Identification of Matrine as a Kirsten rats Arcomaviral oncogene homolog inhibitor alleviating chemotherapy-induced neuropathic pain.

BACKGROUND: Chemotherapy-induced peripheral neuropathy (CIPN) represents a prevailing and severe clinical concern, characterized by limited availability of clinically effective treatment strategies. Current evidence endorses matrine's potential as a neuroprotective and analgesic agent for CIPN. Nevertheless, the precise targets and mechanisms of action of matrine remain insufficiently explored, impeding comprehensive pharmacological investigation and clinical application. OBJECTIVE: This study endeavors to elucidate the analgesic and neuroprotective effects of matrine in mice with vincristine-induced neuropathic pain. A focal point is the identification of matrine's specific target and the underlying molecular mechanisms governing its analgesic and neuroprotective actions. METHODS: To discern matrine's analgesic effects in CIPN mice, we conducted behavioral experiments encompassing the Von Frey filament test and Hargreaves Test. Furthermore, we conducted electrophysiological and histopathological assessments involving HE staining, Nissl staining, and Fluoro-Jade B staining to evaluate matrine's effects on neuroprotection within dorsal root ganglia and the spinal cord of CIPN mice. Sequentially, thermal shift assay, GTP hydrolysis assay, and nucleotide exchange assay were executed to validate matrine's inhibitory effects on KRAS. Molecular docking and site-directed mutagenesis experiments were implemented to identify the precise binding pocket of matrine on KRAS. Lastly, matrine's inhibitory effects on downstream signaling pathways of KRAS were confirmed through experiments conducted at animal model. RESULTS: Matrine exhibited a notable increase in mechanical withdrawal threshold and thermal withdrawal latency in vincristine-treated mice. This compound substantially ameliorated the neurofunctional blockade associated with sensory and motor functions induced by vincristine. Moreover, matrine mitigated pathological damage within DRG and the L4-L5 spinal cord regions. The study's MST experiments indicated matrine's substantial elevation of KRAS's melting temperature. The GTP hydrolysis and nucleotide exchange assays revealed concentration-dependent inhibition of KRAS activity by matrine. Molecular docking provided insight into the binding mode of matrine with KRAS, while site-directed mutagenesis verified the specific binding site of matrine on KRAS. Lastly, matrine's inhibition of downstream Raf/Erk1/2 and PI3K/Akt/mTOR signaling pathways of KRAS was confirmed in VCR mice. CONCLUSION: Compared to previous studies, our research has identified matrine as a natural inhibitor of the elusive protein KRAS, often considered "undruggable." Furthermore, this study has revealed that matrine exerts its therapeutic effects on chemotherapy-induced peripheral neuropathy (CIPN) by inhibiting KRAS activation, subsequently suppressing downstream signaling pathways such as Raf/Erk1/2 and PI3K/Akt/mTOR. This investigation signifies the discovery of a novel target for matrine, thus expanding the potential scope of its involvement in KRAS-related biological functions and diseases. These findings hold the promise of providing a crucial experimental foundation for forthcoming drug development initiatives centered around matrine, thereby advancing the field of pharmaceutical research.

spatiAlign: an unsupervised contrastive learning model for data integration of spatially resolved transcriptomics.

BACKGROUND: Integrative analysis of spatially resolved transcriptomics datasets empowers a deeper understanding of complex biological systems. However, integrating multiple tissue sections presents challenges for batch effect removal, particularly when the sections are measured by various technologies or collected at different times. FINDINGS: We propose spatiAlign, an unsupervised contrastive learning model that employs the expression of all measured genes and the spatial location of cells, to integrate multiple tissue sections. It enables the joint downstream analysis of multiple datasets not only in low-dimensional embeddings but also in the reconstructed full expression space. CONCLUSIONS: In benchmarking analysis, spatiAlign outperforms state-of-the-art methods in learning joint and discriminative representations for tissue sections, each potentially characterized by complex batch effects or distinct biological characteristics. Furthermore, we demonstrate the benefits of spatiAlign for the integrative analysis of time-series brain sections, including spatial clustering, differential expression analysis, and particularly trajectory inference that requires a corrected gene expression matrix.

