The promise and peril of using a large language model to obtain clinical information: ChatGPT performs strongly as a fertility counseling tool with limitations.

OBJECTIVE: To compare the responses of the large language model-based "ChatGPT" to reputable sources when given fertility-related clinical prompts. DESIGN: The "Feb 13" version of ChatGPT by OpenAI was tested against established sources relating to patient-oriented clinical information: 17 "frequently asked questions (FAQs)" about infertility on the Centers for Disease Control (CDC) Website, 2 validated fertility knowledge surveys, the Cardiff Fertility Knowledge Scale and the Fertility and Infertility Treatment Knowledge Score, as well as the American Society for Reproductive Medicine committee opinion "optimizing natural fertility." SETTING: Academic medical center. PATIENT(S): Online AI Chatbot. INTERVENTION(S): Frequently asked questions, survey questions and rephrased summary statements were entered as prompts in the chatbot over a 1-week period in February 2023. MAIN OUTCOME MEASURE(S): For FAQs from CDC: words/response, sentiment analysis polarity and objectivity, total factual statements, rate of statements that were incorrect, referenced a source, or noted the value of consulting providers. FOR FERTILITY KNOWLEDGE SURVEYS: Percentile according to published population data. FOR COMMITTEE OPINION: Whether response to conclusions rephrased as questions identified missing facts. RESULT(S): When administered the CDC's 17 infertility FAQ's, ChatGPT produced responses of similar length (207.8 ChatGPT vs. 181.0 CDC words/response), factual content (8.65 factual statements/response vs. 10.41), sentiment polarity (mean 0.11 vs. 0.11 on a scale of -1 (negative) to 1 (positive)), and subjectivity (mean 0.42 vs. 0.35 on a scale of 0 (objective) to 1 (subjective)). In total, 9 (6.12%) of 147 ChatGPT factual statements were categorized as incorrect, and only 1 (0.68%) statement cited a reference. ChatGPT would have been at the 87th percentile of Bunting's 2013 international cohort for the Cardiff Fertility Knowledge Scale and at the 95th percentile on the basis of Kudesia's 2017 cohort for the Fertility and Infertility Treatment Knowledge Score. ChatGPT reproduced the missing facts for all 7 summary statements from "optimizing natural fertility." CONCLUSION(S): A February 2023 version of "ChatGPT" demonstrates the ability of generative artificial intelligence to produce relevant, meaningful responses to fertility-related clinical queries comparable to established sources. Although performance may improve with medical domain-specific training, limitations such as the inability to reliably cite sources and the unpredictable possibility of fabricated information may limit its clinical use.

The exciting potential for ChatGPT in obstetrics and gynecology.

Natural language processing-the branch of artificial intelligence concerned with the interaction between computers and human language-has advanced markedly in recent years with the introduction of sophisticated deep-learning models. Improved performance in natural language processing tasks, such as text and speech processing, have fueled impressive demonstrations of these models' capabilities. Perhaps no demonstration has been more impactful to date than the introduction of the publicly available online chatbot ChatGPT in November 2022 by OpenAI, which is based on a natural language processing model known as a Generative Pretrained Transformer. Through a series of questions posed by the authors about obstetrics and gynecology to ChatGPT as prompts, we evaluated the model's ability to handle clinical-related queries. Its answers demonstrated that in its current form, ChatGPT can be valuable for users who want preliminary information about virtually any topic in the field. Because its educational role is still being defined, we must recognize its limitations. Although answers were generally eloquent, informed, and lacked a significant degree of mistakes or misinformation, we also observed evidence of its weaknesses. A significant drawback is that the data on which the model has been trained are apparently not readily updated. The specific model that was assessed here, seems to not reliably (if at all) source data from after 2021. Users of ChatGPT who expect data to be more up to date need to be aware of this drawback. An inability to cite sources or to truly understand what the user is asking suggests that it has the capability to mislead. Responsible use of models like ChatGPT will be important for ensuring that they work to help but not harm users seeking information on obstetrics and gynecology.

Adiponectin concentration in mid-trimester amniotic fluid varies with the α-amylase level and maternal and neonatal outcomes.

OBJECTIVE: Factors influencing intraamniotic adiponectin levels and their functional significance remain incompletely elucidated. We prospectively measured adiponectin in amniotic fluid and identified its associations with maternal parameters, mediators in amniotic fluid and pregnancy outcomes. STUDY DESIGN: Mid-trimester amniotic fluid from 571 women was tested for adiponectin, interleukin (IL)-6, IL-8 and α-amylase by enzyme-linked immunosorbant assay (ELISA), after which clinical data were obtained. Correlations between adiponectin and clinical or laboratory variables were analyzed by the Kruskal-Wallis, Mann-Whitney and Spearman rank correlation tests. RESULTS: As compared to median levels in 462 women with a term delivery (7.8 ng/mL), adiponectin was elevated in 14 women who subsequently developed preterm premature rupture of membranes (pPROM) (17.3 ng/mL) and 24 women with an iatrogenic preterm birth (IPTB) (13.9 ng/mL) (P=0.0003), but not in 30 women who subsequently had a spontaneous preterm birth with intact membranes (8.1 ng/mL) (P>0.05). Median adiponectin was also elevated in 13 women whose babies developed fetal growth restriction (FGR) (20.6 ng/mL) (P=0.0055) and in 22 women whose babies had respiratory distress syndrome (RDS) (23.0 ng/mL) (P<0.0001). The adiponectin concentration was positively correlated with amylase (P=0.0089) and inversely correlated with maternal body mass index (P=0.0045). CONCLUSION: Adiponectin is a component of mid-trimester amniotic fluid and its concentration varies with maternal body mass index and subsequent development of pPROM, IPTB, FGR and RDS.

Influence of mid-trimester amniotic fluid on endogenous and lipopolysaccharide-mediated responses of mononuclear lymphoid cells.

PROBLEM We evaluated the influence of amniotic fluid (AF) on immune mediator production by mononuclear leukocytes. METHOD OF STUDY Thirty mid-gestation AFs were incubated with peripheral blood mononuclear cells (PBMCs) in the presence or absence of lipopolysaccharide (LPS). Supernatants were tested for interleukin (IL) - 6, 10, 12, 23, tumor necrosis factor-α (TNF-α) and monocyte chemotactic protein (MCP)-1. RESULTS Endogenous mediator production was minimal or non-detectable. AF stimulated endogenous MCP-1, IL-6 and TNF-α release. In the presence of LPS, production of MCP-1 and IL-10 by PBMCs was enhanced eight- to ninefold by AF. Release of IL-6 and IL-23 was enhanced less than twofold by the addition of AF while TNF-α production was unchanged. AF-stimulated mediator production was similar irrespective of pregnancy outcome. CONCLUSION Selective AF stimulation of LPS-mediated MCP-1 and IL-10 release may be a mechanism to promote antibody production and the influx of phagocytic cells to engulf pathogens while downregulating the production of pro-inflammatory cytokines.