35 results about "Biomedical information" patented technology

The invention discloses a method and system for analyzing a bacterium-drug interaction panoramic dynamic mechanism by using a causal enlightenment hypergraph neural network, and electronic equipment, and belongs to the technical field of artificial intelligence and biomedical informatics. The method comprises the following steps: constructing a disease / syndrome / disease-bacterium / microorganism / metabolite-prescription / drug / component interlocking hypergraph; analyzing a bacteria-drug interaction dynamic mechanism based on a dynamic hypergraph network algorithm; optimizing the hypergraph neural network based on the prescription compatibility rule and the traditional Chinese medicine dosage; and predicting the microbial marker based on the causal enlightenment hypergraph neural network. According to the method, a multi-entity high-order relation is captured through an interlocking hypergraph structure, a time sliding window and an attenuation mechanism are introduced to analyze dynamic evolution, the traditional Chinese medicine dosage is used as an attention priori optimization network, and causal reasoning is achieved through alternate optimization of an intervention discovery and response prediction module. In a public data set test, the performance evaluation index is obviously superior to that in the prior art, and the problem of analyzing the panoramic dynamic mechanism of the bacterium-drug interaction is effectively solved.

The present disclosure generally relates to an integrated approach for retrieving biomedical information from clinical video presentations. In particular, the present disclosure is directed to video retrieval systems and methods of text-video retrieval from clinical video presentations.

The invention relates to the technical field of biomedical information, in particular to an umbilical cord mesenchymal stem cell donor matching database construction method based on HLA (human leukocyte antigen) typing, which comprises the following steps of: recording and storing data of a donor in a database to finish initial construction of the database; regularly rechecking the donor data record stored in the database, and updating the state identifier of the donor data record when the rechecked data is changed by a preset degree; and receiving a matching request containing the human leukocyte antigen typing data of the target patient and the target disease type, performing donor screening in the database by using a preset comprehensive matching degree model based on the matching request, and optimizing parameters of the comprehensive matching degree model through a machine learning model based on the matching record and the clinical curative effect data corresponding to the matching record. According to the method, the accuracy and effectiveness of umbilical cord mesenchymal stem cell treatment are improved, and meanwhile, the system has continuous self-evolution ability.

The application discloses a kind of gene microarray data analysis processing methods, it is related to biomedical information processing technical field, comprising: using microarray technology, obtain gene expression data;Multiple low-dimensional feature selection tasks are constructed;Optimal feature subset is obtained by multi-task pseudo-affine transformation algorithm;According to optimal feature subset, neural network model is trained;Analysis processing is carried out to the gene to be predicted and whether it is predicted to be ill.The gene microarray data analysis processing method of the application is analyzed to gene microarray data characteristics, and gene microarray data is helpful to dig disease characteristic gene information, and has a key role to early detection, clinical treatment and disease prevention of disease.

The invention discloses a cross-type biomedical named entity recognition method and device based on knowledge distillation in the technical field of natural language processing and biomedical information processing. The method comprises the following steps: firstly, acquiring a plurality of single-type data sets for different biomedical entity types, and independently training a teacher model for each data set; meanwhile, gathering all single-type data into a unified training set, and constructing a unified multi-entity type label space; then, for the data in the unified training set, obtaining prediction results of all teacher models, and fusing prediction distributions of a plurality of teacher models into a unified aggregation probability distribution (namely a soft label) by adopting a probability aggregation strategy based on an independence hypothesis; and finally, constructing a student model, and training the student model by using a mixed loss function jointly formed by the aggregation probability distribution and the original label. According to the invention, through knowledge fusion and compression, the problem of label conflict in multi-entity type learning is effectively solved, high identification precision is ensured, a lightweight student model is supported, and the balance between identification performance and calculation requirements is realized.

