It is not a matter of "if"; rather the pressing inquiry is "when" AI devices will be able to replace doctors. It appears that for certain tasks, the medical future is close at hand. The most difficult challenge that needs to be addressed is the setting of suitable regulations.Dr. Scott Gottlieb is a CNBC contributor and sits on the boards of Pfizer, Tempus – a genetic testing startup – health-care tech company Aetion Inc. and biotech company Illumina. Additionally, he is a partner at the venture capital firm New Enterprise Associates.Recently, researchers at Harvard revealed a remarkable feat: their language model, ChatGPT, successfully passed the U.S. Medical Licensing Exam, outperforming the 10 percent of medical students who fail annually.This achievement has prompted speculation about the possibilities of artificial intelligence (A.I.) stepping into the shoes of doctors. An understanding of the different A.I. technologies employed in healthcare – as well as the regulatory environment surrounding them – is needed to fully comprehend the potential for revolutionizing the practice of medicine.In general, these technologies can be divided into two categories: machine learning algorithms, which enable computers to learn patterns from data; and natural language processing tools, which allow computers to interpret and generate human language.The A.I. tools currently being applied to healthcare fall into four main paths. First are large language models that process medical claims and create and analyze medical records. Next is supervised machine learning, which interprets clinical data, such as for image analysis in radiology, pathology, and cardiology. As this type of A.I. can directly impact patient care, they are often regulated under the same laws as medical devices.The third category is AI tools that use large language models to extract clinical information from patient-specific data and provide diagnoses or treatments. The fourth path is fully automated large language models that parse a patient's entire medical record to diagnose and prescribe treatments.Though the technology has not quite reached this level yet, it holds significant promise for addressing financial challenges, such as those presented by Baumol's cost disease. As the quality and scope of data continues to improve, these tools will become increasingly capable of enhancing the productivity of providers while, in many cases, beginning to substitute for them.
Comentários