Current Literature

Article Reviews Evidence on Opportunistic Salpingectomy for Ovarian Cancer Prevention

Rius M, Carugno J, Abrao MS, Carmona F. Opportunistic Salpingectomy for Ovarian Cancer Prevention: A Call for Action. J Am Coll Surg. 2023, in press.

Mariona Rius, MD, PhD, and coauthors presented an important review of evidence supporting the value of opportunistic bilateral salpingectomy as a means of reducing ovarian cancer risk in women. They noted that ovarian cancer is the leading cause of gynecologic disease-related death and is the fifth-ranked cause of death among women. Symptoms of ovarian cancer are nonspecific and delayed diagnosis resulting in advanced disease is common.

Research has shown that ovarian cancers originate in the fallopian tube and studies of preventive bilateral salpingectomy have confirmed a decrease in ovarian cancer rates in women undergoing the procedure. Starting in 2010, gynecologic professional societies began to encourage addition of bilateral salpingectomy to other abdominal gynecologic procedures.

In this article, other surgical specialists who perform abdominal procedures are encouraged to add bilateral salpingectomy to the procedure in women who are at risk for developing ovarian cancer. Data cited in the article confirmed that addition of opportunistic bilateral salpingectomy to a common abdominal procedure such as laparoscopic cholecystectomy would prolong the procedure by 13 minutes and there would be minimal risk for complications associated with the added salpingectomy. The data also confirmed that the risk of symptomatic postoperative ovarian dysfunction was minimal. Additional data show that adoption of opportunistic bilateral salpingectomy by surgeons performing abdominal procedures could potentially reduce ovarian cancer mortality by 14.5%, preventing 1,854 deaths annually.

The authors encouraged surgeons to discuss opportunistic bilateral salpingectomy with women who are scheduled to undergo abdominal surgical procedures.

Artificial Intelligence and Machine Learning in Clinical Medicine

Haug CJ, Drazen JM. Artificial Intelligence and Machine Learning in Clinical Medicine, 2023. N Engl J Med 2023;388(13):1201-08. doi: 10.1056/NEJMra2302038

This article reviewed data on the use of artificial intelligence (AI) in healthcare and introduced a series of articles focusing on this topic that will appear over the coming months in the New England Journal of Medicine.

The authors noted that the use of AI to assist in diagnosing diseases began in the 1950s, but progress was slow and interrupted by intermittent failures of the computer programs. As the power of computers has increased, natural language processing became possible and the corpus of data that became available to deep learning programs increased markedly making AI a realistic approach for improving the quality and efficiency of healthcare.

Published research has confirmed the value of AI as a means of interpreting laboratory studies and radiographic images. Expansion of AI into areas such as developing care plans and entering data into the medical record is occurring. Concerns exist, however, in areas such as diagnostic accuracy and proper selection of data to support healthcare decisions. The authors emphasized that research supporting the integration of AI into healthcare practice should focus on clinically relevant questions, be scalable and applicable for problems encountered in everyday practice, and lead to patient benefit.

The series of articles on AI uses in medical practice will seek to provide valuable information for improving patient care.

Benefits, Limits, and Risks of GPT-4 as an AI Chatbot for Medicine

Lee P, Bubeck S, Petro J. Benefits, Limits, and Risks of GPT-4 as an AI Chatbot for Medicine. N Engl J Med 2023;388(13):1233-39. doi: 10.1056/NEJMsr2214184

In the second article of the series on uses of AI in healthcare, Lee and coauthors explored the benefits, limits, and risks of chatbots.

These programs use AI to support a software that provides an interactive communication process where the healthcare professional submits a “prompt” and the AI program devises and delivers a response. The prompt can be a question (“What are the important features of the approach to diagnosis and management of _______ described in these articles?”) or a request (“Please construct a note for the medical record based on this recorded interaction with a patient”). Examples of prompts and responses are provided in the article.

Currently available chatbots (GPT-4 and the Google chatbot LaMDA) have not been specifically programmed to aid healthcare providers, but the programs have proved effective at answering medical questions (USMLE examination questions, for example) and composing medical record entries. The responses of the chatbots can contain errors (known as “hallucinations”); failures in calculating medical formulas and omissions of key data to support recommendations in medical record entries have been discovered frequently. Often, these can be corrected by submitting a separate prompt asking the chatbot to correct errors. It is clear, however, that currently available chatbots will need carefully designed quality control protocols to guarantee safety and benefit for patients.

The future of chatbot use in healthcare will likely be exciting.