GSC: efficient lossless compression of VCF files with fast query.

BACKGROUND: With the rise of large-scale genome sequencing projects, genotyping of thousands of samples has produced immense variant call format (VCF) files. It is becoming increasingly challenging to store, transfer, and analyze these voluminous files. Compression methods have been used to tackle these issues, aiming for both high compression ratio and fast random access. However, existing methods have not yet achieved a satisfactory compromise between these 2 objectives. FINDINGS: To address the aforementioned issue, we introduce GSC (Genotype Sparse Compression), a specialized and refined lossless compression tool for VCF files. In benchmark tests conducted across various open-source datasets, GSC showcased exceptional performance in genotype data compression. Compared with the industry's most advanced tools (namely, GBC and GTC), GSC achieved compression ratios that were higher by 26.9% to 82.4% over GBC and GTC on the datasets, respectively. In lossless compression scenarios, GSC also demonstrated robust performance, with compression ratios 1.5× to 6.5× greater than general-purpose tools like gzip, zstd, and BCFtools-a mode not supported by either GBC or GTC. Achieving such high compression ratios did require some reasonable trade-offs, including longer decompression times, with GSC being 1.2× to 2× slower than GBC, yet 1.1× to 1.4× faster than GTC. Moreover, GSC maintained decompression query speeds that were equivalent to its competitors. In terms of RAM usage, GSC outperformed both counterparts. Overall, GSC's comprehensive performance surpasses that of the most advanced technologies. CONCLUSION: GSC balances high compression ratios with rapid data access, enhancing genomic data management. It supports seamless PLINK binary format conversion, simplifying downstream analysis.

Advances and Challenges in Interfacial Binding Forces for Electrocatalysts.

As a practical chemical energy conversion technology, electrocatalysis could be used in fields of energy conversion and environmental protection. In recent years, significant research efforts have been devoted to the design and development of high-performance electrocatalysts because the rational design of catalysts is crucial for enhancing electrocatalytic performance. Creating electrocatalysts by forming interactions between different components at the interface is an important means of controlling and improving performance. Therefore, several common interfacial binding forces used for synthesizing electrocatalysts was systematically summarized in this review for the first time. The discussion revolves around the crucial roles these binding forces play in various electrocatalytic reaction processes. Various characterization techniques capable of proving the existence of these interfacial binding forces was also involved in the review. Finally, some prospects and challenges for designing and researching materials through the utilization of interfacial binding forces were presented.

Light-Triggered Anti-ambipolar Transistor Based on an In-Plane Lateral Homojunction.

Currently, the construction of anti-ambipolar transistors (AATs) is primarily based on asymmetric heterostructures, which are challenging to fabricate. AATs used for photodetection are accompanied by dark currents that prove difficult to suppress, resulting in reduced sensitivity. This work presents light-triggered AATs based on an in-plane lateral WSe2 homojunction without van der Waals heterostructures. In this device, the WSe2 channel is partially electrically controlled by the back gate due to the screening effect of the bottom electrode, resulting in a homojunction that is dynamically modulated with gate voltage, exhibiting electrostatically reconfigurable and light-triggered anti-ambipolar behaviors. It exhibits high responsivity (188 A/W) and detectivity (8.94 × 1014 Jones) under 635 nm illumination with a low power density of 0.23 µW/cm2, promising a new approach to low-power, high-performance photodetectors. Moreover, the device demonstrates efficient self-driven photodetection. Furthermore, ternary inverters are realized using monolithic WSe2, simplifying the manufacturing of multivalued logic devices.

LAPTM4B counteracts ferroptosis via suppressing the ubiquitin-proteasome degradation of SLC7A11 in non-small cell lung cancer.

