196 results about "Medical research" patented technology

An intelligent analysis method based on medical document structure perception and multi-modal fusion comprises the following steps: carrying out structure topology modeling on a medical document, extracting visual layout, text meta-information, space coordinates and semantic keyword features, constructing a semantic topological graph and dynamically shielding irrelevant contents; selecting an extraction path according to a document type, performing deep semantic analysis and entity recognition on a text-type document, and performing visual enhancement OCR recognition on a scanning-type document; the features are injected into a medical knowledge graph, and feature fusion, semantic verification, relation reasoning and information completion are achieved through a graph neural network; a three-stage strategy optimization model of basic pre-training, domain adaptation and online reinforcement learning is adopted; and large-scale processing is realized through a dynamically aggregated distributed architecture. The method is used for intelligent analysis and structured conversion of documents of hospitals, medical insurance and medical scientific research. The problems that heterogeneous medical document analysis adaptability is poor, multi-modal fusion is difficult, medical knowledge utilization is insufficient, and large-scale processing efficiency is low are solved.

A spatial omics multi-modal fusion method under a single cell level comprises the following steps: extracting spatial morphological characteristics of differential expression genes and cell nucleuses from spatial transcriptome data, single cell sequencing data and histological images, and realizing field adaptation among different platforms by using a conditional variation auto-encoder. And based on a probability inference model, fusing spatial transcriptome expression, unicellular omics and morphological characteristics, and jointly inferring the type and gene expression level of each cell. A spatial cell network is constructed through a graph attention mechanism, and spatial diffusion and recognition of cell types in a full slice range are realized. In combination with a multi-omics enhancement module, undetected gene and protein expression is completed based on expression similarity, and prediction consistency is improved through spatial correction. According to the method, high-resolution reconstruction of single-cell multi-omics information in a three-dimensional space is realized, the information coverage and spatial resolution of spatial omics data are improved, and an efficient and low-cost solution is provided for spatial biology and precise medical research.

The invention relates to the technical field of ICU signal processing, and discloses a signal data cleaning method and system based on ICU monitoring. The method comprises the following steps: receiving original data streams of different vital sign monitoring devices through a multi-channel physiological signal acquisition interface, and converting the original data streams into standardized data with a unified timestamp, sampling frequency and amplitude dimension; evaluating the signal stability of the standardized data in real time, triggering multi-dimensional feature extraction when the signal stability is lower than a preset threshold value, and obtaining time domain, frequency domain and nonlinear feature vectors; inputting the feature vector into an anomaly detection model, and outputting an anomaly confidence score and a type label; calling a dynamic cleaning strategy, executing interpolation restoration, noise suppression or abnormal segment elimination, and generating cleaned data; finally, self-adaptive reconstruction and storage optimization are carried out on the cleaned data, final storage signal data are generated, the ICU monitoring data quality and utilization efficiency can be improved, and the requirements of clinical diagnosis and treatment and medical research are met.

The invention discloses a Meta analysis method and device, equipment and a storage medium, according to the scheme, automatic semantic analysis is carried out on a research problem input by a user through a preset large language model, target literature is processed through a pre-training language model, an efficient and reliable automatic tool is provided for evidence-based medicine, and the efficiency is improved. According to the method, the dependency of Meta statistical analysis on professionals is reduced, the analysis period is shortened, personal errors are reduced, the intelligent degree, efficiency, accuracy and repeatability of Meta statistical analysis are remarkably improved, and standardized and intelligent development of evidence-based medical research is promoted; in addition, retrieval is carried out based on the target structured form, the retrieval strategy is optimized, the retrieval accuracy is improved, and therefore the performance and accuracy of Meta statistical analysis are further improved.

The invention discloses a magnetic source positioning method based on a spin-exchange relaxation-free atom magnetometer, and aims to provide a high-precision brand-new solution for magnetic source positioning in a near-zero magnetic environment by performing physical modeling and least square estimation on space magnetic field distribution of a target magnetic source. The magnetometer is enabled to be simultaneously sensitive to a three-axis magnetic field, a light absorption detection method is utilized to ensure that complete three-axis magnetic field information is obtained, and the sensitivity of the three axes reaches fT / Hz1 / 2 (1fT = 10 <-15 > T) magnitude. In the positioning process, the target magnetic source is abstracted into a physical model of the magnetic dipole, and the space magnetic field information generated by the target magnetic source is obtained through the atom magnetometer, so that the long-term stability of the magnetic source positioning process and the high precision of the positioning result can be ensured in the zero magnetic environment; therefore, a high-precision magnetic positioning technical means with great potential is provided for various application scenes such as paleomagnetic sample testing and magnetic source positioning in medical research, and technical progress and application expansion in related fields are expected to be promoted.

