22 results about "Main diagnosis" patented technology

The present application relates to the technical field of natural language processing, and discloses a large language model clinical diagnosis capability fine-grained evaluation method and system. The method comprises: obtaining a patient medical record data set; synchronously performing the following tasks: scoring a question-standard answer-model answer triple based on a predetermined scoring dimension, calculating the medical knowledge score of the evaluated model; the evaluated model performs a recall task and a differential task, confirms or excludes a differential diagnosis list, and obtains a final main diagnosis; linking the differential diagnosis list and the final main diagnosis output by the evaluated model in the patient medical record; and analyzing factors affecting the clinical diagnosis capability of the evaluated model based on the linking result. The present application aims to deeply and synchronously evaluate the medical knowledge and reasoning capability of a large language model based on the same clinical diagnosis situation.

The invention discloses an APG settlement method and system based on outpatient disease category grouping, and aims to solve the problems of difficulty in automatic identification of outpatient main diagnosis and main operation and insufficient consistency with policy versions. The method comprises the following steps: constructing a heterogeneous evidence graph comprising a doctor-seeing inner layer, an individual longitudinal layer and an APG and policy layer, adopting a graph Transform constrained by relation perception and a monotonous time sequence attention mask, setting diagnosis and operation interactive decoding, introducing a probability soft logic rule bound with a policy version, and carrying out joint differentiable first K item selection, and main items are determined, and main group and additional group mapping and settlement calculation are completed, so that the technical effects of accurate main item identification, consistent time sequence, policy compliance and explainable end-to-end settlement are realized.

The invention discloses a circular non-coding RNA (Ribonucleic Acid) molecule circSENP6 and application thereof. The nucleotide sequence of the circSENP6 is shown as SEQ ID NO. 1. The invention also discloses a preparation method of the circular non-coding RNA molecule circSENP6. The circular non-coding RNA molecule circSENP6 can be used as a diagnostic biomarker for coronary atherosclerosis, is used for biological identification and clinical detection of coronary atherosclerotic heart disease, and can be used for early discovering coronary heart disease patients, reducing the death rate of coronary heart disease and improving the prognosis of patients by using the expression level of the gene for diagnosis. At present, the main diagnosis method for coronary atherosclerotic heart disease has the risks of high operation danger coefficient, large radiation and the like, and by collecting peripheral blood of a patient for detection, the method has the advantages of small trauma, good repeatability and the like, and is easier to accept.

The invention discloses application of FGL1 protein in preparation of a kit for diagnosing vertigo. Clinically, diagnosis of vertigo depends on diagnosis experience of doctors, complaint of patients becomes a main diagnosis basis, the doctors perform preliminary diagnosis according to physical examination, medical history inquiry and the like and then perform clear diagnosis according to etiological examination, diagnosis time is long, and preliminary diagnosis subjectivity is high. It is found that the FGL1 protein can effectively diagnose and distinguish healthy subjects and patients with vertigo, the diagnosis accuracy is high, the FGL1 protein can be used for being developed into a kit for diagnosing vertigo, and compared with an existing diagnosis method, the diagnosis efficiency and objectivity are obviously improved.

The application provides a method and device for training a model, and a method and device for predicting a sub-diagnosis intention. An electronic device supports obtaining a target main diagnosis intention of a patient in a process of queuing for diagnosis, and predicting and outputting, based on a prediction model of a sub-diagnosis intention, a sub-diagnosis intention in which a patient inputs a sub-named entity from a plurality of sub-diagnosis intentions cascaded with the target main diagnosis intention, so that the patient can input a sub-named entity to the output sub-diagnosis intention, to collect answers of the patient to questions in advance automatically by using time of the patient in the process of queuing for diagnosis, and facilitate automatic recording of a disease of the patient, so as to avoid omission and misdiagnosis or missed diagnosis. Furthermore, after the patient enters a diagnosis stage, a doctor can directly obtain answers of the patient to questions, without real-time questioning of the patient by the doctor, so as to save time of the doctor and the patient in the process of diagnosis.

