A major impediment to using AI in otolaryngology stems from the fact that current electronic health records (EHRs) are designed primarily for documentation and billing purposes, Dr. Bur said. Although they contain an enormous amount of patient data, most data is unstructured and not directly usable by ML algorithms.
Explore This IssueFebruary 2019
EHR developers also tend to incorporate new features that are driven by federally mandated requirements; a clear return on investment would be needed to support deployment of new AI, Dr. Bur said. Unfortunately, current fee-for-service payment models may actually deter health systems from adopting AI. Under fee-for-service, hospitals bill for each clinical activity and, in the case of an incorrect diagnosis, may perform and bill for follow-up tests and care. This means AI that reduces incorrect diagnoses may actually reduce a health system’s revenue if they adopt it.
Dr. Friedland said adequate funding is another detriment. “Developing meaningful AI for clinical use in otolaryngology will require years, if not decades, of research to develop expert systems,” he said. Some problems in the field can only be identified by those on the front lines. We need research funding to perform this work and to incentivize young physicians to study AI and ML in otolaryngology practice.”
Dr. Chowdhury added that there is a large learning curve needed to truly understand the scope of AI and ML research, as it combines advanced elements of statistics, mathematics, computer science, and probability theory, with which most physicians are unfamiliar. “This creates a barrier to entry that may make it challenging for AI models to gain broader acceptance within the medical community,” he said. “People tend to reject what they don’t understand.”
The small datasets that otolaryngologists are accustomed to investigating in the specialty pose yet another challenge, because ML requires large data sets. To overcome this challenge, Dr. Chowdhury recommended collaborating with others to collect large amounts of high-quality data. In fact, it is probably the only way, given the small size of the otolaryngology specialty. “We also need to collaborate with skilled individuals specializing in AI and ML to incorporate this technology into our field,” he said. “It’s probably unrealistic for most otolaryngologists to have a deep understanding of the technology’s nuances, especially considering how quickly things change. Models and algorithms are often outdated within a year or less, so it takes a lot of effort to stay on the leading edge.”
A Promising Future
Despite challenges, otolaryngologists have many reasons to be optimistic about AI’s future, Dr. Bur said. AI has seen tremendous growth, paralleled by advances in computing power, in the last decades, and all indicators signal that this trend will continue. As CMS and private insurers continue to explore alternative payment models, it is likely that incentives for health systems and EHR developers will shift in favor of technologies—such as AI—with the potential to improve healthcare quality and cost. In launching the AI-driven clinical data registry Reg-ENT, which is designed to harness the power of data to guide the best otolaryngology care, the American Academy of Otolaryngology–Head and Neck Surgery is taking an important step to prepare for alternative payment models and increased requirements for quality reporting that will affect reimbursement.
By participating in clinical data registries such as Reg-ENT and by seeking to collaborate with data scientists in developing new AI, otolaryngologists can ensure the field is a pioneer in the development and deployment of AI technologies in clinical practice.
To get more otolaryngologists on board with using AL and ML, Dr. Chowdhury sees general awareness of AI’s potential as the first step. “We hope to have dedicated sessions at upcoming national meetings to talk about and explore AI and ML possibilities, and try to build more collaborations,” he said. For clinicians interested in being actively involved in AI and ML research, he recommended reviewing the principles behind the statistical models like linear regression and logistic regression, as these are often simpler (and better) options for predictive analytics than non-linear models. “With this foundation, it is much easier to understand how ML algorithms work, and their limitations.”