Exploring the potential of ChatGPT in medical dialogue summarization: a study on consistency with human preferences.

BACKGROUND: Telemedicine has experienced rapid growth in recent years, aiming to enhance medical efficiency and reduce the workload of healthcare professionals. During the COVID-19 pandemic in 2019, it became especially crucial, enabling remote screenings and access to healthcare services while maintaining social distancing. Online consultation platforms have emerged, but the demand has strained the availability of medical professionals, directly leading to research and development in automated medical consultation. Specifically, there is a need for efficient and accurate medical dialogue summarization algorithms to condense lengthy conversations into shorter versions focused on relevant medical facts. The success of large language models like generative pre-trained transformer (GPT)-3 has recently prompted a paradigm shift in natural language processing (NLP) research. In this paper, we will explore its impact on medical dialogue summarization. METHODS: We present the performance and evaluation results of two approaches on a medical dialogue dataset. The first approach is based on fine-tuned pre-trained language models, such as bert-based summarization (BERTSUM) and bidirectional auto-regressive Transformers (BART). The second approach utilizes a large language models (LLMs) GPT-3.5 with inter-context learning (ICL). Evaluation is conducted using automated metrics such as ROUGE and BERTScore. RESULTS: In comparison to the BART and ChatGPT models, the summaries generated by the BERTSUM model not only exhibit significantly lower ROUGE and BERTScore values but also fail to pass the testing for any of the metrics in manual evaluation. On the other hand, the BART model achieved the highest ROUGE and BERTScore values among all evaluated models, surpassing ChatGPT. Its ROUGE-1, ROUGE-2, ROUGE-L, and BERTScore values were 14.94%, 53.48%, 32.84%, and 6.73% higher respectively than ChatGPT's best results. However, in the manual evaluation by medical experts, the summaries generated by the BART model exhibit satisfactory performance only in the "Readability" metric, with less than 30% passing the manual evaluation in other metrics. When compared to the BERTSUM and BART models, the ChatGPT model was evidently more favored by human medical experts. CONCLUSION: On one hand, the GPT-3.5 model can manipulate the style and outcomes of medical dialogue summaries through various prompts. The generated content is not only better received than results from certain human experts but also more comprehensible, making it a promising avenue for automated medical dialogue summarization. On the other hand, automated evaluation mechanisms like ROUGE and BERTScore fall short in fully assessing the outputs of large language models like GPT-3.5. Therefore, it is necessary to research more appropriate evaluation criteria.

RNA G-quadruplex regulates microRNA-26a biogenesis and function.

BACKGROUND & AIMS: RNA G-quadruplexes (RG4s) appear to be important in post-transcriptional gene regulation, but their pathophysiological functions remain unknown. MicroRNA-26a (miR-26a) is emerging as a therapeutic target for various human diseases, however the mechanisms underlying endogenous miR-26a regulation are poorly understood. Herein, we study the role of RG4 in miR-26a expression and function in vitro and in vivo. METHODS: Putative RG4s within liver-enriched miRNAs were predicted by bioinformatic analysis, and the presence of an RG4 structure in the miR-26a-1 precursor (pre-miR-26a-1) was further analyzed by biophysical and biochemical methods. RG4 stabilizers, pre-miR-26a-1 overexpression plasmids, and luciferase reporter assays were used to assess the effect of RG4 on pre-miR-26a-1 maturation. Both miR-26a knock-in and knockout mouse models were employed to investigate the influence of this RG4 on miR-26a expression and function. Moreover, the interaction between RG4 in pre-miR-26a-1 and DEAH-box helicase 36 (DHX36) was determined by biophysical and molecular methods. Finally, miR-26a processing and DHX36 expression were quantified in the livers of obese mice. RESULTS: We identify a guanine-rich sequence in pre-miR-26a-1 that can fold into an RG4 structure. This RG4 impairs pre-miR-26a-1 maturation, resulting in a decrease in miR-26a expression and subsequently an increase in miR-26a cognate targets. In line with known miR-26a functions, this RG4 can regulate hepatic insulin sensitivity and lipid metabolism in vitro and in vivo. Furthermore, we reveal that DHX36 can bind and unwind this RG4 structure, thereby enhancing miR-26a maturation. Intriguingly, there is a concordant decrease of miR-26a maturation and DHX36 expression in obese mouse livers. CONCLUSIONS: Our findings define a dynamic DHX36/RG4/miR-26a regulatory axis during obesity, highlighting an important role of RG4 in physiology and pathology. LAY SUMMARY: Specific RNA sequences called G-quadruplexes (or RG4) appear to be important in post-transcriptional gene regulation. Obesity leads to the formation of these RG4 structures in pre-miR-26a-1 molecules, impairing the maturation and function of miR-26a, which has emerged as a therapeutic target in several diseases. This contributes to hepatic insulin resistance and the dysregulation of liver metabolism.

Metabolism-induced tumor activator 1 (MITA1), an Energy Stress-Inducible Long Noncoding RNA, Promotes Hepatocellular Carcinoma Metastasis.

Metastasis is the main cause of cancer-related death, yet the underlying mechanisms are still poorly understood. Long noncoding RNAs (lncRNAs) are emerging as crucial regulators of malignancies; however, their functions in tumor metastasis remain largely unexplored. In this study, we identify a lncRNA, termed metabolism-induced tumor activator 1 (MITA1), which is up-regulated in hepatocellular carcinoma (HCC) and contributes to metastasis. MITA1, a chromatin-enriched lncRNA discovered by our nuclear RNA sequencing, is significantly induced by energy stress. This induction of MITA1 is governed by the liver kinase B1-adenosine monophosphate-activated protein kinase (LKB1-AMPK) pathway and DNA methylation. Knockdown of MITA1 dramatically inhibits the migration and invasion of liver cancer cells in vitro and HCC metastasis in vivo. Mechanistically, MITA1 promotes the epithelial-mesenchymal transition, an early and central step of metastasis, which may partly attribute to an increase in Slug (snail family zinc finger 2) transcription. MITA1 deficiency reduces the expression of the mesenchymal cell markers, especially Slug, whereas Slug overexpression greatly impairs the effects of MITA1 deficiency on HCC migration and invasion. Correspondingly, there is a positive correlation between the levels of MITA1 and Slug precursors in HCC tissues. Conclusion: Our data reveal MITA1 as a crucial driver of HCC metastasis, and highlight the identified AMPK-MITA1-Slug axis as a potential therapeutic strategy for HCC.

Nonlinear ultrasonic imaging in pulse-echo mode using Westervelt equation: a preliminary research.

Acoustic nonlinear parameter ß, was of great interest in tissue characterization in recent years. Nonlinear imaging methods have been reported to provide improved spatial and contrast resolution. We introduce a nonlinear imaging method derived from nonlinear wave equation based on Gaussian-form solution assumption, which can be applied in pulse-echo mode on diagnostic ultrasound. Through making the use of two pulse transmission, only nonlinear effects are reserved and other effects like scattering, diffraction and linear attenuation can be eliminated. For validation of this method a set of simulation results are generated with a nonlinear simulator. Simulated images also indicate that our method clearly describes the spatial distribution of B/A in the medium.