Explore the vast range of titles available.

Long nanopore reads are advantageous in de novo genome assembly. However, nanopore reads usually have broad error distribution and high-error-rate subsequences. Existing error correction tools cannot correct nanopore reads efficiently and effectively. Most methods trim high-error-rate subsequences during error correction, which reduces both the length of the reads and contiguity of the final assembly. Here, we develop an error correction, and de novo assembly tool designed to overcome complex errors in nanopore reads. We propose an adaptive read selection and two-step progressive method to quickly correct nanopore reads to high accuracy. We introduce a two-stage assembler to utilize the full length of nanopore reads. Our tool achieves superior performance in both error correction and de novo assembling nanopore reads. It requires only 8122 hours to assemble a 35X coverage human genome and achieves a 2.47-fold improvement in NG50. Furthermore, our assembly of the human WERI cell line shows an NG50 of 22 Mbp. The high-quality assembly of nanopore reads can significantly reduce false positives in structure variation detection. Nanopore reads have been advantageous for de novo genome assembly; however these reads have high error rates. Here, the authors develop an error correction and de novo assembly tool, NECAT, which produces efficient, high quality assemblies of nanopore reads.

Large‐scale artificial intelligence (AI) models such as ChatGPT have the potential to improve performance on many benchmarks and real‐world tasks. However, it is difficult to develop and maintain these models because of their complexity and resource requirements. As a result, they are still inaccessible to healthcare industries and clinicians. This situation might soon be changed because of advancements in graphics processing unit (GPU) programming and parallel computing. More importantly, leveraging existing large‐scale AIs such as GPT‐4 and Med‐PaLM and integrating them into multiagent models (e.g., Visual‐ChatGPT) will facilitate real‐world implementations. This review aims to raise awareness of the potential applications of these models in healthcare. We provide a general overview of several advanced large‐scale AI models, including language models, vision‐language models, graph learning models, language‐conditioned multiagent models, and multimodal embodied models. We discuss their potential medical applications in addition to the challenges and future directions. Importantly, we stress the need to align these models with human values and goals, such as using reinforcement learning from human feedback, to ensure that they provide accurate and personalized insights that support human decision‐making and improve healthcare outcomes. This review provides an overview of large‐scale AI models, including language models (e.g., ChatGPT), vision‐language models, and language‐conditioned multiagent models, and discusses their potential applications in medicine, as well as their limitations and future trends. We also propose how large‐scale AI models can be integrated into various scenarios of clinical applications.

Although long-read single-cell RNA isoform sequencing (scISO-Seq) can reveal alternative RNA splicing in individual cells, it suffers from a low read throughput. Here, we introduce HIT-scISOseq, a method that removes most artifact cDNAs and concatenates multiple cDNAs for PacBio circular consensus sequencing (CCS) to achieve high-throughput and high-accuracy single-cell RNA isoform sequencing. HIT-scISOseq can yield >10 million high-accuracy long-reads in a single PacBio Sequel II SMRT Cell 8M. We also report the development of scISA-Tools that demultiplex HIT-scISOseq concatenated reads into single-cell cDNA reads with >99.99% accuracy and specificity. We apply HIT-scISOseq to characterize the transcriptomes of 3375 corneal limbus cells and reveal cell-type-specific isoform expression in them. HIT-scISOseq is a high-throughput, high-accuracy, technically accessible method and it can accelerate the burgeoning field of long-read single-cell transcriptomics. Long-read single-cell RNA isoform sequencing can elucidate the intricate landscape of alternative RNA splicing in individual cells, but it suffers from a low read throughput. Here, the authors develop circular consensus sequencing methods to allow high-throughput and high-accuracy single-cell RNA isoform sequencing.

Background/aimsLarge language models (LLMs) have substantial potential to enhance the efficiency of academic research. The accuracy and performance of LLMs in a systematic review, a core part of evidence building, has yet to be studied in detail.MethodsWe introduced two LLM-based approaches of systematic review: an LLM-enabled fully automated approach (LLM-FA) utilising three different GPT-4 plugins (Consensus GPT, Scholar GPT and GPT web browsing modes) and an LLM-facilitated semi-automated approach (LLM-SA) using GPT4’s Application Programming Interface (API). We benchmarked these approaches using three published systematic reviews that reported the prevalence of diabetic retinopathy across different populations (general population, pregnant women and children).ResultsThe three published reviews consisted of 98 papers in total. Across these three reviews, in the LLM-FA approach, Consensus GPT correctly identified 32.7% (32 out of 98) of papers, while Scholar GPT and GPT4’s web browsing modes only identified 19.4% (19 out of 98) and 6.1% (6 out of 98), respectively. On the other hand, the LLM-SA approach not only successfully included 82.7% (81 out of 98) of these papers but also correctly excluded 92.2% of 4497 irrelevant papers.ConclusionsOur findings suggest LLMs are not yet capable of autonomously identifying and selecting relevant papers in systematic reviews. However, they hold promise as an assistive tool to improve the efficiency of the paper selection process in systematic reviews.

