The advent of generative artificial intelligence (AI) and natural language processing (NLP) tools, such as ChatGPT, have put the power of AI into the palms of our hands. AI and NLP have hit the mainstream and are all the rage both within and outside of medicine. Doctors and hospitals have begun exploring ways that AI and NLP can help solve some of the biggest challenges in health care: reducing physician burnout, improving physician efficiency and decreasing administrative burdens.
As AI continues to mature, its potential impact on ophthalmological care becomes increasingly evident, from its ability to enhance diagnostic accuracy and optimize treatment plans to improve surgical efficiency and safety. However, AI also stands to make major positive impacts on clinical workflows in the office.
Physicians spend countless hours using their EMR systems to review clinical data, draft patient notes, communicate with colleagues and send letters to patients and insurance companies. Unfortunately, the advent of EMRs has led to a decline in face-to-face time between patients and physicians as well as a rising tide of burnout in the profession of medicine.
AI-enabled workflows hold the potential to reinvigorate the doctor-patient relationship by offloading and automating administrative tasks, drafting clinical notes, responding to patient inbox messages and so much more.
The future of AI in ophthalmology will go far beyond disease diagnostics. Using tools like ChatGPT and NLP, physicians will be put back in the driver seat of health care and will be allowed to practice medicine the way we were trained in medical school — face-to-face with our patients. Here are the ways physicians will benefit from these tools as well as some of the challenges and factors to consider.
REDEFINING COMMUNICATION AND DOCUMENTATION
NLP has revolutionized the way we interact with technology by enabling AI-based machines to understand and respond to human language in a conversational tone. This opens up the door for NLP to automate or improve any language-based task in the clinic. In the context of ophthalmology, NLP and advanced models like ChatGPT will play a pivotal role in several key areas:
Clinical notes
Clinical documentation is an expensive and time-intensive endeavor. In many offices, clinical note writing has been offloaded to a well-trained scribe, though the cost of such a full-time staff member’s salary is large. Doctors without scribes are forced to eat up precious clinic time to complete clinical notes during the day, or else take their work home and complete notes in the evening during “pajama time.” Either way, the cost on the practice, on patient throughput and on the physician’s professional and personal lives is high.
The burden of clinical documentation can be alleviated through the integration of ChatGPT and NLP. These models can assist ophthalmologists in generating comprehensive and accurate clinical notes, promoting standardized and efficient record-keeping. By having computers “listen in” to the clinical encounter, documentation can precisely match the discussions physicians have with their patients in the office. The ability of ChatGPT to understand context and generate coherent, detailed notes reduces the burden on physicians, allowing them to focus more on patient care and less on administrative tasks.
Patient inbox messages
Patient portals and EMR systems allow patients to ping their ophthalmologist with clinical questions at any time. While this ease of access is excellent for patients requiring urgent attention, the burden on a busy ophthalmologist can be sizable.
NLP and ChatGPT can streamline patient communication by handling initial responses to patient queries, providing information on treatment plans and ensuring timely responses through the EMR. Patients can receive instant, human-like responses to their inquiries, improving communication efficiency, enhancing patient engagement and satisfaction, and fostering a sense of connectivity between patients and their health-care providers. Additionally, studies have shown that the longer responses generated by ChatGPT to patient inbox messages are perceived as more empathetic.1
INTERPRETATION OF CLINICAL IMAGING AND TESTING
Ophthalmologists employ a litany of imaging tests, including fundus photographs and OCTs. Manual review and report generation for each of these imaging tests can take up valuable physician time.
NLP and Chat-GPT, alongside complex AI-based image analysis algorithms, could review ophthalmic images, extract key information from images and reports, and provide a concise summary that is automatically entered into the EMR. This would be similar to the automated reports generated on EKGs and would be editable by the physician.
As workflows shift toward remote patient monitoring and telemedical models of care, automated interpretation of diagnostic tests can aid ophthalmologists in making well-informed decisions about patient care for larger amounts of data in a shorter time interval.
TELEMEDICINE AND REMOTE EYE CARE
Telemedicine has emerged as a transformative care solution, breaking down geographical barriers and ensuring broader access to specialized eye care. Remote consultations, virtual follow-ups and telemonitoring are set to become integral components of ophthalmic practice over the next 5 to 10 years. This shift is driven not only by technological advancements, but also by the increasing demand for convenient and patient-centric healthcare solutions as well as the increasing number of patients requiring eye care despite the relatively stable number of eye doctors.
