing recovery times.
- Virtual health assistants: AI-enabled virtual assistants can provide round-the-clock patient support and monitor chronic conditions.
- Drug discovery: AI can accelerate the drug discovery process by predicting molecular behavior and potential side effects.
Programs like this might one day be able to accurately assign current procedure terminology (CPT) codes to patient interaction records, streamlining invoicing and reimbursements. This is a responsibility that some medical assistants may currently have—but many will be happy to leave it to AI if it means they can concentrate on other, more patient-focused tasks. On the other hand, medical assistants who are interested in data management and analysis may have an opportunity for professional advancement to use (or acquire) the skills to manage AI systems.
Concerns Among Healthcare Professionals
While AI offers numerous advantages, it also raises potential issues. One is the so-called “black box” problem: Clinicians are wary about trusting a process that cannot demonstrate how it is working or the source data it is relying upon. This natural reluctance is especially important in cases where diagnosing an illness could be a life-and-death matter for a patient. Even though certain applications have been in use for five years or more, clinicians are still in a “testing” phase with many of these products, especially as offices or hospitals try new systems.
Medical professionals can, however, rely on approvals from the U.S. Food and Drug Administration (FDA) for any AI program or system that is offered for sale. The FDA regularly checks that such programs have demonstrated they are safe and effective in relation to manufacturers’ claims. According to one source, the FDA has approved more than 600 AI-based and machine-learning-based medical devices since 1995. But FDA approval is not required for any “homegrown” AI, such as that in development at medical centers at places like Duke, the Mayo Clinic and Stanford.
Allied health professionals may be able to carve out an important niche in the AI-enabled landscape of healthcare, acting as liaisons between patients and AI. With more personalized care plans arising out of AI and wearable tech, allied healthcare workers may actually be the ones to provide the behavioral support that will help patients follow through and achieve the best health outcomes. After all, no matter how the care is crafted, it is only good if it helps people get better.