Impact of pre-transplant immune checkpoint inhibitor use on post-transplant outcomes in HCC: A systematic review and individual patient data meta-analysis.

BACKGROUND AND AIM: Treatment with immune checkpoint inhibitors (ICIs) for hepatocellular carcinoma (HCC) prior to liver transplantation (LT) has been reported; however, ICIs may elevate the risk of allograft rejection and impact other clinical outcomes. This study aims to summarize the impact of ICI use on post-LT outcomes. MATERIALS AND METHODS: In this individual patient data meta-analysis, we searched databases to identify HCC cases treated with ICIs before LT, detailing allograft rejection, HCC recurrence, and overall survival. We performed Cox regression analysis to identify risk factors for allograft rejection. RESULTS: Among 91 eligible patients, with a median (interquartile range [IQR]) follow-up of 690.0 (654.5) days, there were 24 (26.4%) allograft rejections, 9 (9.9%) HCC recurrences, and 9 (9.9%) deaths. Age (adjusted hazard ratio [aHR] per 10 years=0.72, 95% confidence interval [CI]=0.53, 0.99, P=0.044) and ICI washout time (aHR per 1 week=0.92, 95% CI=0.86, 0.99, P=0.022) were associated with allograft rejection. The median (IQR) washout period for patients with ≤20% probability of allograft rejection was 94 (196) days. Overall survival did not differ between cases with and without allograft rejection (log-rank test, p=0.2). Individuals with HCC recurrence had fewer median (IQR) ICI cycles than those without recurrence (4.0 [1.8]) vs. 8.0 [9.0]); p=0.025). The proportion of patients within Milan post-ICI was lower for those with recurrence vs. without (16.7% vs. 65.3%, p=0.032) CONCLUSION: Patients have acceptable post-LT outcomes after ICI therapy. Age and ICI washout length relate to the allograft rejection risk, and a 3-month washout may reduce it to that of patients without ICI exposure. Number of ICI cycles and tumor burden may affect recurrence risk. Large prospective studies are necessary to confirm these associations. IMPACT AND IMPLICATIONS: This systematic review and individual patient data meta-analysis of 91 patients with hepatocellular carcinoma and immune checkpoint inhibitors use prior to liver transplantation suggests acceptable overall post-transplant outcomes. Older age and longer immune checkpoint inhibitor washout period have a significant inverse association with the risk of allograft rejection. A 3-month washout may reduce it to that of patients without ICI exposure. Additionally, a higher number of immune checkpoint inhibitor cycles and tumor burden within Milan criteria at the completion of immunotherapy may predict a decreased risk of hepatocellular carcinoma recurrence, but this observation requires further validation in larger prospective studies. CODE FOR INTERNATIONAL PROSPECTIVE REGISTER OF SYSTEMATIC REVIEWS (PROSPERO): CRD42023494951.

Exploring the underlying molecular mechanisms of acute myocardial infarction after SARS-CoV-2 infection.

An increase in acute myocardial infarction (AMI)-related deaths has been reported during the COVID-19 pandemic. Despite evidence suggesting the association between severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infection and AMI, the underlying mechanisms remain unclear. Here, we integrated mRNA and microRNA expression profiles related to SARS-CoV-2 infection and AMI from public databases. We then performed transcriptomic analysis using bioinformatics and systems biology approaches to explore the potential molecular mechanisms of SARS-CoV-2 infection affects AMI. First, twenty-one common differentially expressed genes (DEGs) were identified from SARS-CoV-2 infection and AMI patients in endothelial cells datasets and then we performed functional analysis to predict the roles of these DEGs. The functional analysis emphasized that the endothelial cell response to cytokine stimulus due to excessive inflammation was essential in these two diseases. Importantly, the tumor necrosis factor and interleukin-17 signaling pathways appeared to be integral factors in this mechanism. Interestingly, most of these common genes were also upregulated in transcriptomic datasets of SARS-CoV-2-infected cardiomyocytes, suggesting that these genes may be shared in cardiac- and vascular-related injuries. We subsequently built a protein-protein interaction network and extracted hub genes and essential modules from this network. At the transcriptional and post-transcriptional levels, regulatory networks with common DEGs were also constructed, and some key regulator signatures were further identified and validated. In summary, our research revealed that a highly activated inflammatory response in patients with COVID-19 might be a crucial factor for susceptibility to AMI and we identified some candidate genes and regulators that could be used as biomarkers or potential therapeutic targets.