Non-small cell lung cancer (NSCLC) is a leading cause of cancer-related deaths worldwide, necessitating the identification of novel therapeutic targets. Lysosome Associated Protein Transmembrane 4B (LAPTM4B) is involved in biological processes critical to cancer progression, such as regulation of solute carrier transporter proteins and metabolic pathways, including mTORC1. However, the metabolic processes governed by LAPTM4B and its role in oncogenesis remain unknown. In this study, we conducted unbiased metabolomic screens to uncover the metabolic landscape regulated by LAPTM4B. We observed common metabolic changes in several knockout cell models suggesting of a role for LAPTM4B in suppressing ferroptosis. Through a series of cell-based assays and animal experiments, we demonstrate that LAPTM4B protects tumor cells from erastin-induced ferroptosis both in vitro and in vivo. Mechanistically, LAPTM4B suppresses ferroptosis by inhibiting NEDD4L/ZRANB1 mediated ubiquitination and subsequent proteasomal degradation of the cystine-glutamate antiporter SLC7A11. Furthermore, metabolomic profiling of cancer cells revealed that LAPTM4B knockout leads to a significant enrichment of ferroptosis and associated metabolic alterations. By integrating results from cellular assays, patient tissue samples, an animal model, and cancer databases, this study highlights the clinical relevance of the LAPTM4B-SLC7A11-ferroptosis signaling axis in NSCLC progression and identifies it as a potential target for the development of cancer therapeutics.

Donor-Acceptor Copolymer with a Linear Backbone Induced Ordered and Robust Doping Morphology for Efficient and Stable Organic Electrochemical Devices.

Donor (D)-acceptor (A) copolymer-based organic mixed ionic-electronic conductors (OMIECs) exhibit intrinsic environmental stability for they have tailored energy levels. However, their figure-of-merit (µC*) is still falling behind the D-D polymers because of morphology deterioration during the electrochemical doping process. Herein, we developed two D-A copolymers with precisely regulated backbone curvature, namely PTBT-P and PTTBT-P. Compared to the curved PTBT-P and previously reported copolymers, PTTBT-P better keeps its backbone linear, leading to a long-range ordered doping morphology, which is revealed by the in operando X-ray technique. This optimized doping morphology enables a significantly improved operando charge mobility (µ) of 2.44 cm2 V-1 s-1 and a µC* value of 342 F cm-1 V-1 s-1, one of the highest values in D-A copolymer based on OECTs. Besides, we fabricated PTTBT-P-based electrochemical random-access memories and achieved ideal and robust conductance modulation. This study highlights the critical role of backbone curvature control in the optimization of doping morphology for efficient and robust organic electrochemical devices.

Photocatalytic Reduction of Perrhenate and Pertechnetate in a Strongly Acidic Aqueous Solution.

Pertechnetate (99TcO4-), a physiologically toxic radioactive anion, is of great concern due to its high mobility in environmental contamination remediation. Although the soluble oxyanion can be photoreduced to sparingly soluble TcO2·nH2O, its effective removal from a strongly acidic aqueous solution remains a challenge. Here, we found that low-crystalline nitrogen-doped titanium oxide (N-TiO2, 0.6 g L-1) could effectively uptake perrhenate (ReO4-, 10 mg L-1, a nonradioactive surrogate for TcO4-) with 50.8% during 360 min under simulated sunlight irradiation at pH 1.0, but P25 and anatase could not. The nitrogen active center formed by trace nitrogen doping in N-TiO2 can promote the separation and transfer of photogenerated carriers. The positive valence band value of N-TiO2 is slightly higher than those of P25 and anatase, which means that the photogenerated holes have a stronger oxidizability. These holes are involved in the formation of strong reducing •CO2- radicals from formic acid oxidation. The active radicals convert ReO4- to Re(VI), which is subsequently disproportionated to Re(IV) and Re(VII). Effective photocatalytic reduction/removal of Re(VII)/Tc(VII) is performed on the material, which may be considered a potential and convenient strategy for technetium decontamination and extraction in a strongly acidic aqueous solution.