We evaluated the prevalence of visual impairment (VI) and blindness among Chinese adults in the Singapore Chinese Eye Study (SCES, 2009–2011), and compared the trends with the Tanjong Pagar Survey, Singapore (TPS), conducted a decade earlier. The SCES comprised of 3,353 Chinese adults aged ≥40 years (response rate, 72.8%). Participants underwent standardized examinations, including measurements of presenting, and best-corrected visual acuity (VA). Bilateral VI (VA < 20/40 to ≥20/200) and blindness (VA < 20/200) were defined based on the United States definition (better-seeing eye). Age-standardized prevalence was calculated using the 2010 Singapore Chinese Population Census. Primary causes and factors associated with VI and blindness were evaluated. In SCES, the age-standardized prevalence of presenting bilateral VI and blindness were 17.7% and 0.6%, respectively; the age-standardised prevalence of best-corrected bilateral VI and blindness were 3.4% and 0.2%, respectively. The previous TPS reported similar rates of best-corrected bilateral VI (3.8%) and blindness (0.3%). In SCES, cataract remains the main cause for both best-corrected bilateral VI (76.0%) and blindness (50.0%). Older age, female, lower income, lower educational level, and smaller housing type were associated with presenting bilateral VI or blindness (all P  ≤ 0.025). These findings will be useful for the planning of eye care services and resource allocation.

Large language models (LLMs) offer the potential to improve virtual patient-physician communication and reduce health care professionals' workload. However, limitations in accuracy, outdated knowledge, and safety issues restrict their effective use in real clinical settings. Addressing these challenges is crucial for making LLMs a reliable health care tool. This study aimed to evaluate the efficacy of Med-RISE, an information retrieval and augmentation tool, in comparison with baseline LLMs, focusing on enhancing accuracy and safety in medical question answering across diverse clinical domains. This comparative study introduces Med-RISE, an enhanced version of a retrieval-augmented generation framework specifically designed to improve question-answering performance across wide-ranging medical domains and diverse disciplines. Med-RISE consists of 4 key steps: query rewriting, information retrieval (providing local and real-time retrieval), summarization, and execution (a fact and safety filter before output). This study integrated Med-RISE with 4 LLMs (GPT-3.5, GPT-4, Vicuna-13B, and ChatGLM-6B) and assessed their performance on 4 multiple-choice medical question datasets: MedQA (US Medical Licensing Examination), PubMedQA (original and revised versions), MedMCQA, and EYE500. Primary outcome measures included answer accuracy and hallucination rates, with hallucinations categorized into factuality (inaccurate information) or faithfulness (inconsistency with instructions) types. This study was conducted between March 2024 and August 2024. The integration of Med-RISE with each LLM led to a substantial increase in accuracy, with improvements ranging from 9.8% to 16.3% (mean 13%, SD 2.3%) across the 4 datasets. The enhanced accuracy rates were 16.3%, 12.9%, 13%, and 9.8% for GPT-3.5, GPT-4, Vicuna-13B, and ChatGLM-6B, respectively. In addition, Med-RISE effectively reduced hallucinations, with reductions ranging from 11.8% to 18% (mean 15.1%, SD 2.8%), factuality hallucinations decreasing by 13.5%, and faithfulness hallucinations decreasing by 5.8%. The hallucination rate reductions were 17.7%, 12.8%, 18%, and 11.8% for GPT-3.5, GPT-4, Vicuna-13B, and ChatGLM-6B, respectively. The Med-RISE framework significantly improves the accuracy and reduces the hallucinations of LLMs in medical question answering across benchmark datasets. By providing local and real-time information retrieval and fact and safety filtering, Med-RISE enhances the reliability and interpretability of LLMs in the medical domain, offering a promising tool for clinical practice and decision support.