The invention provides an elevator brake fault diagnosis method based on multi-scale feature distillation, and relates to the field of fault diagnosis of elevator brakes. The method comprises the steps that multi-channel signals related to a braking event of an elevator brake are collected, the multi-channel signals comprise PVDF strain, acceleration time domain and high-frequency acoustic emission transient pulse multi-channel signals, and standardized signals are obtained through time alignment, signal slicing, windowing and normalization processing; inputting the data into a main diagnosis network, extracting and fusing multi-scale features, completing fault type and severity double-task diagnosis, training the network in combination with a real label, and generating a soft label; a lightweight edge diagnosis network is constructed, and a deployable diagnosis model is obtained based on real label and soft label distillation training; and collecting real-time signals, processing the signals and inputting the signals into the diagnosis model to obtain fault types and severity. According to the invention, the fault identification and severity evaluation precision can be improved, and the real-time online diagnosis of the elevator brake can be realized.

The application provides a DRG grouping recommendation method, device, equipment and medium based on a large model, wherein the method comprises the following steps: based on a medical model, extracting a disease from a medical record of a patient to be recommended to obtain a diagnosed disease, and selecting a main diagnosis based on the diagnosed disease, a diagnosis disease list and the medical record to obtain a main diagnosis disease; and performing DRG grouping recommendation based on the main diagnosis disease to obtain a DRG grouping recommendation result of the patient to be recommended, which overcomes the defects of poor clinical adaptability, easy omission of important information, and insufficient accuracy and reliability of the current DRG grouping, starts from the perspective of diagnosis and treatment, excludes diagnoses based on insufficient medical models, thereby optimizing and improving the diagnosis disease list, and on this basis, the main diagnosis is selected and the DRG grouping is recommended according to the medical record, which can realize accurate and reliable DRG grouping, has strong explainability and adaptability in the clinical use process, and can meet the actual application requirements.

The invention discloses a main diagnosis quality control method based on an inference thinking mode, which comprises the following steps: firstly, collecting massive original electronic medical record data, and then combining the collected medical record data with different LLM models to construct a fine tuning data set; then selecting an LLM base model, and performing SFT fine tuning on the LLM base model by adopting a LoRA technology to obtain an LLM fine tuning base model; continuously carrying out GRPO reinforcement learning on the LLM fine tuning base model by adopting a LoRA technology to obtain a quality control model after reinforcement learning; and finally inputting to-be-analyzed electronic medical record data into the quality control model, and outputting a corresponding quality control analysis result by the quality control model. The method has the remarkable effects that various demands of high efficiency, high precision, high interpretability, high coverage rate, low maintenance, clinical easiness in use and the like are simultaneously met in the same technical framework, and the technical blank of the main diagnosis quality control field is filled; and a landing, sustainable and extensible normal form-level solution is provided for medical insurance settlement, performance assessment and medical data management.

The invention provides a medical record homepage main diagnosis recommendation method and system based on rule-model collaboration, and belongs to the technical field of medical information. The method comprises the following steps: extracting, analyzing and standardizing structured and unstructured data in a home page of a medical record to generate structured data for feature calculation; based on the structured data, constructing a multi-dimensional feature vector including clinical priority, ICD coding attributes, resource consumption and patient basic information; enhancing the multi-dimensional feature vector through a rule-model cooperation mechanism to form an enhanced feature vector; and inputting the enhanced feature vector into a trained machine learning sorting model to obtain a recommendation probability of each candidate main diagnosis, and outputting a main diagnosis recommendation list based on the recommendation probability. According to the invention, the accuracy and efficiency of main diagnosis selection are effectively improved, and the risk of medical insurance payment refusing and the workload of doctors are reduced.