The invention relates to the technical field of image processing, and discloses a fluorescent lifetime microscopic image large-view-field splicing method and system and electronic equipment, and the method comprises the steps: constructing a tissue region mask for a to-be-spliced image, and constructing a vignetting model in a tissue region to complete vignetting correction; counting brightness indexes in an organization area to determine a global brightness reference, and realizing image group brightness unification through global zooming and single image brightness adaptive correction; further realizing alignment of adjacent images through feature point matching, and finishing image block splicing in combination with a minimum color difference suture fusion method to obtain an image band; an overlapping area is extracted from an image belt to generate an effective content mask, relative displacement is estimated by using a phase correlation method after low-frequency suppression, image belt splicing is completed by multiplexing a minimum color difference suture fusion method, a large-view-field spliced image is output, the automation degree and robustness of splicing are improved, the spliced image is geometrically consistent and visually seamless, and the splicing efficiency is improved. And the requirements of medical research on high-resolution and large-field-of-view fluorescence lifetime microscopic images are met.

The invention discloses an anorectal postoperative rehabilitation intelligent supervision method and system based on multiple modes, and belongs to the technical field of intelligent supervision. A temperature, humidity, air quality and other multi-mode sensor group is arranged in each anorectal postoperative rehabilitation management inpatient area, the body temperature, heart rate and wound surface images of a patient are synchronously collected, and the wound surface area and seepage color are extracted through image processing; the method comprises the following steps: by taking a time node as a unit, adding a time label to an environment and patient basic data set, constructing a multi-modal comprehensive data set in a unified format, calculating an environment comprehensive index and a physiological comprehensive index, fusing the environment comprehensive index and the physiological comprehensive index to calculate a postoperative rehabilitation risk value, setting a threshold interval to generate low, medium and high three-level risk labels, and the patient risk label set is pushed to the doctor end in real time, a dynamic supervision system is constructed, the anomaly recognition rate and the clinical response speed are improved, the complication occurrence rate is reduced, the rehabilitation period is shortened, and reliable data support is provided for nursing quality improvement and evidence-based medical research.

The invention relates to the technical field of medical data security, and discloses an electronic medical record data security sharing method based on a block chain. The method comprises the steps of receiving original data of different medical institutions through a multi-source medical data acquisition module and performing standardized cleaning to eliminate data format differences; generating a medical data block chain network by adopting a distributed account book construction module, and generating data blocks from the standardized data according to a time sequence; based on an attribute-based encryption strategy generation module, dynamically configuring patient node access permission granularity, and defining a medical record access range and a data service life of each node; calling the encrypted storage contract through an intelligent contract execution module, writing the electronic medical record into a block, and generating a chained hash verification value; when an external query request is received, the zero-knowledge proof verification module is triggered to generate validity verification parameters. According to the method, safe and controllable sharing of electronic medical records is realized, and cross-mechanism medical cooperation and medical research development are assisted.

The invention relates to the technical field of image modeling, in particular to a three-dimensional modeling method and system for medical image data, and the method comprises the steps: carrying out the preprocessing of the obtained two-dimensional or three-dimensional medical image data, extracting a blood vessel region based on the image texture and gray features, and generating a preliminary blood vessel structure segmentation image; in combination with a blood flow simulation or real acquisition means, obtaining dynamic feature data from the vascular structure segmentation map, and mapping the dynamic feature data to a corresponding vascular structure position to form a vascular distribution map with blood flow constraint; and inputting the blood vessel distribution diagram into a three-dimensional modeling module, expanding and complementing small blood vessel branches by utilizing blood flow constraint, dynamically correcting the form of a main blood vessel, and generating a three-dimensional blood vessel structure model with real fluid physiological characteristics. The method can be widely applied to medical research and clinical aid decision-making scenes such as preoperative simulation, individualized intervention path planning and blood flow reconstruction analysis, and has good popularization and application prospects and engineering practical value.

