Background: Large language models (LLMs) have gained prominence since the release of ChatGPT in late 2022. Objective: The aim of this study was to assess the accuracy of citations and references generated by ChatGPT (GPT-3.5) in two distinct academic domains: the natural sciences and humanities. Methods: Two researchers independently prompted ChatGPT to write an introduction section for a manuscript and include citations; they then evaluated the accuracy of the citations and Digital Object Identifiers (DOIs). Results were compared between the two disciplines. Results: Ten topics were included, including 5 in the natural sciences and 5 in the humanities. A total of 102 citations were generated, with 55 in the natural sciences and 47 in the humanities. Among these, 40 citations (72.7%) in the natural sciences and 36 citations (76.6%) in the humanities were confirmed to exist (P=.42). There were significant disparities found in DOI presence in the natural sciences (39/55, 70.9%) and the humanities (18/47, 38.3%), along with significant differences in accuracy between the two disciplines (18/55, 32.7% vs 4/47, 8.5%). DOI hallucination was more prevalent in the humanities (42/55, 89.4%). The Levenshtein distance was significantly higher in the humanities than in the natural sciences, reflecting the lower DOI accuracy. Conclusions: ChatGPT’s performance in generating citations and references varies across disciplines. Differences in DOI standards and disciplinary nuances contribute to performance variations. Researchers should consider the strengths and limitations of artificial intelligence writing tools with respect to citation accuracy. The use of domain-specific models may enhance accuracy.

Background: The gap between the demand and supply of donor livers is still a considerable challenge. Since static cold storage is not sufficient in marginal livers, machine perfusion is being explored as an alternative. The objective of this study was to assess (dual) hypothermic oxygenated machine perfusion (HOPE/D-HOPE) and normothermic machine perfusion (NMP) in contrast to static cold storage (SCS). Methods: Three databases were searched to identify studies about machine perfusion. Graft and patient survival and postoperative complications were evaluated using the random effects model. Results: the incidence of biliary complications was lower in HOPE vs. SCS (OR: 0.59, 95% CI: 0.36–0.98, p = 0.04, I²: 0%). There was no significant difference in biliary complications between NMP and SCS (OR: 0.76, 95% CI: 0.41–1.40, p = 0.38, I²: 55%). Graft and patient survival were significantly better in HOPE than in SCS (HR: 0.40, 95% CI: 0.23–0.71, p = 0.002, I²: 0%) and (pooled HR: 0.43, 95% CI: 0.20–0.93, p = 0.03, I²: 0%). Graft and patient survival were not significantly different between NMP and SCS. Conclusion: HOPE/D-HOPE and NMP are promising alternatives to SCS for donor liver preservation. They may help address the widening gap between the demand for and availability of donor livers by enabling the rescue and transplantation of marginal livers.

Objective: The aim of this study is to determine the clinical efficacy of bile-derived liquid biopsy compared with plasma and tumor tissue biopsy in patients with biliary tract carcinoma (BTC).Methods: A total of 13 patients with BTC were enrolled in this cohort. Tumor tissue, bile, and plasma samples were obtained and analyzed using next-generation sequencing for genomic profiling.Results: Bile and plasma samples were collected from all 13 patients, and 11 patients also had matched tumor tissues available. The cell-free DNA (cfDNA) concentration was significantly higher in the bile supernatant than in plasma (median: 1918 vs. 63.1 ng/ml, p = 0.0017). The bile supernatant and pellet had a significantly higher mean mutation allele frequency (MF) than plasma (median: 3.84% vs. 4.22% vs. 0.16%; p < 0.001). Genomic alterations were predominantly missense. Both bile supernatant and pellet had significantly more genomic alterations than plasma (average: 9.3 vs. 7.2 vs. 2.3 alterations per sample; p < 0.01). Among the top 10 most frequent genomic alterations, the consistency between bile supernatant and tumor tissue was 90.00% (18/20), that between bile pellet and tumor tissue was 85.00% (17/20), and that between the plasma and tissue was only 35.00% (7/20). MAF of both bile supernatant and pellet was positively correlated with that in tissue samples (ρ < 0.0001, spearman r = 0.777, and ρ < 0.0001, spearman r = 0.787, respectively), but no significant correlation with tissue was found in the plasma (ρ = 0.966, spearman r = 0.008). Furthermore, additional genomic alterations could be detected in bile supernatant and pellet than in tissue. Potential targets for targeted therapy were identified in bile supernatant and pellet. Regarding copy number variation (CNV) and chromosomal instability (CIN) detection, four additional CNVs from two patients were detected in the bile supernatant that was not detected in tissues (i.e., amplification of TERC, IL7R, RICTOR, and TERT). CIN was significantly higher in tumor tissue than in plasma. The CIN of the bile was also significantly higher than that of plasma. There was no significant difference in CIN between the tissue and the bile supernatant.Conclusion: The consistency of all genomic alterations and tumor tissue-determined genomic alteration in the bile supernatant/pellet was significantly higher than in plasma. Bile supernatants/pellets are better for genetic sequencing and may also have potential clinical value to guide targeted therapy and evaluate prognosis. Bile cfDNA may be a feasible substitute for tumor tissue in the genetic testing of patients with BTC.