Liver resection had better disease-free survival rates compared with radiofrequency ablation in hepatocellular carcinoma: A meta-analysis based on randomized clinical trials.

BACKGROUND: Liver resection (LR) and radiofrequency ablation (RFA) are the most commonly used treatment modalities for early-stage hepatocellular carcinoma (ES-HCC). The comparative efficacy of LR and RFA in ES-HCC remains debated. We conducted a meta-analysis based on randomized trials to compare the outcomes of LR and RFA. METHODS: We searched PubMed, Embase, the Cochrane Library, and ClinicalTrials.gov for randomized controlled trials (RCTs) comparing RFA and LR interventions for the treatment of ES-HCC. The primary outcomes were overall survival (OS) and disease-free survival (DFS). We used meta-regression to determine the source of heterogeneity and conducted a trial sequential analysis to examine whether the outcome was statistically reliable. RESULTS: Our meta-analysis included nine RCTs with a total of 1,516 HCC patients. Compared with patients receiving RFA, those receiving LR did not have significantly different 2-year OS (HR=1.28, 95% CI: 0.73-2.23) and 5-year OS (HR=1.49, 95% CI: 0.99-2.24). However, patients receiving LR showed a favorable trend in 2-year DFS (HR=1.40, 95% CI: 1.16-1.69) and 5-year DFS (HR=1.37; 95% CI: 1.05-1.77), although these results are not conclusive due to underpowered significance. The heterogeneity was low, and the outcomes were statistically reliable. DISCUSSION: Meta-analysis suggests that while LR shows a favorable trend in DFS compared to RFA for ES-HCC, the present evidence does not thoroughly support recommending LR over RFA. The inconclusive nature of these findings highlights the need for further large-scale RCTs to establish definitive comparative efficacy.

Development and validation of a biomarker index for HCC treatment response.

BACKGROUND: Serum AFP-L3%, AFP, and DCP are useful biomarkers for HCC detection, but their utility in assessing treatment response remains unknown. We aim to evaluate the accuracy of a biomarker model in the detection of posttreatment viable tumors. METHODS: For model derivation, recipients with HCC undergoing liver transplant from 2018 to 2022 who had biomarkers collected within 3 months before transplant were included. We developed a generalized linear model for detecting posttreatment viable tumors with the 3 biomarkers as covariates, which we termed the "LAD Score." An independent cohort of 117 patients with HCC was used for external validation. RESULTS: Among 205 recipients of transplant, 70.2% had evidence of viable tumor on explant. The median LAD score was higher among patients with viable versus nonviable tumors (1.06 vs. 0.465, p < 0.001). The LAD score had a sensitivity of 55.6% and a specificity of 85.1% at the cutoff of 0.927, which was more accurate than imaging for detecting posttreatment viable tumors (AUROC 0.736 vs. 0.643, respectively; p = 0.045). The superior performance of the LAD score over imaging is primarily driven by its greater accuracy in detecting tumors <2 cm in diameter (AUROC of the LAD score 0.721 vs. imaging 0.595, p = 0.02). In the validation data set, the LAD score had an AUROC of 0.832 (95% CI: 0.753, 0.911) with a sensitivity of 72.5% and a specificity of 89.4% at the cutoff of 0.927. CONCLUSIONS: Our findings suggest the utility of LAD score in treatment response assessment after locoregional therapy for HCC, particularly in detecting small tumors. A larger prospective study is in progress to validate its accuracy and evaluate its performance in recurrence monitoring.

Age and urban-rural disparities in cutaneous melanoma mortality rates in the United States during the COVID-19 pandemic.