The invention relates to the technical field of medicines, in particular to application of lasiokaurin and cis-platinum combined medicine in preparation of a medicine for treating liver cancer. As a diterpenoid compound, lasiokaurin shows significant potential in the anti-cancer field. However, in medical research and clinical application at present, related reports about combined use of lasiokaurin and a classical anti-cancer drug cis-platinum for treating liver cancer do not exist. Based on the current research situation, the invention creatively provides the application of the lasiokaurin and cis-platinum combined medicine in the aspect of preparing the liver cancer treatment medicine. In-depth study finds that lasiokaurin and cis-platinum can generate a synergistic effect and jointly play a strong anti-tumor role, and a brand new way is opened up for treatment of liver cancer.

The invention provides a target character prediction method suitable for low-depth sequencing and a whole genome selection model suitable for low-depth sequencing, and relates to the technical field of biology. According to the target character prediction method suitable for low-depth sequencing and the whole-genome selection model provided by the invention, the target character of the to-be-detected sample can be accurately predicted under a limited data volume by combining low-depth whole-genome re-sequencing data with a machine learning technology, and the genome selection efficiency can be improved by matching the two; the time and economic cost of breeding or medical research are reduced, and the method is particularly suitable for large-scale genetic analysis scenes needing rapid iteration.

The invention discloses a training method, system and device of a disease auxiliary diagnosis model and a medium. The method comprises the steps of preprocessing local multi-modal medical data of each mechanism; performing adversarial training on the multi-modal medical data through a generator and a discriminator to generate a high-fidelity synthetic sample; updating gradient parameters of the disease auxiliary diagnosis model based on the synthetic sample; the gradient parameters are encrypted and uploaded to a federal center server; the server aggregates the gradients and issues the gradients to each mechanism; and generating a disease auxiliary diagnosis model after iterative optimization. According to the method, scarce samples are supplemented through multi-modal data standardization and adversarial generation, cross-institution privacy protection cooperation is realized in combination with federated learning and a differential privacy technology, a global optimal model can be constructed without sharing original data, the model diagnosis accuracy and generalization ability are improved, the problem of data imbalance is effectively relieved, and the user experience is improved. And clinical auxiliary diagnosis and medical scientific research requirements are met.

The invention provides a multi-agent-driven pet medical research literature generation method and device, and the method comprises the steps: obtaining pet medical research task information inputted by a user; based on the pet medicine research task information, identifying a user task intention through a master control agent, and calling a language model to perform structured understanding on the user task intention to obtain a task analysis result; performing task division on the task analysis result to obtain a plurality of sub-tasks, and calling a plurality of sub-agents to process the plurality of sub-tasks in parallel or in series to obtain a plurality of pieces of return information corresponding to the plurality of sub-tasks; wherein each sub-agent has independent context memory and tool use permission, and the plurality of sub-agents share the intermediate result; based on a plurality of pieces of return information, deep pet medical research literatures are generated by writing the intelligent agent, so that more flexible and intelligent research strategy planning is realized, and the adaptability and task coverage capability of the intelligent agent system are remarkably improved.

According to the multi-source retrieval and agent collaborative decision-making medical question-answering system and method, a three-layer retrieval framework is constructed, a local medical knowledge base vector retriever, a PubMed intelligent retriever and an academic deep crawling device are integrated, dynamic fusion of multi-source heterogeneous medical information is achieved, and the medical question-answering efficiency is improved. Through the synergistic effect of the three, the information coverage range is remarkably expanded, more comprehensive and multi-dimensional evidence support is provided for various medical problems, the retrieval efficiency and accuracy of a professional database are effectively improved, and the use threshold of non-professional users is reduced. Meanwhile, the technical scheme of the invention supports dual-scene task processing, can meet diversified medical information requirements, provides reliable reference for diagnosis and treatment decisions of medical staff, also provides structured and high-quality information support for subject investigation of medical researchers, and powerfully promotes intelligence and precision of medical information processing.

This invention discloses a method for constructing an online prediction website based on cis-regulatory elements of biological sequences. This method is applicable to building online prediction websites from models trained using machine learning on biological sequences. With the continuous development of artificial intelligence technology, machine learning algorithms have flourished and been successfully applied in multiple fields. In particular, in the fields of biological and medical research, researchers have constructed numerous prediction models to serve various research tasks, such as the prediction of cis-regulatory elements, biomedical pathogenic elements, and proteins. Although many prediction models have been successfully constructed, professionals who frequently use computers are more concerned with how to apply these models to practical work. Therefore, constructing the prediction models in this paper into user-friendly and easy-to-use online websites can greatly enhance the practical application value of the models.

