Search

CN-122025084-A - Health data processing system, method and computer readable storage medium

CN122025084ACN 122025084 ACN122025084 ACN 122025084ACN-122025084-A

Abstract

The application relates to the technical field of artificial intelligence, and discloses a health data processing system, method and medium based on multi-source verification, which can be applied to intelligent interaction scenes in financial and medical scenes. The system part comprises an acquisition module, a data confidence verification module, an analysis module and a traceability feedback output module, wherein the acquisition module is configured with a heterogeneous sensor array aiming at the same target physiological parameter and acquires multiple paths of original physiological signals of a user, the data confidence verification module is used for executing consistency verification on the multiple paths of original physiological signals and determining effective physiological parameter data, the analysis module is used for extracting time sequence characteristics from the effective physiological parameter data, matching the dynamic medical knowledge graph with corresponding clinical phenotype nodes, inputting the clinical phenotype nodes and covariate information of the user into an inference model for calculation to obtain posterior risk probability values of target health abnormal conditions, and the traceability feedback output module is used for generating a health monitoring report and outputting medical basis chain information corresponding to the posterior risk probability values. The application aims to solve the problem that the health monitoring result in the prior art lacks authority and interpretability.

Inventors

  • WANG JIANZONG
  • LI JIALIN

Assignees

  • 平安科技（深圳）有限公司

Dates

Publication Date
20260512
Application Date
20260115

Claims (10)

  1. 1. A health data processing system based on multi-source verification, comprising: the heterogeneous acquisition module is configured with heterogeneous sensor arrays aiming at the same target physiological parameter, and the heterogeneous sensor arrays comprise at least three sensors based on different measurement principles and are used for acquiring multiple paths of original physiological signals of a user; The data confidence verification module is used for receiving the multiple paths of original physiological signals, executing dynamic consistency verification on the multiple paths of original physiological signals, and determining effective physiological parameter data according to a verification result; The knowledge fusion analysis module is used for storing a pre-constructed dynamic medical knowledge graph, extracting time sequence features from the effective physiological parameter data, matching corresponding clinical phenotype nodes in the dynamic medical knowledge graph based on the time sequence features, inputting the matched clinical phenotype nodes and covariate information of a user into a medical knowledge fusion reasoning model for calculation, and obtaining a posterior risk probability value of a target health abnormal condition; And the traceability feedback output module is used for generating a health monitoring report containing the posterior risk probability value, retrieving medical basis chain information associated with the clinical phenotype node from the dynamic medical knowledge graph and outputting the medical basis chain information and the health monitoring report in an associated mode.
  2. 2. The system of claim 1, wherein the data confidence verification module, when performing a dynamic consistency check, is configured to: Calculating the deviation value between any two paths of signal output results in the multipath original physiological signals in real time, and judging whether the deviation value is smaller than a preset consistency threshold value or not; If the deviation value of the output results of at least two independent sensors is recognized to be smaller than the consistency threshold value, corresponding data are recognized as the effective physiological parameter data; If the deviation values of the output results of all the sensors are recognized to be larger than the consistency threshold value, a confidence degradation prompt signal is generated, and prompt information is output to guide a user to adjust the contact state of the heterogeneous sensor array and the user or keep the static state in response to the confidence degradation prompt signal.
  3. 3. The system of claim 1, wherein the system further comprises a controller configured to control the controller, the system further includes an adaptive calibration module configured to: monitoring real-time physiological state data of a user, and identifying a time period in a stable physiological period as a reference calibration window; acquiring sensor measurement data within the reference calibration window to establish an individualized baseline model; And calculating the parameter offset of the heterogeneous sensor array based on the personalized baseline model, and correcting the original physiological signals acquired later by utilizing the parameter offset.
  4. 4. The system of claim 1, wherein the pre-constructed dynamic medical knowledge graph in the knowledge fusion analysis module is constructed and updated by: Acquiring a medical guideline, a disease classification standard and a clinical decision support library as a structured knowledge source, carrying out semantic analysis on the structured knowledge source to extract medical entities and association relations among the entities, and constructing dynamic medical knowledge maps covering different human body systems based on the medical entities and the association relations; And monitoring the information updating state of an international medical organization release channel, automatically capturing new edition of guide content when the release of a new edition of medical guide is monitored, and synchronously mapping the new edition of guide content to a corresponding node of the dynamic medical knowledge graph so as to update the association relation or the medical entity.
  5. 5. The system of claim 1, wherein the knowledge fusion analysis module, when inputting the matched covariate information of the clinical phenotype node and user to a medical knowledge fusion inference model for computation, is configured to: The matched clinical phenotype nodes are used as input features of a graph neural network model in the medical knowledge fusion reasoning model; in the graph neural network model, feature propagation and aggregation are carried out by combining the covariate information so as to output intermediate feature representation; performing Bayesian inference calculation on the intermediate feature representation to obtain the prior probability of the abnormal condition of the target health; and combining the fluctuation amplitude of the time sequence characteristic with the prior probability to calculate the final posterior risk probability value.
  6. 6. The system of claim 1, wherein the trace-source feedback output module, when outputting medical evidence chain information, is configured to: Tracing back source nodes in the dynamic medical knowledge graph according to graph paths activated by calculating the posterior risk probability values; Extracting guide clauses, study document numbers and recommendation level information stored in the source node; and combining the extracted guideline clause, study document number and recommendation level information into the medical basis chain information.
  7. 7. The system of any of claims 1-6, further comprising a hierarchical response interaction module configured to: Performing interval judgment on the posterior risk probability value; When the posterior risk probability value is in a first risk interval, automatically encrypting and transmitting the time sequence characteristic and the effective physiological parameter data to a preset remote medical service end, and receiving feedback data returned by the remote medical service end; When the posterior risk probability value is in a second risk interval higher than the first risk interval and a specific critical index is monitored, triggering emergency intervention logic, and sending a help seeking signal and positioning information to a preset external terminal.
  8. 8. The system of claim 7, wherein the hierarchical response interaction module is further configured to: Converting the feedback data into a standard document format and storing the feedback data into a digital archive storage area of a user; and responding to the evaluation instruction of the user, collecting the effectiveness score of the user on the feedback data or the health monitoring report, and transmitting the effectiveness score to the knowledge fusion analysis module for optimizing the parameters of the medical knowledge fusion inference model.
  9. 9. A health data processing method based on a system according to any of claims 1-8, comprising: controlling the heterogeneous sensor array to acquire multiple paths of original physiological signals aiming at the same target physiological parameter; Performing dynamic consistency test on the multiple paths of original physiological signals, and determining effective physiological parameter data according to test results; extracting time sequence features from the effective physiological parameter data, and matching corresponding clinical phenotype nodes in a pre-constructed dynamic medical knowledge graph based on the time sequence features; Inputting the matched clinical phenotype nodes and the covariate information of the user into a medical knowledge fusion reasoning model for calculation to obtain posterior risk probability values of the target health abnormal conditions; and generating a health monitoring report containing the posterior risk probability value, and outputting medical evidence chain information corresponding to the posterior risk probability value.
  10. 10. A computer readable storage medium, on which a computer program is stored which, when being executed by a processor, carries out the steps of the method according to claim 9.

