Abstract
In the study by Javid et al., the potential of a Large Language Model (LLM) ChatGPT to deliver complete, accurate information to urolithiasis patients in a meaningful, empathetic, relevant manner is explored. In this initial foray into the world of office ChatGPT, two evaluating urologists found the computer platform to be largely inferior to the in-person encounter in all domains, except for “relevance.” The authors in the penultimate paragraph of their discussion begin: “It is evident that while ChatGPT and similar LLMs can serve as beneficial adjuncts in healthcare practices, they cannot fully substitute the crucial expertise and experience inherent to humans.”
1
While I agree with the statement, it is clear to me that LLM ChatGPT and other aspects of artificial intelligence (AI) will soon transform the physician–patient experience, making the patient's visit more efficient and more effective. With the up-to-date AUA and EAU Guidelines built into the LLM, there will slowly evolve a degree of greater uniformity regarding how individuals with urolithiasis, and all other patients with urologic maladies, are counseled. In more general terms, in the realm of urolithiasis, ChatGPT during a visit will provide the patient with detailed dietary information on how to avoid recurrent stone disease thereby freeing the urologist to devote more time to higher taxonomy questions from the patient.
At present, it would seem that the amount of time physicians in general spend with their urolithiasis patients discussing diet and hydration is highly variable and this in turn inversely impacts the stone recurrence rate. Nowhere was this more evident than in the recent New England Journal of Medicine article declaring little benefit with hydrochlorothiazide therapy among urolithiasis patients. 2 In this study, patients were not uniformly counseled on dietary measures before beginning the trial nor was it mandatory for them to even have hypercalciuria to enter the trial. Indeed, 40% were normocalciuric and thus in many kidney stone clinics would never have even been offered a thiazide treatment, let alone one with such a short half-life.
In the not-too-distant future, one can well envision a situation in which every patient presenting with urolithiasis begins the initial counseling part of the visit, with their physician empathetically imparting meaningful dietary advice via an artificial intelligence (AI) doppelganger displayed on the computer screen ubiquitous in exam rooms throughout the country. The appearance, mannerisms, and intonation will seamlessly be those of the specific physician seeing the patient. AI will also come into play to the benefit of the physician as it may well, upon entry of the physician's note, assess the data at hand and potentially provide “suggestions” about further evaluation or treatment given what eventually will be an all-encompassing PubMed knowledge base.
To be sure, there will always be those who seek to ignore or destroy this “loom” with its encyclopedic “knowledge” of the literature/guidelines and ability to bring to light diagnoses otherwise not considered or progress in treatment modalities possibly overlooked. For many though, this will enhance their ability to offer the most informed and best care for their patients and in that regard, AI will humbly be embraced.
In the final analysis, despite our own pride in practice, it is so important to accept the adage that “no one of us is as smart as all of us” …and that, the encyclopedic “all of us,” via progress in AI, will soon be present at our side to guide and assist, as we seek the very best outcome for each person seeking our care.