The application discloses a thyroid cancer gene detection report intelligent risk assessment method, and relates to the technical field of biomedical information processing, and comprises the following steps: risk conclusion extraction is carried out on obtained multiple thyroid cancer gene detection reports, and the risk conclusion in each detection report is arranged in order of detection time, the time interval between adjacent detection reports is marked, and continuous risk change records are formed. Through time sequence processing and trend reversal identification, the application realizes dynamic analysis of risk assessment, can capture reversal signals in the early stage of risk increase, and improves the timeliness and reliability of risk identification. By setting a reference point and a risk control entrance, combining time interval insertion and detection frequency improvement, the application realizes adaptive correction and continuous updating of risk results, and enhances the real-time performance and safety of risk assessment.

The invention belongs to the technical field of artificial intelligence and biomedical information, and particularly relates to a biomedical knowledge extraction, fusion and sharing method and system based on federal map data. The biomedical knowledge extraction, fusion and sharing method based on federal map data comprises the following steps: S10, deploying localized mechanism nodes in a participating medical mechanism, and establishing a communication network architecture between each mechanism node and a central node; the mechanism nodes jointly construct a central knowledge graph according to the stored biomedical knowledge, and the central node stores the central knowledge graph; and S20, the mechanism node obtains the diagnosis and treatment data of the local patient. According to the scheme, through a communication network architecture and multi-modal processing, the problem that the privacy risk is high during patient diagnosis and treatment data processing in an existing method is solved, meanwhile, the accuracy rate in the center knowledge graph expansion process is improved, the characteristics of a special mechanism sub-graph are reserved, the updating delay of the center knowledge graph can be shortened, and the updating efficiency of the center knowledge graph is improved. Deployment time of new mechanism nodes is shortened, and medical conjunct expansion is adapted.

The application relates to the technical field of biomedical information processing, in particular to an autism classification system based on resting-state electroencephalogram signals, which comprises an EEG signal acquisition and preprocessing unit and an EEG signal classification unit, the EEG signal classification unit is used for classification by using a trained Rest-HGCN network model; the Rest-HGCN network model comprises a resting-state mixed graph network module, an attention learning module and a classification module; the resting-state mixed graph network module comprises a cognitive graph branch and a data-driven graph branch and is used for extracting corresponding feature mappings; the attention learning module is used for fusing the feature mappings extracted by the resting-state mixed graph network module to obtain a final feature mapping; and the classification module is used for classifying the final feature mapping to obtain a classification result. Through the classification system, the problems of ASD patient EEG feature extraction difficulty and low recognition rate in the prior art can be effectively solved, and only a small amount of features are needed to achieve the purpose of more efficient ASD classification recognition.

To correctly determine the mental condition of a subject.SOLUTION: A biological information analysis method acquires, for a plurality of subjects who have a specific mental tendency over a prescribed period, biological information including autonomic nerve data which includes sympathetic nerve data, parasympathetic nerve data, and total power data which is consolidation of the sympathetic nerve data and parasympathetic nerve data, behavior recording information in which multiple behaviors of the subjects are recorded, and inquiry answer information answered by the subjects, and performs machine learning using the biological information, the behavior recording information and the inquiry answer information, thus constructing a learned model.SELECTED DRAWING: Figure 1

The invention discloses a system, method and device for premature delivery risk prediction and a storage medium, and belongs to the technical field of biomedical information. The system comprises a data acquisition module and a data processing device. The data acquisition module is used for acquiring hemoglobin concentration, serum ferritin concentration, age, BMI, pregnancy times, delivery times, marriage conditions, education degree and other clinical characteristic data of the pregnant woman in the early pregnancy period, the middle pregnancy period and the late pregnancy period; the input characteristics of the model comprise all data of the three stages of pregnancy; the risk score is then compared with a specific threshold value set based on a Youden index, and a risk level of preterm delivery, spontaneous preterm delivery, or iatrogenic preterm delivery is output. The Hb-SF-Clinical multi-feature prediction model is constructed by fusing the three-stage pregnancy dynamic biomarkers and clinical features, the AUC (early birth prediction) of the Hb-SF-Clinical multi-feature prediction model in a test set reaches up to 96.0%, the Hb-SF-Clinical multi-feature prediction model is remarkably superior to the prior art, high-precision and subtype early birth risk assessment is realized, and the clinical applicability is high.