Most recent studies on the coronavirus disease 2019 (COVID-19) pandemic and cutaneous melanoma (CM) focused more on delayed diagnosis or advanced presentation. We aimed to ascertain mortality trends of CM between 2012 and 2022, focusing on the effects of the COVID-19 pandemic. In this serial population-based study, the National Vital Statistics System dataset was queried for mortality data. Excess CM-related mortality rates were estimated by calculating the difference between observed and projected mortality rates during the pandemic. Totally there were 108,853 CM-associated deaths in 2012-2022. CM-associated mortality saw a declining trend from 2012 to 2019 overall. However, it increased sharply in 2020 (ASMR 3.73 per 100,000 persons, 5.95% excess mortality), and remained high in 2021 and 2022, with the ASMRs of 3.82 and 3.81, corresponding to 11.17% and 13.20% excess mortality, respectively. The nonmetro areas had the most pronounced rise in mortality with 12.20% excess death in 2020, 15.33% in 2021 and 20.52% in 2022, corresponding to a 4-6 times excess mortality risk compared to large metro areas during the pandemic. The elderly had the most pronounced rise in mortality, but the mortality in the younger population was reduced.

Machine Learning-Based Models for Advanced Fibrosis and Cirrhosis Diagnosis in Chronic Hepatitis B Patients With Hepatic Steatosis.

BACKGROUND AND AIMS: The global rise of chronic hepatitis B (CHB) superimposed on hepatic steatosis (HS) warrants noninvasive, precise tools for assessing fibrosis progression. This study leveraged machine learning (ML) to develop diagnostic models for advanced fibrosis and cirrhosis in this patient population. METHODS: Treatment-naive CHB patients with concurrent HS who underwent liver biopsy in 10 medical centers were enrolled as a training cohort and an independent external validation cohort (NCT05766449). Six ML models were implemented to predict advanced fibrosis and cirrhosis. The final models, derived from SHAP (Shapley Additive exPlanations), were compared with Fibrosis-4 Index, nonalcoholic fatty liver disease Fibrosis Score, and aspartate aminotransferase-to-platelet ratio index using the area under receiver-operating characteristic curve (AUROC) and decision curve analysis (DCA). RESULTS: Of 1,198 eligible patients, the random forest model achieved AUROCs of 0.778 (95% confidence interval [CI], 0.749-0.807) for diagnosing advanced fibrosis (random forest advanced fibrosis model) and 0.777 (95% CI, 0.748-0.806) for diagnosing cirrhosis (random forest cirrhosis model) in the training cohort, and maintained high AUROCs in the validation cohort. In the training cohort, the random forest advanced fibrosis model obtained an AUROC of 0.825 (95% CI, 0.787-0.862) in patients with hepatitis B virus DNA ≥105 IU/mL, and the random forest cirrhosis model had an AUROC of 0.828 (95% CI, 0.774-0.883) in female patients. The 2 models outperformed Fibrosis-4 Index, nonalcoholic fatty liver disease Fibrosis Score, and aspartate aminotransferase-to-platelet ratio index in the training cohort, and also performed well in the validation cohort. CONCLUSIONS: The random forest models provide reliable, noninvasive tools for identifying advanced fibrosis and cirrhosis in CHB patients with concurrent HS, offering a significant advancement in the comanagement of the 2 diseases. CLINICALTRIALS: gov, Number: NCT05766449.

Impact of the Early COVID-19 Pandemic on Incidence and Outcomes of Hepatocellular Carcinoma in the United States.