The present application relates to the technical field of medical image computing and artificial intelligence, in particular to a sugar metabolism cross-modal image generation method based on a WARPNet framework, based on the WARPNet framework, through three core steps of multi-scale feature coding, weighted attention cross-modal reasoning and context perception generation, precise synthesis from MRI to PET is realized. The innovation lies in introducing a learnable sugar metabolism semantic prior weight in cross-modal reasoning, modulating feature attention, thereby explicitly guiding the model to focus on key metabolic areas, effectively solving the semantic gap problem between anatomical and functional modalities. The method improves the quality, physiological reasonableness and cross-dataset generalization ability of the synthesized image, and has application potential in auxiliary diagnosis, treatment planning and medical research.

The utility model relates to the technical field of medical equipment, in particular to a microbial incubator for medical research, which comprises an incubator, an LED (light-emitting diode) lamp, a light-transmitting opening, a shading sheet, a culture drawer and a controller, a rectangular groove is formed in the outer side wall of the incubator at the culture drawer, and a storage groove is formed in the inner side wall of the rectangular groove. According to the anti-dazzling screen, through the arranged limiting assembly, after the shifting block is pushed to adjust the position of the anti-dazzling screen, the shifting block can be limited through the limiting assembly, the shifting block is prevented from moving due to external force, and the situation that the position of the shielding plate is changed is avoided; a temperature and humidity control panel can be conveniently protected through the arranged protection assembly, and the situation that the set temperature and humidity are changed due to the fact that the temperature and humidity control panel is easily touched by mistake in the cultivation process is prevented; and the phenomenon of influencing the culture of microorganisms also occurs.

The application provides a magnetocardiogram registration system and method based on optical scanning, and the system comprises a bed body registration module and a CT image registration module. The bed body registration module comprises a first arc-shaped marker and an L-shaped base, and the CT image registration module comprises a second arc-shaped marker which is consistent in shape with the first arc-shaped marker of the bed body registration module. Both the bed body registration module and the CT image registration module are made of non-magnetic photosensitive resin material. Before magnetocardiogram (MCG) acquisition, the two registration modules are digitized by using a handheld optical scanner. According to the bed body registration module, the sensor array coordinate system and the human body coordinate system are aligned. When CT scanning is performed, the image structure of the trunk part of a subject is obtained. According to the CT image registration module, the human body coordinate system and the CT coordinate system are aligned. Correspondingly, the application also provides a registration method corresponding to the system, which significantly improves the efficiency and accuracy of the MCG registration process and promotes the subsequent medical research and application of MCG.

The present invention pertains to a system and method for a medical research instrument device consisting of instrument equipment and computer controls. The instrument equipment consists of electronic boards that are used to control and monitor the various actuators and sensors within the instrument. Software is executed on the computer to drive the human interface and control assay methods. The present invention uses dielectrophoresis ("DEP") to preferentially collect specific particles from a biological sample. The type of particles collected are driven by the parameters used to define the electric field used in DEP. The device passes the samples and various reagents across a patterned silicon die capable of creating the electric fields used within the DEP process.

This invention relates to the field of data processing technology and specifically discloses a method for constructing a clinical terminology knowledge graph. The method includes: acquiring multi-source clinical data and medical domain dictionary data; determining the clinical entity set and data block dictionary set for each clinical data block of each medical clinical data from each data source; determining multiple business domain clustering data for each business domain label and the clustered medical entity set for each business domain clustering data; calculating the statistical entity relationship set and graph entity relationship set between every two medical entities in the clustered medical entity set of each business domain label; and constructing a clinical knowledge graph. This method enables deep collaboration between multi-source clinical data and medical dictionaries, improving the professionalism and reliability of medical entity associations, balancing the independence of business domain knowledge with the interoperability of global knowledge, and providing high-quality core support for intelligent medical assistance, medical research, and clinical knowledge transmission.