The invention provides a vehicle fault prediction method based on AUTOSAR, and the method comprises the steps: A, carrying out enhanced data collection and diagnosis execution, recording key parameter snapshots and supporting a periodic broadcast mode through expanding a diagnosis event manager DEM and a diagnosis communication manager DCM of a basic software layer of the AUTOSAR; b, light-weight in-vehicle data aggregation, wherein prediction required data signals distributed on different electronic control units (ECU) are aggregated to an application layer of a main diagnosis ECU through an environment RTE during operation of the AUTOSAR; step C, deploying and executing loosely coupled intelligent prediction, packaging a prediction algorithm into an independent AUTOSAR software component SWC, and subscribing real-time data flow and accessing historical fault data by the SWC through an RTE port; and D, performing fault prediction by adopting a hybrid prediction model in combination with the physical model and the data-driven model. The safety and reliability of the vehicle are effectively improved, and the full-life-cycle maintenance cost is reduced.

The invention relates to the technical field of magneto-rheological seat diagnosis, and discloses a fault diagnosis method based on a magneto-rheological seat. The method comprises the following steps: acquiring current state data of the magnetorheological seat, and establishing a fault feature database mapped with the current state data; configuring diagnosis targets according to the fault feature database, wherein the diagnosis targets comprise vibration abnormity diagnosis, pressure distribution diagnosis and response time diagnosis; then determining a target classification identifier in the diagnosis target, wherein the target classification identifier comprises a main diagnosis target, an auxiliary diagnosis target and a reference target; after key diagnosis parameters are selected through the main diagnosis target and the auxiliary diagnosis target, diagnosis target influence analysis of the key diagnosis parameters is executed; and finally, after parameter optimization constraints are established according to the diagnosis target influence analysis result, parameter directional optimization is executed, and magneto-rheological seat fault diagnosis is completed by using a parameter directional optimization result. According to the method, the diagnosis efficiency and accuracy can be improved, and stable operation of the magneto-rheological seat is guaranteed.

The invention provides an electronic medical record disease classification coding method and device, and the method comprises the steps: carrying out the preprocessing of electronic medical record data, and obtaining diagnosis and treatment information; inputting the diagnosis and treatment information and an external standard data set into the trained first large model, and outputting at least one piece of first diagnosis information; inputting the at least one piece of first diagnosis information and an external standard data set into a plurality of trained second large models, and respectively obtaining at least one piece of second diagnosis information output by each second large model; and judging whether the proportion of the same second diagnosis name in the first sequence and the second ICD-11 code combination in the plurality of second large models is greater than or equal to a specified threshold value or not, and if so, combining the same second diagnosis name in the first sequence and the second ICD-11 code combination as the main diagnosis. The invention provides an electronic medical record disease classification coding method and device, which are used for solving the problems of low coding accuracy and incomplete disease coverage in the prior art.

The utility model relates to the technical field of medical instruments, in particular to a disposable hard bronchoscope which comprises an operation end and a bronchoscope tube, one end of the bronchoscope tube is an inclined opening, the operation end is detachably connected with the bronchoscope tube, the bronchoscope tube is a hard tube, the operation end is provided with a main examination port, a jet vent, an auxiliary vent and a blocking portion, the main examination port is communicated with the bronchoscope tube, and the jet vent is communicated with the auxiliary vent. The spraying ventilation opening, the blocking part and the auxiliary ventilation opening are formed in the two sides of the main examination opening, the spraying ventilation opening is communicated with the main examination opening, an air bag assembly used for expanding the trachea is arranged on the endoscope tube and comprises an annular air bag and an air inlet tube, the other end of the air inlet tube is connected with a connecting assembly, and the connecting assembly comprises a three-way connector. The disinfection device disclosed by the utility model has the beneficial effects that the disinfection cost is effectively reduced, and cross infection is prevented; the endoscope tube is communicated with the main examination port, one light source port is omitted, the number of connectors is small, the inner wall is smooth, and airflow is stable.