The release of ChatGPT in 2022 has catalyzed the adoption of large language models (LLMs) across diverse writing domains, including academic writing. However, this technological shift has raised critical questions regarding the prevalence of LLM usage in academic writing and its potential influence on the quality and impact of research articles. Here, we address these questions by analyzing preprint articles from arXiv, bioRxiv, and medRxiv between 2022 and 2024, employing a novel LLM usage detection tool. Our study reveals that LLMs have been widely adopted in biomedical and other types of academic writing since late 2022. Notably, we noticed that LLM usage is linked to an enhanced impact of research articles after examining their correlation, as measured by citation numbers. Furthermore, we observe that LLMs influence specific content types in academic writing, including hypothesis formulation, conclusion summarization, description of phenomena, and suggestions for future work. Collectively, our findings underscore the potential benefits of LLMs in scientific communication, suggesting that they may not only streamline the writing process but also enhance the dissemination and impact of research findings across disciplines. This study analyzed articles from arXiv, bioRxiv, and medRxiv published between 2022 and 2024, using a novel tool for detecting LLM usage. We found that (1) LLMs have been widely adopted since late 2022, and (2) their usage appears to enhance the impact of research papers. Overall, our findings suggest that LLMs may positively associated with scientific communication without altering core research content.

OBJECTIVE: To examine the association between early retinal arteriolar abnormalities and diabetic peripheral neuropathy (DPN). RESEARCH DESIGN AND METHODS: Data from 608 people (aged 40–80 years) with diabetes from the population-based Singapore Malay Eye Study were analyzed. Participants underwent binocular two-field digital retinal photography and quantitative sensory testing. DPN was defined as an abnormal response to a monofilament or neurothesiometer test. Quantitative changes of retinal vascular caliber and arteriolar bifurcation geometry were measured using a computer-based program. Qualitative retinal signs of retinopathy and retinal arteriolar wall signs were graded by standardized methods. RESULTS: DPN was present in 155 people (25.5%). After adjusting for age, sex, diabetes duration, HbA1c, cardiovascular risk factors, antihypertensive medication use, and peripheral arterial disease, people with suboptimal arteriolar caliber (odds ratio 1.94 [95% CI 1.22–3.10]), larger arteriolar branching coefficient (1.58 [1.03–2.42]), diabetic retinopathy (1.82 [1.20–2.75]), and focal arteriolar narrowing (2.92 [1.48–5.76]) were more likely to have DPN. Participants with a greater number of retinal microvascular signs were more likely to have DPN than those without retinal changes (6.11 [2.11–17.71] for two or more signs and 3.47 [1.18–10.21] for one sign compared with none). CONCLUSIONS: Individuals with diabetes with early retinal arteriolar abnormalities are more likely to have DPN, independent of hyperglycemia and major vascular risk factors. These data support the hypothesis that early microvascular dysfunction, evident in the retina, is an independent risk factor for DPN.

Gigantochloa verticillata is produced in Mengla and Jinghong, Yunnan Province, China, and cultivated in Hong Kong. Vietnam, Thailand, India, Indonesia, and Malaysia are distributed and cultivated. We determined the complete chloroplast genome sequence for G. verticillata using Illumina sequencing data. The complete chloroplast sequence is 139,489 bp, including large single-copy (LSC) region of 83,062 bp, small single-copy (SSC) region of 12,877 bp, and a pair of invert repeats (IR) regions of 21,775 bp. Plastid genome contain 132 genes, 85 protein-coding genes, 39 tRNA genes, and 8 rRNA genes. Phylogenetic analysis based on 23 chloroplast genomes indicates that G. verticillata is closely related to Dendrocalamus latiflorus in Bambusodae.

The detection of cysteine (Cys) and homocysteine (Hcy) in biological fluids has great significance for early diagnosis, including Alzheimer's and Parkinson's disease. The simultaneous determination of Cys and Hcy with a single probe is still a huge challenge. To enlarge the differences in space structure (line and ring) and energy (-721.78 and -761.08 Hartree) between Cys and Hcy, and to cause a difference of aggregation kinetics, gold nanoparticles (AuNPs) are capped with hydrophilic and low-toxic polyvinylpyrrolidone (PVP) (named AuNPs@PVP) and some surface-active sites of AuNPs are masked, the active area for the binding between AuNPs and the detection object is confined, meanwhile, the stability of AuNPs is improved. A novel nanosensor based on confined active area and aggregation kinetics of AuNPs@PVP, is designed for the identification and determination of Cys and Hcy in 1 and 3 min, respectively, with sufficiently low detection limit (4.12 and 4.35 μM) and linear range (4.12–100 μM) for health evaluation. This single colorimetric sensor was applied successfully to the determination of urine and serum, evidencing high anti-interference ability. Graphical Abstract