Textured Perovskite/Silicon Tandem Solar Cells Achieving Over 30% Efficiency Promoted by 4-Fluorobenzylamine Hydroiodide.

Monolithic textured perovskite/silicon tandem solar cells (TSCs) are expected to achieve maximum light capture at the lowest cost, potentially exhibiting the best power conversion efficiency. However, it is challenging to fabricate high-quality perovskite films and preferred crystal orientation on commercially textured silicon substrates with micrometer-size pyramids. Here, we introduced a bulky organic molecule (4-fluorobenzylamine hydroiodide (F-PMAI)) as a perovskite additive. It is found that F-PMAI can retard the crystallization process of perovskite film through hydrogen bond interaction between F- and FA+ and reduce (111) facet surface energy due to enhanced adsorption energy of F-PMAI on the (111) facet. Besides, the bulky molecular is extruded to the bottom and top of perovskite film after crystal growth, which can passivate interface defects through strong interaction between F-PMA+ and undercoordinated Pb2+/I-. As a result, the additive facilitates the formation of large perovskite grains and (111) preferred orientation with a reduced trap-state density, thereby promoting charge carrier transportation, and enhancing device performance and stability. The perovskite/silicon TSCs achieved a champion efficiency of 30.05% based on a silicon thin film tunneling junction. In addition, the devices exhibit excellent long-term thermal and light stability without encapsulation. This work provides an effective strategy for achieving efficient and stable TSCs.

Haplotype-resolved assembly of diploid and polyploid genomes using quantum computing.

Precision medicine's emphasis on individual genetic variants highlights the importance of haplotype-resolved assembly, a computational challenge in bioinformatics given its combinatorial nature. While classical algorithms have made strides in addressing this issue, the potential of quantum computing remains largely untapped. Here, we present the vehicle routing problem (VRP) assembler: an approach that transforms this task into a vehicle routing problem, an optimization formulation solvable on a quantum computer. We demonstrate its potential and feasibility through a proof of concept on short synthetic diploid and triploid genomes using a D-Wave quantum annealer. To tackle larger-scale assembly problems, we integrate the VRP assembler with Google's OR-Tools, achieving a haplotype-resolved local assembly across the human major histocompatibility complex (MHC) region. Our results show encouraging performance compared to Hifiasm with phasing accuracy approaching the theoretical limit, underscoring the promising future of quantum computing in bioinformatics.

Extraction, Isolation and Biological Activity of Two Glycolipids from Bangia fusco-purpurea.

In order to explore the extraction and activity of macroalge glycolipids, six macroalgae (Bangia fusco-purpurea, Gelidium amansii, Gloiopeltis furcata, Gracilariopsis lemaneiformis, Gracilaria sp. and Pyropia yezoensis) glycolipids were extracted with five different solvents firstly. Considering the yield and glycolipids concentration of extracts, Bangia fusco-purpurea, Gracilaria sp. and Pyropia yezoensis were selected from six species of marine macroalgae as the raw materials for the extraction of glycolipids. The effects of the volume score of methanol, solid-liquid ratio, extraction temperature, extraction time and ultrasonic power on the yield and glycolipids concentration of extracts of the above three macroalgae were analyzed through a series of single-factor experiments. By analyzing the antioxidant activity in vitro, moisture absorption and moisturizing activity, the extraction process of Bangia fusco-purpurea glycolipids was further optimized by response surface method to obtain suitable conditions for glycolipid extraction (solid-liquid ratio of 1:27 g/mL, extraction temperature of 48 °C, extraction time of 98 min and ultrasonic power of 450 W). Bangia fusco-purpurea extracts exhibited a certain scavenging effect on DPPH free radicals, as well as good moisture-absorption and moisture retaining activities. Two glycolipids were isolated from Bangia fusco-purpurea by liquid-liquid extraction, silica gel column chromatography and thin-layer chromatography, and they showed good scavenging activities against DPPH free radicals and total antioxidant capacity. Their scavenging activities against DPPH free radicals were about 60% at 1600 µg/mL, and total antioxidant capacity was better than that of Trolox. Among them, the moisturizing activity of a glycolipid was close to that of sorbierite and sodium alginate. These two glycolipids exhibited big application potential as food humectants and antioxidants.