Undifferentiated pancreatic carcinomas are rare malignant tumors of the pancreas that are very aggressive and challenging to diagnose. The WHO categorizes them into undifferentiated osteoclast-like giant cell, sarcomatoid, and rhabdoid pancreatic carcinomas. Patients present with nonspecific symptoms such as jaundice, vague abdominal or back pain and itchy skin. Their histological characteristics include positive pan-cytokeratin mononuclear pleomorphic cells, osteoclast-like giant cells and CD68. Patients may have KRAS, TP53 and SMAD4 alterations, homozygous deletions of CDKN2A and CDKN2B, as well as INI1 deficiency. Surgical resection is the only curative treatment. Patients may benefit from postoperative adjuvant therapy. There are no widely accepted guidelines specific to this type of tumor; however, some chemotherapy regimens may be promising. The patient prognosis is mostly poor, especially in patients with unresectable tumors. However, several studies have shown patients achieving long-term survival with adjuvant therapy. In conclusion, although undifferentiated pancreatic carcinoma is rare and very aggressive, there is still potential for improved patient survival with proper diagnosis and treatment.

Background/Objectives: Liver-on-a-chip (LiOC) technology is increasingly recognized as a transformative platform for modeling liver biology, disease mechanisms, drug metabolism, and toxicity screening. Traditional two-dimensional (2D) in vitro models lack the complexity needed to replicate the liver’s unique microenvironment. This review aims to summarize recent advancements in LiOC systems, emphasizing their potential in biomedical research and translational applications. Methods: This narrative review synthesizes findings from key studies on the development and application of LiOC platforms. We explored innovations in material science and bioengineering, including microfluidic design, 3D printing, stem cell– and tissue-derived liver organoid integration, and co-culture strategies. Commercially available LiOC systems and their regulatory relevance were also evaluated. Results: LiOC systems have evolved from simple PDMS-based chips to complex, multicellular constructs incorporating hepatocytes, endothelial cells, Kupffer cells, and hepatic stellate cells. Recent studies demonstrate their superior ability to replicate liver-specific architecture and functions. Applications span cancer research, drug toxicity assessment (e.g., drug-induced liver injury prediction with >85% sensitivity), disease modeling, and regenerative medicine. Several platforms have gained FDA recognition and are in active use for preclinical drug testing. Conclusions: LiOC technology offers a more physiologically relevant alternative to traditional models and holds promise for reducing reliance on animal studies. While challenges remain, such as vascularization and long-term function, ongoing advancements are paving the way toward clinical and pharmaceutical integration. The technology is poised to play a key role in personalized medicine and next-generation therapeutic development.

The hypoxic environment in tumors is closely linked to tumor structure, function, dissemination, invasion, metastasis, and drug resistance. Nitroreductase (NTR) is often recognized as a biomarker to evaluate the hypoxia degree for tumor cells. Traditional detection methods such as PET, MRI and multispectral photoacoustic tomography have limitations. Fluorescent probes have garnered attention due to their high sensitivity, rapid response, specificity, and non-invasive nature. In this study, we introduced a novel small molecule fluorescent probe, T-TPE-NO₂, designed with an AIE molecular framework TPE and successfully targeted to the mitochondria of tumor cells. The probe had high selectivity and could detect NTR activity in a broad pH range. Additionally, the probe exhibits high sensitivity with a LOD of 46.3 ng/mL. Under tumor NTR, the probe emitted strong fluorescence signals and generated a substantial amount of reactive oxygen species upon laser irradiation, thereby inducing tumor cell death and enabling photodynamic therapy. The synthesis, structural and morphological characterization of the probe were rigorously validated. Experimental results demonstrate that T-TPE-NO₂ exhibited high sensitivity and selectivity for tumor cells, highlighting its potential application in photodynamic therapy. This research offers a new approach for the detection and treatment of tumor hypoxia.