INTRODUCTION: Access to hepatocellular carcinoma (HCC) surveillance and treatments were disrupted during the coronavirus disease 2019 (COVID-19) pandemic. We aimed to characterize the impact of the pandemic on HCC incidence and mortality rates, treatment, and outcomes in the United States. METHODS: Two nationwide databases, the United States Cancer Statistics and the National Vital Statistics System, were used to investigate HCC incidence and mortality between 2001 and 2020. Trends in age-adjusted incidence rate (aIR) and adjusted mortality rate (aMR) were assessed using joinpoint analysis. The 2020 aIR and aMR were projected based on the prepandemic data and compared with actual values to assess the extent of underdiagnosis. We assessed differences in HCC characteristics, treatment, and overall survival between 2020 and 2018-2019. RESULTS: The aIR of HCC in 2020 was significantly reduced compared with 2019 (5.22 vs 6.03/100K person-years [PY]), representing a 12.2% decrease compared with the predicted aIR in 2020 (5.94/100K PY). The greatest extent of underdiagnosis was observed in Black (-14.87%) and Hispanic (-14.51%) individuals and those with localized HCC (-15.12%). Individuals staged as regional or distant HCC were also less likely to receive treatment in 2020. However, there was no significant difference in short-term overall survival in 2020 compared with 2018-2019, with HCC mortality rates remaining stable (aMR: 2.76 vs 2.73/100K PY in 2020 vs 2019). DISCUSSION: The COVID-19 pandemic resulted in underdiagnosis of HCC, particularly early stage disease and racial ethnic minorities, and underuse of HCC-directed treatment. Longer follow-up is needed to determine the impact of the COVID-19 pandemic on HCC-related mortality.

Excess non-COVID-19-related mortality among inflammatory bowel disease decedents during the COVID-19 pandemic.

BACKGROUND: The coronavirus disease 2019 (COVID-19) pandemic disrupted healthcare in the United States. AIM: To investigate COVID-19-related and non-COVID-19-related death and characteristics associated with excess death among inflammatory bowel disease (IBD) decedents. METHODS: We performed a register-based study using data from the National Vital Statistics System, which reports death data from over 99% of the United States population, from January 1, 2006 through December 31, 2021. IBD-related deaths among adults 25 years and older were stratified by age, sex, race/ethnicity, place of death, and primary cause of death. Predicted and actual age-standardized mortality rates (ASMRs) per 100000 persons were compared. RESULTS: 49782 IBD-related deaths occurred during the study period. Non-COVID-19-related deaths increased by 13.14% in 2020 and 18.12% in 2021 [2020 ASMR: 1.55 actual vs 1.37 predicted, 95% confidence interval (CI): 1.26-1.49; 2021 ASMR: 1.63 actual vs 1.38 predicted, 95%CI: 1.26-1.49]. In 2020, non-COVID-19-related mortality increased by 17.65% in ulcerative colitis (UC) patients between the ages of 25 and 65 and 36.36% in non-Hispanic black (NHB) Crohn's disease (CD) patients. During the pandemic, deaths at home or on arrival and at medical facilities as well as deaths due to neoplasms also increased. CONCLUSION: IBD patients suffered excess non-COVID-19-related death during the pandemic. Excess death was associated with younger age among UC patients, and with NHB race among CD patients. Increased death at home or on arrival and due to neoplasms suggests that delayed presentation and difficulty accessing healthcare may have led to increased IBD mortality.

MASLD/MetALD and mortality in individuals with any cardio-metabolic risk factor: A population-based study with 26.7 years of follow-up.

BACKGROUND AND AIMS: A new term, metabolic dysfunction-associated steatotic liver disease (MASLD), has been proposed by a multi-society expert panel. However, it remains unclear whether hepatic steatosis per se in MASLD contributes to an increased risk of mortality in individuals with any cardio-metabolic risk factor (CMRF), which is also a significant risk factor for increased mortality. This study aimed to compare all-cause and cause-specific mortality between the "MASLD/MetALD" and "no steatotic liver disease (SLD)" groups in individuals with any CMRF. APPROACH AND RESULTS: A population-based cohort study was conducted using 10,750 participants of the Third National Health and Nutrition Examination Survey. All-cause and cause-specific (cardiovascular, cancer, diabetes, and liver) mortality risks were compared between the "MASLD," "MetALD," and "no SLD" groups using the Cox proportional hazards model with complex survey design weights, adjusted for confounders. Over 26 years, the "MASLD" group did not show significantly increased all-cause (adjusted HR 1.04[95% CI: 0.95-1.14], p = 0.413), cardiovascular (0.88 [0.75-1.04], p = 0.139), or cancer (1.06[0.84-1.33], p = 0.635) mortality risk compared to the "no SLD" group in individuals with any CMRF. The MetALD group was associated with increased all-cause (1.41 [1.05-1.89], p = 0.022), cancer (2.35 [1.33-4.16], p = 0.004), and liver (15.04 [2.96-76.35], p = 0.002) mortality risk compared with the no SLD group. This trend was more pronounced in the MetALD group with advanced fibrosis assessed by Fibrosis-4 (FIB-4). CONCLUSIONS: In individuals with CMRF, the presence of steatotic liver disease (MASLD) alone did not increase the risk of mortality, except in cases with more alcohol consumption (MetALD). Therefore controlling metabolic risk factors and reducing alcohol consumption in people with MASLD or MetALD will be crucial steps to improve long-term health outcomes.