An all-optical multidirectional mechano-sensor inspired by biologically mechano-sensitive hair sensilla.

Mechano-sensitive hair-like sensilla (MSHS) have an ingenious and compact three-dimensional structure and have evolved widely in living organisms to perceive multidirectional mechanical signals. Nearly all MSHS are iontronic or electronic, including their biomimetic counterparts. Here, an all-optical mechano-sensor mimicking MSHS is prototyped and integrated based on a thin-walled glass microbubble as a flexible whispering-gallery-mode resonator. The minimalist integrated device has a good directionality of 32.31 dB in the radial plane of the micro-hair and can detect multidirectional displacements and forces as small as 70 nm and 0.9 µN, respectively. The device can also detect displacements and forces in the axial direction of the micro-hair as small as 2.29 nm and 3.65 µN, respectively, and perceive different vibrations. This mechano-sensor works well as a real-time, directional mechano-sensory whisker in a quadruped cat-type robot, showing its potential for innovative mechano-transduction, artificial perception, and robotics applications.

A spatiotemporal atlas of mouse liver homeostasis and regeneration.

The mechanism by which mammalian liver cell responses are coordinated during tissue homeostasis and perturbation is poorly understood, representing a major obstacle in our understanding of many diseases. This knowledge gap is caused by the difficulty involved with studying multiple cell types in different states and locations, particularly when these are transient. We have combined Stereo-seq (spatiotemporal enhanced resolution omics-sequencing) with single-cell transcriptomic profiling of 473,290 cells to generate a high-definition spatiotemporal atlas of mouse liver homeostasis and regeneration at the whole-lobe scale. Our integrative study dissects in detail the molecular gradients controlling liver cell function, systematically defining how gene networks are dynamically modulated through intercellular communication to promote regeneration. Among other important regulators, we identified the transcriptional cofactor TBL1XR1 as a rheostat linking inflammation to Wnt/ß-catenin signaling for facilitating hepatocyte proliferation. Our data and analytical pipelines lay the foundation for future high-definition tissue-scale atlases of organ physiology and malfunction.

Effects of Sodium Vacancies and Concentrations in Na3SO4F Solid Electrolyte.

The sodium-rich solid electrolyte, Na3SO4F (NSOF), holds promise for eco-friendly and resource-abundant energy storage. While the introduction of heterovalent dopants has the potential to enhance its suitability for battery applications by creating Na vacancies, the effect of vacancies and sodium concentrations on sodium conduction remains unclear. In this work, Mg2+ was introduced into Na+ sites in Na3SO4F, generating sodium vacancies with different contents by using solid-state synthesis method. Among the resulting materials, Na2.96Mg0.02SO4F exhibited an ionic conductivity that is two-order-of-magnitude higher than NSOF at 298 K. Notably, as the sodium concentration decreased, the ionic conductivity also declined, revealing an equilibrium between Na vacancies and concentrations. To further investigate the influence of sodium concentration, excess Na+ was introduced into NaMgSO4F, which inherently possesses a lower sodium content by using solid-state synthesis method. However, this adjustment only led to an approximately one-order-of-magnitude enhancement in optimal ionic conductivity at 298 K. Combined with an in situ X-ray diffraction analysis, our findings underscore the greater sensitivity of sodium conduction to variations in sodium vacancies. This study paves the way for the development of ultrafast sodium ion conductors, offering exciting prospects for advanced energy storage solutions.

Tailoring the Molecular Planarity of Perylene Diimide-Based Third Component toward Efficient Ternary Organic Solar Cells.