The scheme belongs to the field of artificial intelligence and biomedical information technology, and specifically relates to a biomedical knowledge extraction fusion sharing method and system based on federated graph data. The biomedical knowledge extraction fusion sharing method based on federated graph data comprises the following steps: S10: deploying a localized institution node in a participating medical institution, and establishing a communication network architecture between each institution node and a center node; the institution nodes jointly construct a center knowledge graph according to stored biomedical knowledge, and the center node stores the center knowledge graph; S20: the institution node obtains the diagnosis and treatment data of local patients. Through the communication network architecture and multi-modal processing, the scheme solves the problem of high privacy risk in the existing method when processing patient diagnosis and treatment data, improves the accuracy in the expansion process of the center knowledge graph, retains the characteristics of special institution sub-graphs, shortens the update delay of the center knowledge graph, shortens the deployment time of new institution nodes, and adapts to the expansion of medical alliances.

The application discloses a multi-modal graph fusion driven spatial transcriptome region identification method, and belongs to the field of biomedical informatics and computational biology. The method comprises the following steps: step 1, data acquisition; step 2, data processing; step 3, multi-modal adjacency matrix generation; step 4, differentiable multi-modal graph structure fusion; step 5, variational graph autoencoder training; and step 6, clustering. By constructing a spatial domain identification framework based on multi-modal graph fusion and variational autoencoder, the application significantly improves the accuracy, continuity and biological interpretability of spatial transcriptome clustering analysis, effectively solves the deficiencies of existing methods in multi-modal fusion, graph structure optimization and feature discriminability, and provides a more reliable computing tool for in-depth revealing of tissue spatial heterogeneity in the fields of brain science, tumor microenvironment analysis and other frontier biomedical researches.

The invention provides an autism diagnosis method based on multi-modal super-dimensional calculation and electronic equipment, and relates to the technical field of biomedical information. The method comprises the following steps: acquiring resting-state electroencephalogram data, eye movement tracking data and demographic data of a tested object; respectively carrying out typical feature extraction on the resting-state electroencephalogram data and the eye movement tracking data, respectively encoding electroencephalogram feature data and eye movement feature data into an electroencephalogram super-dimensional vector and an eye movement super-dimensional vector through super-dimensional calculation, and encoding the demographic data into a demographic super-dimensional vector; integrating the electroencephalogram super-dimensional vector and the eye movement super-dimensional vector through a weight fusion and residual enhancement mechanism to obtain a feature fusion super-dimensional vector, and integrating the feature fusion super-dimensional vector and a demographic super-dimensional vector through an adaptive fusion strategy to obtain a tested object super-dimensional vector of each tested object; and determining a diagnosis result of the target tested object according to a similarity calculation result with the plurality of class super-dimensional vectors. The autism diagnosis accuracy can be improved.

This invention relates to the field of biomedical information technology, and more particularly to a method for constructing an autism spectrum disorder (ASD) identification model based on deep learning. The method includes the following steps: acquiring videos of conversations with autistic patients; continuously monitoring the eye fixation frequency in the videos; when the eye fixation frequency is detected to be below a threshold, determining it as a reduced eye fixation state and recording the duration of eye fixation; performing sociolinguistic analysis on the conversation videos to obtain sociolinguistic data; and identifying facial reaction features of the autistic patients in the conversation videos based on the sociolinguistic data and recording facial muscle reaction data. This invention, through data processing and deep learning technologies, assesses communication and social interaction impairments in autistic patients and constructs an ASD identification model, thereby improving the efficiency and accuracy of ASD identification.