Appropriateness of ChatGPT in Answering Heart Failure Related Questions.

BACKGROUND: Heart failure requires complex management, and increased patient knowledge has been shown to improve outcomes. This study assessed the knowledge of Chat Generative Pre-trained Transformer (ChatGPT) and its appropriateness as a supplemental resource of information for patients with heart failure. METHOD: A total of 107 frequently asked heart failure-related questions were included in 3 categories: "basic knowledge" (49), "management" (41) and "other" (17). Two responses per question were generated using both GPT-3.5 and GPT-4 (i.e., two responses per question per model). The accuracy and reproducibility of responses were graded by two reviewers, board-certified in cardiology, with differences resolved by a third reviewer, board-certified in cardiology and advanced heart failure. Accuracy was graded using a four-point scale: (1) comprehensive, (2) correct but inadequate, (3) some correct and some incorrect, and (4) completely incorrect. RESULTS: GPT-4 provided 107/107 (100%) responses with correct information. Further, GPT-4 displayed a greater proportion of comprehensive knowledge for the categories of "basic knowledge" and "management" (89.8% and 82.9%, respectively). For GPT-3, there were two total responses (1.9%) graded as "some correct and incorrect" for GPT-3.5, while no "completely incorrect" responses were produced. With respect to comprehensive knowledge, GPT-3.5 performed best in the "management" category and "other" category (prognosis, procedures, and support) (78.1%, 94.1%). The models also provided highly reproducible responses, with GPT-3.5 scoring above 94% in every category and GPT-4 with 100% for all answers. CONCLUSIONS: GPT-3.5 and GPT-4 answered the majority of heart failure-related questions accurately and reliably. If validated in future studies, ChatGPT may serve as a useful tool in the future by providing accessible health-related information and education to patients living with heart failure. In its current state, ChatGPT necessitates further rigorous testing and validation to ensure patient safety and equity across all patient demographics.

Disparities in Palliative Care Use for Patients With Blood Cancer Who Died in the Hospital.

Background: Palliative care can enhance quality of life during a terminal hospitalization. Despite advances in diagnostic and treatment tools, blood cancers lag behind solid malignancies in palliative use. It is not clear what factors affect palliative care use in blood cancer. Methods: We used the 2016 to 2019 National Inpatient Sample to identify demographic and socioeconomic factors associated with receiving palliative care among patients over age 18 with any malignant hematological diagnosis during a terminal hospitalization lasting at least 3 days, excluding those receiving a stem cell transplant. Results: Palliative care use was documented 54% of the time among 49,720 weighted cases (9944 distinct individual hospitalizations), approximately evenly distributed across the years 2016-2019. Palliative care use was lowest in 2016 (51%) and highest in 2018 (58%), and increased with age, reaching 58% for those 80 years and older. Men and women were similarly likely to receive care. Patients of Hispanic ethnicity and African Americans received less palliative care (47% and 49%, respectively), as did those insured by Medicaid (48%), and those admitted to small or rural hospitals (52% and 47%, respectively). Charges for hospitalizations with palliative care were 19% lower than for those without it. Conclusions: This study highlights disparities in palliative care use among blood-cancer patients who died in the hospital. It seems likely that many of the 46% who did not receive palliative care could have benefitted from it. Interventions are likely needed to achieve equitable access to ideal levels of palliative care services in late-stage blood cancer.