The fusion of AI and telemedicine in ophthalmology amplifies the impact of both technologies. Already, AI is capable of autonomously diagnosing diabetic retinopathy using FDA-approved AI-enabled fundus cameras meant to be placed in primary care offices. In the future, similar technology will undoubtedly exist for a litany of ophthalmic diagnoses, such as glaucoma and macular degeneration. Physicians will still be responsible for treating these patients, but telemedicine platforms embedded with NLP and AI algorithms can facilitate patient education, patient scheduling and disease prognostication. Ophthalmologists will be able to leverage AI-driven insights to enhance the quality of remote consultations, ensuring that patients receive timely and accurate assessments of their eye health.
The integration of AI and telemedicine extends beyond virtual consultations. AI algorithms can analyze data collected through remote patient monitoring, detecting subtle changes in health parameters that may indicate disease progression. Imagine patients obtaining home-based OCT images or IOP readings. AI-based real-time data analysis enables proactive interventions, preventing potential complications and improving overall patient outcomes. NLP models will be helpful in communicating “normal” or “healthy” results to patients through automated patient messages or letters instead of relying on one-on-one telemedical communications.
CHALLENGES AND ETHICAL CONSIDERATIONS
While the integration of AI, NLP and ChatGPT into ophthalmology and EMR workflows offers numerous benefits, ethical considerations such as patient privacy, data security and potential biases in language processing must be diligently addressed.
The processing of sensitive patient information through NLP and ChatGPT raises concerns about privacy. ChatGPT is not HIPAA compliant, and, at present, no patient information should ever be uploaded into the system. Future models that are HIPAA compliant, encrypted and secure should be made available to health-care practices in order to realize the promise of NLP in ophthalmology.
Additionally, patients may not fully comprehend the extent to which NLP and ChatGPT are utilized in their health care. This is important, as NLP can generate inaccurate statements about patient’s health or in response to patient concerns, which may lead to unnecessary stress or false reassurance, potentially impacting clinical decision-making and patient outcomes. Inaccurate statements and lack of transparency may erode patient trust in the healthcare system. Ultimately, physician practices should attempt to inform patients about the use of NLP in their practice, as well as the risks of such technologies.
Furthermore, NLP models, including ChatGPT, can inadvertently perpetuate biases present in training data, potentially leading to disparate and unfair treatment of certain demographic groups. It is important to continually assess for biases in these models, and developers should employ techniques to reduce bias in language models to ensure fair and equitable outcomes for all patient populations.
Overreliance on AI diagnostics and language models might erode physician autonomy, clinical judgment and decision-making skills over time. Physicians should ensure novel AI tools are used as aids to enhance rather than replace their expertise.
Lastly, it is unclear where the liability will fall when ChatGPT is employed in the clinical setting. Physicians will likely be considered responsible for clinical decision making, but what happens when an NLP model provides a patient with incorrect instruction on the use of their glaucoma drops? Who is at fault if the NLP response to a patient’s urgent concern leads to a delay in care for a vision-threatening issue? As AI and NLP tools become more ubiquitous, these questions will likely be answered over time. Until then, care and caution should be taken when choosing to employ these tools.
CONCLUSION
In the evolving landscape of ophthalmology, the integration of AI, NLP and ChatGPT within clinical workflows and the EMR has the potential to redefine patient care and improve physician efficiency. The promise of automated documentation and improved communication efficiency may ultimately reduce physician burnout, increase efficiency and improve patient outcomes. As ophthalmologists navigate the transformative incorporation of these technologies into practices, a careful approach that embraces technological advancements while upholding ethical standards of patient care will pave the way for a brighter future for eye care.
The synergistic impact of AI, NLP, and telemedicine is poised to shape the future of ophthalmology, offering a paradigm shift in the delivery of eye-care services. OM
Reference
- Ayers JW, Poliak A, Dredze M, et al. Comparing physician and artificial intelligence chatbot responses to patient questions posted to a public social media forum. JAMA Intern Med. 2023;183:589-596.