The invention relates to the technical field of biomedical information detection and processing, in particular to an infusion monitoring device and a monitoring prompt method thereof, comprising: a control box with a built-in main control unit; the movable clamp is used for clamping an infusion drip cup; the sensing unit is arranged in the movable clamp and is electrically connected with the main control unit; wherein the sensing unit detects liquid drops in the infusion drip cup and feeds back the liquid drops to the main control unit, and the main control unit calculates the number of the liquid drops in the infusion drip cup according to the number of feedback signals of the sensing unit so as to pre-judge that liquid medicine is about to be used up according to the number of the liquid drops and give an alarm to the outside. According to the method, the amount of infused liquid medicine can be estimated more accurately by conducting real-time quantitative accumulation on the liquid drops in the infusion process, a predictive alarm is given out on the basis of the preset time point or the residual capacity point before the liquid medicine is completely exhausted, sufficient response and treatment time is reserved for medical staff, and therefore the medical staff can take an alarm more accurately. And the safety of a clinical infusion process and the efficiency of nursing work are improved.

The invention discloses a brain function evaluation system based on a wireless dry electrode electroencephalograph, and belongs to the technical field of biomedical information. The system comprises an electroencephalogram signal acquisition and analysis sub-module, signals are acquired through a 21-channel wireless dry electrode electroencephalograph, and an electroencephalogram network, a brain topographic map and frequency spectrum energy are synchronously and dynamically visualized during acquisition. After data preprocessing, the analysis sub-module integrates brain network analysis, sample entropy analysis and power spectrum analysis, comprehensively evaluates brain functions from the three dimensions of function connection, signal complexity and frequency band energy distribution, and outputs visual results and quantitative indexes, the system achieves integration of collection and analysis, operation is convenient, and evaluation is comprehensive.

The invention discloses a biomedical information verification model establishment method based on artificial intelligence, and relates to the technical field of biomedical information verification, and the method comprises the following steps: constructing a coherent cell culture process, obtaining cell culture information, analyzing kidney tissues of a selected animal model, and obtaining animal kidney information. The renal toxicity judgment index is generated from the cell culture information and the animal kidney information through a nuclear standardization method, and is compared with the set renal toxicity evaluation threshold to verify whether the biomedical supplies have the renal toxicity, so that the renal toxicity of the biomedical supplies can be quickly and accurately evaluated, and the clinical application prospect is wide. The development period and the development cost can be effectively controlled for the pharmaceutical industry, and the reliability is improved.

The invention belongs to the technical field of biomedical information, and relates to a multi-behavior quantification method and system for autism screening and application. According to the system, normal form presentation is adopted to induce facial expressions and upper limb actions of a subject, facial micro-expressions and limb dynamic change characteristics of the subject are captured through the data acquisition module, and the facial expressions and limb behavior characteristics of the subject are automatically acquired and analyzed through the data analysis module; according to the invention, the method achieves the quantification of facial expressions and limb movement features, outputs autism risk scores and classification judgment results of children through an output module, provides visual and comprehensive child behavior performance and autism risk information for clinicians, and improves the objectivity and ecological effectiveness of screening.

The application relates to the technical field of bioinformatics, in particular to a drug repositioning method based on reinforced symmetric metric learning and a graph convolution network, which comprises the following steps: a drug-disease heterogeneous network is constructed by integrating the correlation of drugs and diseases and biomedical information. The heterogeneous network comprises a drug-drug similarity network, a disease-disease similarity network and a drug-disease correlation network. A graph convolution network is applied to learn the node features of drugs and diseases, and potential drug-disease correlations are predicted to supplement the missing drug-disease correlation information. A reinforced symmetric metric learning method with adaptive margins is used to learn the potential vector representation of drugs and diseases, and the symmetric learning of drug-centered and disease-centered is considered. Based on the potential vector representation learned in the unified metric vector space, new drug-disease correlations are identified through a metric function. The application is simple and effective, and has good performance in drug repositioning prediction.

The invention relates to the related field of biomedical information management, discloses a stem cell quality evaluation system and method based on multi-source data fusion analysis, constructs an intelligent integrated data analysis platform, gets through information synchronization among stages of stem cell products, realizes unified data interaction management, and improves the quality of stem cells. By integrating stem cell associated data of each stage and performing consistent visual supervision processing, convenient query and comprehensive judgment of isolated information are realized, and internal quality control and risk use management of stem cell quality are facilitated.