A Multidisciplinary Assessment of ChatGPT's Knowledge of Amyloidosis: Observational Study.

BACKGROUND: Amyloidosis, a rare multisystem condition, often requires complex, multidisciplinary care. Its low prevalence underscores the importance of efforts to ensure the availability of high-quality patient education materials for better outcomes. ChatGPT (OpenAI) is a large language model powered by artificial intelligence that offers a potential avenue for disseminating accurate, reliable, and accessible educational resources for both patients and providers. Its user-friendly interface, engaging conversational responses, and the capability for users to ask follow-up questions make it a promising future tool in delivering accurate and tailored information to patients. OBJECTIVE: We performed a multidisciplinary assessment of the accuracy, reproducibility, and readability of ChatGPT in answering questions related to amyloidosis. METHODS: In total, 98 amyloidosis questions related to cardiology, gastroenterology, and neurology were curated from medical societies, institutions, and amyloidosis Facebook support groups and inputted into ChatGPT-3.5 and ChatGPT-4. Cardiology- and gastroenterology-related responses were independently graded by a board-certified cardiologist and gastroenterologist, respectively, who specialize in amyloidosis. These 2 reviewers (RG and DCK) also graded general questions for which disagreements were resolved with discussion. Neurology-related responses were graded by a board-certified neurologist (AAH) who specializes in amyloidosis. Reviewers used the following grading scale: (1) comprehensive, (2) correct but inadequate, (3) some correct and some incorrect, and (4) completely incorrect. Questions were stratified by categories for further analysis. Reproducibility was assessed by inputting each question twice into each model. The readability of ChatGPT-4 responses was also evaluated using the Textstat library in Python (Python Software Foundation) and the Textstat readability package in R software (R Foundation for Statistical Computing). RESULTS: ChatGPT-4 (n=98) provided 93 (95%) responses with accurate information, and 82 (84%) were comprehensive. ChatGPT-3.5 (n=83) provided 74 (89%) responses with accurate information, and 66 (79%) were comprehensive. When examined by question category, ChatGTP-4 and ChatGPT-3.5 provided 53 (95%) and 48 (86%) comprehensive responses, respectively, to "general questions" (n=56). When examined by subject, ChatGPT-4 and ChatGPT-3.5 performed best in response to cardiology questions (n=12) with both models producing 10 (83%) comprehensive responses. For gastroenterology (n=15), ChatGPT-4 received comprehensive grades for 9 (60%) responses, and ChatGPT-3.5 provided 8 (53%) responses. Overall, 96 of 98 (98%) responses for ChatGPT-4 and 73 of 83 (88%) for ChatGPT-3.5 were reproducible. The readability of ChatGPT-4's responses ranged from 10th to beyond graduate US grade levels with an average of 15.5 (SD 1.9). CONCLUSIONS: Large language models are a promising tool for accurate and reliable health information for patients living with amyloidosis. However, ChatGPT's responses exceeded the American Medical Association's recommended fifth- to sixth-grade reading level. Future studies focusing on improving response accuracy and readability are warranted. Prior to widespread implementation, the technology's limitations and ethical implications must be further explored to ensure patient safety and equitable implementation.

Alpha-fetoprotein: Past, present, and future.

Alpha-fetoprotein (AFP) is a glycoprotein that plays an important role in immune regulation with critical involvement in early human development and maintaining the immune balance during pregnancy. Postfetal development, the regulatory mechanisms controlling AFP undergo a shift and AFP gene transcription is suppressed. Instead, these enhancers refocus their activity to maintain albumin gene transcription throughout adulthood. During the postnatal period, AFP expression can increase in the setting of hepatocyte injury, regeneration, and malignant transformation. It is the first oncoprotein discovered and is routinely used as part of a screening strategy for HCC. AFP has been shown to be a powerful prognostic biomarker, and multiple HCC prognosis models confirmed the independent prognostic utility of AFP. AFP is also a useful predictive biomarker for monitoring the treatment response of HCC. In addition to its role as a biomarker, AFP plays important roles in immune modulation to promote tumorigenesis and thus has been investigated as a therapeutic target in HCC. In this review article, we aim to provide an overview of AFP, encompassing the discovery, biological role, and utility as an HCC biomarker in combination with other biomarkers and how it impacts clinical practice and future direction.