Incorporating a third component into binary organic solar cells (b-OSCs) has provided a potential platform to boost power conversion efficiency (PCEs). However, gaining control over the non-equilibrium blend morphology via the molecular design of the perylene diimide (PDI)-based third component toward efficient ternary organic solar cells (t-OSCs) still remains challenging. Herein, two novel PDI derivatives are developed with tailored molecular planarity, namely ufBTz-2PDI and fBTz-2PDI, as the third component for t-OSCs. Notably, after performing a cyclization reaction, the twisted ufBTz-2PDI with an amorphous character transferred to the highly planar fBTz-2PDI followed by a semi-crystalline character. When incorporating the semi-crystalline fBTz-2PDI into the D18:L8-BO system, the resultant t-OSC achieved an impressive PCE of 18.56%, surpassing the 17.88% attained in b-OSCs. In comparison, the addition of amorphous ufBTz-2PDI into the binary system facilitates additional charge trap sites and results in a deteriorative PCE of 14.37%. Additionally, The third component fBTz-2PDI possesses a good generality in optimizing the PCEs of several b-OSCs systems are demonstrated. The results not only provided a novel A-DA'D-A motif for further designing efficient third component but also demonstrated the crucial role of modulated crystallinity of the PDI-based third component in optimizing PCEs of t-OSCs.

5-HT7R enhances neuroimmune resilience and alleviates meningitis by promoting CCR5 ubiquitination.

INTRODUCTION: Excessive immune activation induces tissue damage during infection. Compared to external strategies to reconstruct immune homeostasis, host balancing ways remain largely unclear. OBJECTIVES: Here we found a neuroimmune way that prevents infection-induced tissue damage. METHODS: By FACS and histopathology analysis of brain Streptococcus pneumonia meningitis infection model and behavioral testing. Western blot, co-immunoprecipitation, and ubiquitination analyze the Fluoxetine initiate 5-HT7R-STUB1-CCR5 K48-linked ubiquitination degradation. RESULTS: Fluoxetine, a selective serotonin reuptake inhibitor, or the agonist of serotonin receptor 5-HT7R, protects mice from meningitis by inhibiting CCR5-mediated excessive immune response and tissue damage. Mechanistically, the Fluoxetine-5-HT7R axis induces proteasome-dependent degradation of CCR5 via mTOR signaling, and then recruits STUB1, an E3 ubiquitin ligase, to initiate K48-linked polyubiquitination of CCR5 at K138 and K322, promotes its proteasomal degradation. STUB1 deficiency blocks 5-HT7R-mediated CCR5 degradation. CONCLUSION: Our results reveal a neuroimmune pathway that balances anti-infection immunity via happiness neurotransmitter receptor and suggest the 5-HT7R-CCR5 axis as a potential target to promote neuroimmune resilience.

SAW: an efficient and accurate data analysis workflow for Stereo-seq spatial transcriptomics.

The basic analysis steps of spatial transcriptomics require obtaining gene expression information from both space and cells. The existing tools for these analyses incur performance issues when dealing with large datasets. These issues involve computationally intensive spatial localization, RNA genome alignment, and excessive memory usage in large chip scenarios. These problems affect the applicability and efficiency of the analysis. Here, a high-performance and accurate spatial transcriptomics data analysis workflow, called Stereo-seq Analysis Workflow (SAW), was developed for the Stereo-seq technology developed at BGI. SAW includes mRNA spatial position reconstruction, genome alignment, gene expression matrix generation, and clustering. The workflow outputs files in a universal format for subsequent personalized analysis. The execution time for the entire analysis is ∼148 min with 1 GB reads 1 × 1 cm chip test data, 1.8 times faster than with an unoptimized workflow.

Generating single-cell gene expression profiles for high-resolution spatial transcriptomics based on cell boundary images.