This application relates to the field of natural language processing technology, and in particular to a biomedical information extraction method based on a large language model. The method includes: acquiring the biomedical dataset to be processed and performing standardized preprocessing; converting the relation extraction data into high-dimensional vectors and constructing a local vector library; acquiring the medical information text to be processed as the query text, and performing a two-stage example retrieval and filtering in the local vector library to obtain a high-quality example set; generating context examples and performing context learning to understand the current task requirements and generate the information extraction results of the query text; and parsing the information extraction results. Based on the reordering capability of the cross-encoder model, this application designs a two-stage retrieval and reordering mechanism. By ensuring that the ICL examples provided to the large language model have both high semantic relevance and high task guidance, it significantly improves the accuracy and robustness of the model in biomedical named entity recognition and relation extraction tasks.

The present application belongs to the cross technical field of biomedical information processing and artificial intelligence, and specifically relates to a medical prediction model literature quality automatic evaluation method based on TRIPOD+AI standard, obtaining a medical literature file to be evaluated, performing multi-modal analysis and layout analysis on the file, segmenting the document into multiple semantic blocks, and extracting structured data and key text blocks therefrom; constructing a scoring evaluation matrix based on the TRIPOD+AI guideline, the matrix containing statistical hard indicators and descriptive soft indicators; and scoring the structured data and key text blocks in a parallel dual-track mode using a rule calculation engine and a large language model semantic reasoning engine. The present application significantly improves the scoring efficiency, significantly improves the fine granularity and discrimination of the evaluation, and solves the trust problem of "black box" evaluation.

The invention discloses a spatial transcriptome region identification method driven by multi-modal graph fusion, and belongs to the field of biomedical informatics and computational biology. The method comprises the following steps: step 1, data acquisition; 2, data processing; step 3, generating a multi-modal adjacent matrix; 4, carrying out the structure fusion of the differentiable multi-modal graph; step 5, variational graph self-encoder training is carried out; and step 6, clustering. According to the method, the spatial domain recognition framework based on multi-modal graph fusion and the variational autoencoder is constructed, so that the accuracy, continuity and biological interpretability of spatial transcriptome clustering analysis are remarkably improved, and the defects of an existing method in the aspects of multi-modal fusion, graph structure optimization and feature discrimination are effectively overcome; and a more reliable calculation tool is provided for deeply revealing tissue spatial heterogeneity in the field of advanced biomedical research such as brain science, tumor microenvironment analysis and the like.

The invention discloses a standard compound library MetaTag database and application thereof.43120 compounds are recorded in the MetaTag database, the compounds have Xm-Yn type chemical structural characteristics, X represents an electrophilic substituent, Y represents a nucleophilic substituent, m and n represent the number of X and the number of Y respectively, the character '-' represents that X and Y form a covalent bond, and X and Y are both from endogenous metabolites; each compound is composed of a virtual part and an entity part; in the virtual part, each compound corresponds to a unique entry, and each entry comprises but is not limited to sub-entries such as a chemical information class, a biomedical information class, a related literature or an external database link; in the entity part, each compound can find a molecular entity in a mixed standard series of a MetaTag database, and the molecular entity is used for collecting LC-MS data to confirm the metabolite structure.

This application relates to the field of medical information extraction technology, and particularly to a biomedical information extraction method based on two-stage fine-tuning and preference optimization. The method includes: optimizing a base model using a two-stage fine-tuning and preference optimization strategy, and then using the optimized model to extract biomedical information. This application designs a progressive preference learning framework, employing an improved DPO-Positive algorithm and post-fine-tuning to enhance model accuracy for medical information extraction tasks; it also designs a multi-dimensional preference dataset and its automated generation method to reduce the burden of manual annotation and improve model fault tolerance and generalization ability; and combines efficient parameter fine-tuning and dual-model correction verification to achieve high-performance, standardized output with limited computing power, improving knowledge extraction efficiency and reliability.