The invention belongs to the technical field of animal model construction, and particularly provides a construction method and application of a mouse non-alcoholic fatty liver model, and the construction method comprises the following steps: S1, selecting an experimental mouse; s2, a high nitrogen compound aqueous solution with the mass concentration of 4-10 mg / mL is prepared; s3, according to the dosage, the dosage and the administration time, the prepared high-nitrogen compound aqueous solution is fed through the intragastric administration, and the non-alcoholic fatty liver animal model is obtained, the problems that an existing non-alcoholic fatty liver animal model is long in modeling time and expensive in modeling are solved, and the non-alcoholic fatty liver animal model does not need to be fed with high-fat feed, and is simple, convenient and feasible, low in cost, good in repeatability and stable; the establishment period of the non-alcoholic fatty liver animal model can be effectively shortened, the time cost of the medical research of the NAFLD is greatly saved, and a good animal model is provided for researching the pathogenesis and treatment strategy of the NAFLD.

The invention relates to the technical field of medical literature data processing and artificial intelligence, and discloses a random contrast test (RCT) bias risk automatic evaluation method and system, and the method comprises the steps: obtaining RCT literature data containing a text, a table and a flow chart, and a RoB2 evaluation problem set; a problem reconstruction module injected through medical knowledge; performing position decision and multi-modal coordination analysis on the RCT literature through a multi-modal document analysis module; the reconstructed evaluation problem and the analyzed context information are input into a multi-expert cooperation module, cooperative reasoning is performed through a multi-reasoning unit, results are fused through a decision unit, and item-level, field-level and overall bias risk evaluation results are generated. The invention also constructs bias risk entry level and domain level data sets for supporting the training and verification of the method. The system comprises a data acquisition module, a problem reconstruction module, a document analysis module and a collaborative evaluation module, all the modules work cooperatively, automatic evaluation of RoB2 whole-process bias risks is achieved, and the system can be applied to scenes such as evidence-based medical research, system evaluation and clinical research quality control.

The invention belongs to the technical field of experimental animal models, biomedical research and immunology, and discloses a photocoagulation thymus ischemia animal model, a method and application of the photocoagulation thymus ischemia animal model in rapid simulation of thymus degeneration and immune senescence, and the model shows at least one or more characteristics 1-7 days after induction: thymus atrophy, thymus weight or thymus index is significantly reduced; the cortical-medullary structure of thymus is disordered or disappears; the expression distribution of the Keratin 5 / Keratin 8 is abnormal; thymic cell apoptosis is increased, and lipid deposition is enhanced; the proportion of CD4 + CD8 + double positive T cells is reduced, and the proportion of DN2 / DN3 stage T cells is increased; the initial T cell output is reduced; and the immune function is reduced. The pathological mechanism is clear; the modeling period is short, and the immune senescence phenotype can be stably obtained within several days; the injury focus is controllable, and systemic stress or hormone interference is avoided; the model has high repeatability and is suitable for mechanism research and drug screening.

The invention discloses a millimeter wave radar-based pressure sore prevention early warning method and system, and the method comprises the steps: collecting a radar echo signal of a bed body region through a millimeter wave radar, and generating point cloud data containing distance and speed information; whether the patient is in bed is judged; when it is judged that the patient is in bed, processing the point cloud data to track the trunk and limbs of the patient; based on the tracking result, large-amplitude body movement and small-amplitude body movement of the patient are recognized through a multi-feature fusion body movement algorithm, and the effectiveness of body position change is judged; and according to the body movement identification and validity judgment result, performing pressure sore risk assessment, and sending out early warning information when the risk exists. According to the method, the defects of a traditional contact sensor and artificial nursing are overcome. Complete non-contact monitoring is achieved, direct contact with the body of a patient is avoided, discomfort, oppression and skin stimulation are eliminated, the risk of cross infection is reduced, information transmission timeliness and response speed are improved, and support is provided for data-driven decision making and medical research.

The invention is suitable for the technical field of stem cell grading, and provides a stem cell morphology automatic grading method and system based on a convolutional neural network, and the method comprises the steps: obtaining a preprocessed stem cell image; training a convolutional neural network model based on the marked stem cell image data set; a to-be-graded stem cell image is input into a trained convolutional neural network model, a grading result is automatically output and displayed, cell image features are automatically extracted through the convolutional neural network, a data enhancement technology is utilized to expand a training data set, an advanced optimization algorithm is adopted to carry out model training, and finally the grading accuracy of 95% or above is achieved. The method greatly improves the efficiency and accuracy of morphological analysis of stem cells, and shows huge value in the fields of medical research, clinical diagnosis, drug research and development and the like.