The invention relates to the technical field of artificial intelligence, and discloses an intelligent triage system and method based on artificial intelligence questions and answers, and the system comprises an information collection module, a triage question module, a multi-department triage module and a triage response module which are in communication connection in sequence. The method corresponds to the system. According to the application, the patient information is accurately acquired through the information acquisition module, and the symptom information is refined through interdisciplinary dynamic questions and answers of the triage question module to generate the optimized disease information, so that the accuracy of association between the symptom and the department is improved; the multi-department triage module quantifies the distribution probability and support degree of each department to realize advanced pre-judgment of interdisciplinary collaboration; the triage response module is combined with real-time response of departments to dynamically determine main diagnosis and support departments, referral caused by misjudgment of a single department is reduced, triage efficiency is improved, the problems that interdisciplinary disease triage is difficult and cooperation is lagged are solved, medical resource configuration is optimized, and patient treatment experience and medical service quality are improved.

The invention relates to the technical field of natural language processing, and discloses a large language model clinical diagnosis capability fine-grained evaluation method and system. The method comprises the following steps: acquiring a patient medical record data set; synchronously executing the following tasks: scoring the question-standard answer-model answer triple based on a predetermined scoring dimension, and calculating a medical knowledge score of the evaluated model; executing a recall task and an identification task by the evaluated model, and confirming or excluding the identification diagnosis list to obtain final main diagnosis; linking a differential diagnosis list in the medical record of the patient and output by the evaluated model with the final main diagnosis; and analyzing factors influencing the clinical diagnosis capability of the evaluated model based on a link result. The invention aims to deeply and synchronously evaluate the medical knowledge and reasoning ability of the large language model based on the same clinical diagnosis situation.

The application belongs to the technical field of intelligent management of ancient buildings, and provides an ancient building wood component diagnosis and prevention intelligent management method combined with digital twinning, comprising: processing the main diagnosis surface image of the wood component by using a damage diagnosis model to obtain damage state recognition results; integrating the static attribute data of the wood component, and environment exposure data, damage state recognition results and maintenance data to obtain a structured digital twinning archive; obtaining the feature vector of the wood component at each historical time point based on the structured digital twinning archive, and stacking them in time sequence into a three-dimensional feature tensor; inputting the three-dimensional feature tensor into a multi-task time series prediction model to obtain the future damage prediction result of the wood component and the damage influence weight of the historical time point; calculating the SHAP value of the index; generating damage cause description; matching the damage cause description with a decision rule library to obtain the maintenance decision instruction of the wood component and sending it. The accuracy and forward-looking nature of the future damage prediction result of the wood component are improved, and a decision closed loop is formed.

The application discloses a DRG high set identification method, medium and equipment based on bill backtracking and causal imprint, and the method comprises the following steps: acquiring medical charging detail data of a to-be-audited medical record, extracting diagnosis and treatment project features and mapping to a medical knowledge graph to obtain a feature atom set; inputting the feature atom set into a large language model to output a real disease phenotype as a pseudo label, and performing semantic comparison with a reported main diagnosis to identify main diagnosis high set; performing score sensitivity grading on other diagnoses reported, screening high-value complication entities, verifying resource consumption matching degrees of the high-value complication entities through a clinical path constraint library, and identifying fictitious complications; extracting main charging items and a low-value charging imprint rule library, identifying virtual increase of main trunk intervention projects through counterfactual causal comparison; and finally, calculating a comprehensive abnormality index by comprehensively considering deviation, matching degree and deviation, and outputting an audit report. The application automatically identifies DRG high set abnormal behaviors from multiple dimensions, and improves the accuracy and comprehensiveness of the audit.

The invention discloses a meta-learning bearing fault diagnosis method and system based on an adaptive task generator. Defining a performance index sequence, and training a self-adaptive task generator through the historical performance index sequence, so that the self-adaptive task generator has the capability of dynamically generating a task sampling probability; constructing a collaborative meta-learning optimization framework of the main diagnosis model and the self-adaptive task generator, and realizing collaborative evolution of the main diagnosis model and the self-adaptive task generator through alternate iteration of internal loop rapid adaptation and external loop generalization evaluation; the task difficulty is dynamically adjusted by quantifying a difficulty index and an annealing mechanism, so that the task difficulty is gradually increased; and performing multi-round iterative optimization until model performance convergence, and outputting a final main diagnosis model and a self-adaptive task generator model. Through historical performance driven task generation, collaborative optimization of the main model and the generator and dynamic difficulty adjustment, the diagnosis precision and generalization ability of the model in a small sample scene are improved.