Artificial intelligence in liver cancer research: a scientometrics analysis of trends and topics.

Background and aims: With the rapid growth of artificial intelligence (AI) applications in various fields, understanding its impact on liver cancer research is paramount. This scientometrics project aims to investigate publication trends and topics in AI-related publications in liver cancer. Materials and Methods: We employed a search strategy to identify AI-related publications in liver cancer using Scopus database. We analyzed the number of publications, author affiliations, and journals that publish AI-related publications in liver cancer. Finally, the publications were grouped based on intended application. Results: We identified 3950 eligible publications (2695 articles, 366 reviews, and 889 other document types) from 1968 to August 3, 2023. There was a 12.7-fold increase in AI-related publications from 2013 to 2022. By comparison, the number of total publications on liver cancer increased by 1.7-fold. Our analysis revealed a significant shift in trends of AI-related publications on liver cancer in 2019. We also found a statistically significant consistent increase in numbers of AI-related publications over time (tau = 0.756, p < 0.0001). Eight (53%) of the top 15 journals with the most publications were radiology journals. The largest number of publications were from China (n=1156), the US (n=719), and Germany (n=236). The three most common publication categories were "medical image analysis for diagnosis" (37%), "diagnostic or prognostic biomarkers modeling & bioinformatics" (19%), and "genomic or molecular analysis" (18%). Conclusion: Our study reveals increasing interest in AI for liver cancer research, evidenced by a 12.7-fold growth in related publications over the past decade. A common application of AI is in medical imaging analysis for various purposes. China, the US, and Germany are leading contributors.

Large language models and bariatric surgery patient education: a comparative readability analysis of GPT-3.5, GPT-4, Bard, and online institutional resources.

BACKGROUND: The readability of online bariatric surgery patient education materials (PEMs) often surpasses the recommended 6th grade level. Large language models (LLMs), like ChatGPT and Bard, have the potential to revolutionize PEM delivery. We aimed to evaluate the readability of PEMs produced by U.S. medical institutions compared to LLMs, as well as the ability of LLMs to simplify their responses. METHODS: Responses to frequently asked questions (FAQs) related to bariatric surgery were gathered from top-ranked health institutions. FAQ responses were also generated from GPT-3.5, GPT-4, and Bard. LLMs were then prompted to improve the readability of their initial responses. The readability of institutional responses, initial LLM responses, and simplified LLM responses were graded using validated readability formulas. Accuracy and comprehensiveness of initial and simplified LLM responses were also compared. RESULTS: Responses to 66 FAQs were included. All institutional and initial LLM responses had poor readability, with average reading levels ranging from 9th grade to college graduate. Simplified responses from LLMs had significantly improved readability, with reading levels ranging from 6th grade to college freshman. When comparing simplified LLM responses, GPT-4 responses demonstrated the highest readability, with reading levels ranging from 6th to 9th grade. Accuracy was similar between initial and simplified responses from all LLMs. Comprehensiveness was similar between initial and simplified responses from GPT-3.5 and GPT-4. However, 34.8% of Bard's simplified responses were graded as less comprehensive compared to initial. CONCLUSION: Our study highlights the efficacy of LLMs in enhancing the readability of bariatric surgery PEMs. GPT-4 outperformed other models, generating simplified PEMs from 6th to 9th grade reading levels. Unlike GPT-3.5 and GPT-4, Bard's simplified responses were graded as less comprehensive. We advocate for future studies examining the potential role of LLMs as dynamic and personalized sources of PEMs for diverse patient populations of all literacy levels.