The invention discloses a medical research data layered exploration method and device, electronic equipment and a medium. The method comprises the following steps: identifying a project demand document of a medical research project through a demand identification model, determining a data demand of the medical research project, and determining a standard library table matched with the data demand; generating a first-layer structured query statement and a second-layer structured query statement according to a data demand through a statement generation model; based on the standard library table, executing the first-layer structured query statement, and verifying the first query result through a result verification model; if the first query result passes the verification, executing a second-layer structured query statement on the first query result, and verifying a second query result through a result verification model; and taking the second query result passing the verification as a first probing result. The efficiency of medical research data exploration can be remarkably improved, and it is ensured that the first exploration result not only meets the overall requirements of a project, but also has high accuracy and integrity.

The invention belongs to the technical field of cell culture, and discloses a cell culture device for an experiment for treating cerebral ischemia stroke based on long non-coding RNA, the cell culture device comprises an incubator, a cross-shaped partition plate is fixedly connected in the incubator, a culture mechanism is arranged in the incubator, the culture mechanism comprises four stabilizing assemblies and a temperature control assembly, and the temperature control assembly is connected with the cross-shaped partition plate. The stabilizing assembly and the temperature control assembly are matched for use, the stabilizing assembly comprises sliding rails, the sliding rails are fixedly connected to the interior of the incubator, and sliding blocks are symmetrically and slidably connected to the interiors of the two sliding rails. By arranging the culture mechanism, stable, independent and environment-controllable parallel culture is carried out on a plurality of experimental samples in the same culture box, the culture dish can be stably clamped and fixed to prevent accidental overturning, and the culture dish has the partitioned independent temperature control capability; therefore, stable, reliable and flexible hardware support is provided for related frontier biomedical research.

The utility model relates to the technical field of experiment operation tables, and discloses an experiment operation table main body for biomedical research engineering, which comprises two protection boxes, a cleaning assembly and a control assembly, the upper surfaces of the two protection boxes are provided with chutes communicated with the interiors of the protection boxes, the cleaning assembly is arranged on the two protection boxes, and the cleaning assembly comprises two servo motors. The two servo motors are fixedly installed on the front surfaces of the two protection boxes correspondingly, the output ends of the two servo motors are fixedly connected with threaded rods correspondingly, and the rear ends of the two threaded rods rotationally penetrate into the corresponding protection boxes correspondingly and are rotationally connected with the inner walls of the corresponding protection boxes correspondingly. By arranging the cleaning assembly, automatic cleaning of the surface of the operation panel can be achieved, workers do not need to manually operate a cleaning structure to clean the operation panel, the operation panel can be disinfected and secondarily cleaned, the cleaning quality is high, and then the practicability of the experiment operation table body of the biomedical research engineering is improved.

The invention relates to the field of basic medicine research, in particular to a basic medicine pathogen culture dish which comprises an environment regulation and control module, a multi-dimensional monitoring module, a data analysis module, an early warning and feedback module and a data management module. Multi-parameter cooperative control is realized through a gas sensor and a temperature and humidity sensor, multi-dimensional monitoring is performed by combining optical microscopic imaging and chemical sensing technologies, and real-time data analysis and prediction are realized by using an embedded processor and a machine learning algorithm. The device can dynamically balance the gas concentration, temperature and humidity, comprehensively monitor the growth state of pathogens, improve the experiment efficiency and data reliability, and provide an efficient and accurate culture device solution for basic medical research.

The application provides an RNA sequence classification method based on a Mamba model and semi-supervised learning, and belongs to the field of bioinformatics. First, multi-scale sparse features are extracted from the RNA sequence, high-dimensional features are compressed into low-dimensional potential space vectors through an encoder network, and L2 normalization is performed to enhance feature separability. Second, a Mamba model based on a selective state space model is introduced in a residual structure to efficiently model long-range dependencies with low complexity. Finally, a semi-supervised learning of unsupervised feature reconstruction and supervised classification is realized by constructing an encoder-decoder structure, and a weighted loss function is used to improve the generalization performance of the model under small sample conditions. Experimental results show that the method significantly improves the F1 score in the multi-class RNA sequence classification task. The application reduces the cost of long sequence modeling and labeling, and is suitable for the fields of RNA function prediction, biomedical research and disease diagnosis.