The application discloses a disease diagnosis automatic coding method and system, and belongs to the technical field of medical event recognition. The method comprises the following steps: acquiring and structuring preprocessed disease diagnosis documents, displaying the document directory and content; using a diagnosis coding model to extract diagnosis names, evidences and surgical operations from the documents, and displaying the diagnosis names, evidences and surgical operations in a multi-window in a highlighting mode; supporting auditing, commenting, editing, merging and traceability positioning of the diagnosis and evidence information; automatically generating a diagnosis coding result, sorting the diagnosis coding result, modifying the main diagnosis and adjusting the sequence of the associated evidences; the system supports DRG / DIP simulation grouping display, and assists in coding sequence decision-making; the coding result can be saved, collected and uploaded to a related system, and a complete coding closed-loop process is constructed. The application can greatly reduce the work burden of coders, improve medical insurance settlement efficiency and data quality, and has good clinical application value and popularization prospect.

The present application relates to a kind of settlement based on grouping recommendation system and method of list, belong to DRG grouping technical field.The present application discloses grouping recommendation system includes in situ fuzzy grouping module, ectopic fuzzy grouping module, QY (ambiguity) grouping module, missing code recommendation module, grouping fit degree module, and grouping recommendation method steps are as follows: in situ fuzzy grouping is carried out, and grouping range is expanded;Ectopic fuzzy grouping is carried out, and grouping range is further expanded;QY (ambiguity) grouping supplement is carried out, and grouping is carried out with the main diagnosis code of surgical coding collocation, obtains the possible existence of real grouping;Missing code grouping supplement is carried out, and in situ fuzzy grouping is carried out, obtains the possible existence of real grouping;The fit degree of S1-S4 all groupings is calculated respectively, and grouping fit degree value is arranged in descending order form and is arranged in list.The present application can recommend the most suitable grouping to doctor by analyzing the position of ICD coding, classification relationship.

The invention relates to a medical insurance settlement method based on a DRG model, and the method comprises the steps: firstly determining a basic weight according to a basic DRG group of a main diagnosis disease in a target case; then, aiming at the comprehensive disease feature vectors corresponding to the main diagnosis disease and the whole complications, applying a concurrent synergistic effect model to obtain corresponding concurrent superposition coefficients; according to the regional basic rate under the DRG model, the medical insurance payment amount corresponding to the target case is calculated and obtained; according to the design scheme, by introducing a concurrent superposition coefficient, precise quantification is achieved for cost increase caused by the nonlinear synergistic effect among multiple complications, the fairness and precision of DRG payment are remarkably improved, the reasonable income of medical institutions for treating complex cases can be guaranteed, the method can adapt to medical technology progress and cost change, and the method is suitable for popularization and application. And core technical support is provided for scientific and refined treatment of a medical insurance payment system.

The invention relates to the technical field of deep learning, and discloses an equipment fault prediction and diagnosis method in an industrial control interconnection environment based on deep learning, and the method comprises the following steps: a, collecting an original sensor signal and an original working condition parameter in real time from a sensor disposed in industrial equipment, processing the original sensor signals and the original working condition parameters to generate processed sensor features and processed working condition features; and b, inputting the processed working condition characteristics into a pre-trained working condition gating sub-network, wherein the working condition gating sub-network dynamically generates gating signals based on the working condition characteristics. According to the method, the technical scheme that the batch normalization parameters of the main diagnosis network are dynamically modulated through the working condition gating sub-network is adopted, the scaling factor and offset of the features are adjusted in real time, the negative migration effect caused by working condition changes is remarkably eliminated, and therefore the diagnosis precision and stability of the model under the variable working conditions are improved.