In spatially resolved transcriptomics, Stereo-seq facilitates the analysis of large tissues at the single-cell level, offering subcellular resolution and centimeter-level field-of-view. Our previous work on StereoCell introduced a one-stop software using cell nuclei staining images and statistical methods to generate high-confidence single-cell spatial gene expression profiles for Stereo-seq data. With advancements allowing the acquisition of cell boundary information, such as cell membrane/wall staining images, we updated our software to a new version, STCellbin. Using cell nuclei staining images, STCellbin aligns cell membrane/wall staining images with spatial gene expression maps. Advanced cell segmentation ensures the detection of accurate cell boundaries, leading to more reliable single-cell spatial gene expression profiles. We verified that STCellbin can be applied to mouse liver (cell membranes) and Arabidopsis seed (cell walls) datasets, outperforming other methods. The improved capability of capturing single-cell gene expression profiles results in a deeper understanding of the contribution of single-cell phenotypes to tissue biology. Availability & Implementation: The source code of STCellbin is available at https://github.com/STOmics/STCellbin.

The OsEIL1-OsWOX11 transcription factor module controls rice crown root development in response to soil compaction.

Optimizing the root architecture of crops is an effective strategy for improving crop yields. Soil compaction is a serious global problem that limits crop productivity by restricting root growth, but the underlying molecular mechanisms are largely unclear. Here, we show that ethylene stimulates rice (Oryza sativa) crown root development in response to soil compaction. First, we demonstrate that compacted soil promotes ethylene production and the accumulation of ETHYLENE INSENSITIVE 3-LIKE 1 (OsEIL1) in rice roots, stimulating crown root primordia initiation and development, thereby increasing crown root number in lower stem nodes. Through transcriptome profiling and molecular analyses, we reveal that OsEIL1 directly activates the expression of WUSCHEL-RELATED HOMEOBOX 11 (OsWOX11), an activator of crown root emergence and growth, and that OsWOX11 mutations delay crown root development, thus impairing the plant's response to ethylene and soil compaction. Genetic analysis demonstrates that OsWOX11 functions downstream of OsEIL1. In summary, our results demonstrate that the OsEIL1-OsWOX11 module regulates ethylene action during crown root development in response to soil compaction, providing a strategy for the genetic modification of crop root architecture and grain agronomic traits.

An Entity Extraction Pipeline for Medical Text Records Using Large Language Models: Analytical Study.

BACKGROUND: The study of disease progression relies on clinical data, including text data, and extracting valuable features from text data has been a research hot spot. With the rise of large language models (LLMs), semantic-based extraction pipelines are gaining acceptance in clinical research. However, the security and feature hallucination issues of LLMs require further attention. OBJECTIVE: This study aimed to introduce a novel modular LLM pipeline, which could semantically extract features from textual patient admission records. METHODS: The pipeline was designed to process a systematic succession of concept extraction, aggregation, question generation, corpus extraction, and question-and-answer scale extraction, which was tested via 2 low-parameter LLMs: Qwen-14B-Chat (QWEN) and Baichuan2-13B-Chat (BAICHUAN). A data set of 25,709 pregnancy cases from the People's Hospital of Guangxi Zhuang Autonomous Region, China, was used for evaluation with the help of a local expert's annotation. The pipeline was evaluated with the metrics of accuracy and precision, null ratio, and time consumption. Additionally, we evaluated its performance via a quantified version of Qwen-14B-Chat on a consumer-grade GPU. RESULTS: The pipeline demonstrates a high level of precision in feature extraction, as evidenced by the accuracy and precision results of Qwen-14B-Chat (95.52% and 92.93%, respectively) and Baichuan2-13B-Chat (95.86% and 90.08%, respectively). Furthermore, the pipeline exhibited low null ratios and variable time consumption. The INT4-quantified version of QWEN delivered an enhanced performance with 97.28% accuracy and a 0% null ratio. CONCLUSIONS: The pipeline exhibited consistent performance across different LLMs and efficiently extracted clinical features from textual data. It also showed reliable performance on consumer-grade hardware. This approach offers a viable and effective solution for mining clinical research data from textual records.