Search

Abstract Background Large language models like GPT-3.5-turbo and GPT-4 hold promise for healthcare professionals, but they may inadvertently inherit biases during their training, potentially affecting their utility in medical applications. Despite few attempts in the past, the precise impact and extent of these biases remain uncertain. Methods We use LLMs to generate responses that predict hospitalization, cost and mortality based on real patient cases. We manually examine the generated responses to identify biases. Results We find that these models tend to project higher costs and longer hospitalizations for white populations and exhibit optimistic views in challenging medical scenarios with much higher survival rates. These biases, which mirror real-world healthcare disparities, are evident in the generation of patient backgrounds, the association of specific diseases with certain racial and ethnic groups, and disparities in treatment recommendations, etc. Conclusions Our findings underscore the critical need for future research to address and mitigate biases in language models, especially in critical healthcare applications, to ensure fair and accurate outcomes for all patients.

Non-alcoholic fatty liver disease (NAFLD) is increasingly recognized for its global prevalence and potential progression to more severe liver diseases such as non-alcoholic steatohepatitis (NASH). The gut microbiota plays a pivotal role in the pathogenesis of NAFLD, yet the detailed characteristics and ecological alterations of gut microbial communities during the progression from non-alcoholic fatty liver (NAFL) to NASH remain poorly understood. Methods: In this study, we conducted a comparative analysis of gut microbiota composition in individuals with NAFL and NASH to elucidate differences and characteristics. We utilized 16S rRNA sequencing to compare the intestinal gut microbiota among a healthy control group (65 cases), NAFL group (64 cases), and NASH group (53 cases). Random forest machine learning and database validation methods were employed to analyze the data. Results: Our findings indicate a significant decrease in the diversity of intestinal flora during the progression of NAFLD (p

Hepatic ischemia/reperfusion injury (HIRI) is a common and inevitable factor leading to poor prognosis in various liver diseases, making the outcomes of current treatments in clinic unsatisfactory. Metformin has been demonstrated to be beneficial to alleviate HIRI in recent studies, however, the underpinning mechanism remains unclear. In this study, we found metformin mitigates HIRI-induced ferroptosis through reshaped gut microbiota in mice, which was confirmed by the results of fecal microbiota transplantation treatment but showed the elimination of the beneficial effects when gut bacteria were depleted using antibiotics. Detailedly, through 16S rRNA and metagenomic sequencing, we identified that the metformin-reshaped microbiota was characterized by the increase of gamma-aminobutyric acid (GABA) producing bacteria. This increase was further confirmed by the elevation of GABA synthesis key enzymes, glutamic acid decarboxylase and putrescine aminotransferase, in gut microbes of metformin-treated mice and healthy volunteers. Furthermore, the benefit of GABA against HIRI-induced ferroptosis was demonstrated in GABA-treated mice. Collectively, our data indicate that metformin can mitigate HIRI-induced ferroptosis by reshaped gut microbiota, with GABA identified as a key metabolite.