The invention relates to the technical field of biomedical information, in particular to a pancreatitis prognosis evaluation method and system fusing clinical and image data, and the method comprises the steps: obtaining and preprocessing the clinical and image data; carrying out clinical and image feature extraction and multi-modal deep fusion; and the prognosis prediction module outputs an evaluation result. The system comprises a data acquisition module, a preprocessing module, a feature extraction module, a multi-modal fusion module and a prognosis prediction module. According to the technical scheme, the accuracy and objectivity of prognosis evaluation can be remarkably improved, early-stage high-risk recognition is assisted, treatment is optimized, and precise medical treatment is achieved.

The invention discloses a dynamic graph transform electroencephalogram emotion recognition method fusing multi-scale spatial-temporal characteristics, and belongs to the technical field of biomedical information. The method comprises the following steps: acquiring a multi-channel emotion electroencephalogram signal, and preprocessing the multi-channel emotion electroencephalogram signal; dividing the preprocessed electroencephalogram signal into a plurality of continuous dynamic electroencephalogram signal segments according to a sliding time window and a window overlapping range, distributing an original emotion tag for each dynamic electroencephalogram signal segment, and extracting frequency domain features of the electroencephalogram signal segments; constructing a multi-layer dynamic emotion electroencephalogram network based on the preprocessed multi-channel electroencephalogram signal data and the extracted electroencephalogram signal frequency domain features; a multi-scale space-time diagram Transform network is constructed, and multi-scale space structure information and complex time dynamic characteristics in the multi-layer dynamic emotion electroencephalogram network are deeply fused; and training and optimizing the multi-scale space-time diagram Transform network to obtain an emotion recognition model, and performing emotion recognition on the electroencephalogram data by using the emotion recognition model.

The invention belongs to the technical field of biomedical information technology, cancer genomics and clinical diagnosis, and discloses a method and a device for automatically rating tumor somatic mutation carcinogenicity based on multi-dimensional evidence. According to the method, variation data to be rated are acquired and subjected to standardized annotation, crowd frequency, cancer hot spots, functional experiments and prediction information are automatically integrated based on a preset rating rule set, multi-dimensional evidence is calculated and predicted, and mutation is rated as carcinogenic, possibly carcinogenic, indefinite in meaning, possibly benign or benign after quantitative scoring. The device is an interactive system and supports variation retrieval, real-time rating, evidence review and result adjustment. According to the method, the problems of strong subjectivity, poor consistency, low efficiency and insufficient standardization of manual rating in the prior art are solved, efficient, accurate and repeatable rating of whole genome scale variation is realized, and reliable support is provided for genome diagnosis and targeted therapy.

The multi-task large language model training method for the biomedical field belongs to the field of large language model training. In order to solve the problem of single large language model realizing multi-task processing, the technical key points are as follows: constructing a training set of a first instruction data set for a medical intelligent question answering task; constructing a training set of a second instruction data set for a medical report generation task; constructing a training set of a third instruction data set for a biomedical information extraction task; training the large language model using the training sets, and the effect is that the single large language model can realize multi-task accurate and efficient processing.

The invention discloses a heart rate estimation system based on adaptive filtering and a joint sparse spectrum reconstruction model, and belongs to the technical field of biomedical information processing. The method comprises the following steps: firstly, filtering an acquired PPG signal and a synchronous acceleration signal, then judging whether the signals are in an initialization stage, if so, calculating the heart rate at the current moment by utilizing a joint sparse spectrum reconstruction model, if not, judging the signal quality, and when the signal quality meets the requirement, judging the signal quality; if yes, heart rate estimation is conducted through a normalized least-mean-square adaptive filtering algorithm, an estimation result is verified, if verification is passed, the current heart rate is output, and when verification fails and the signal quality does not meet the requirement, the heart rate value at the current moment is obtained through recalculation through a joint sparse spectrum reconstruction model. According to the method, the calculation time is greatly shortened on the premise that the precision equivalent to that of a joint sparse spectrum reconstruction model is met, the requirements of robustness and real-time performance are met at the same time, and the technical problem that high precision and low calculation complexity conflict is solved.