Description

Health data processing system, method and computer readable storage medium Technical Field The application relates to the technical field of medical science and technology and artificial intelligence, in particular to a health data processing system and method based on multi-source verification and a computer readable storage medium. Background In the technical field of medical science and technology, particularly in the application field of digital medical technology, along with the development of Artificial Intelligence (AI) and internet of things technologies, intelligent health detection devices have been widely applied to personal health management. However, existing health detection techniques face serious challenges in terms of reliability and authority of results. Specifically, most of current AI health analysis models adopt an end-to-end black box decision mechanism, only one number or simple risk prompt is output, and a user cannot know the deduction basis and logic process of the result. In addition, existing health advice is often trained based on general rules or small sample data, lacks direct support for authoritative medical guidelines and clinical evidence-based medical evidence, and results in serious generalization of the output advice and even false scientific misleading. The lack of transparency and the processing mode of medical endorsement make it difficult for a user to judge whether the detection result is scientific and reliable, and severely restrict the application value and the user trust degree of the intelligent equipment in disease early warning and auxiliary diagnosis and treatment scenes. Disclosure of Invention The embodiment of the application provides a health data processing system, a health data processing method and a computer readable storage medium based on multi-source verification, aiming at solving the problems of opaque health monitoring analysis logic and lack of credibility of results in the prior art. The embodiment of the application provides a health data processing system based on multi-source verification, which comprises the following components: the heterogeneous acquisition module is configured with heterogeneous sensor arrays aiming at the same target physiological parameter, and the heterogeneous sensor arrays comprise at least three sensors based on different measurement principles and are used for acquiring multiple paths of original physiological signals of a user; The data confidence verification module is used for receiving the multiple paths of original physiological signals, executing dynamic consistency verification on the multiple paths of original physiological signals, and determining effective physiological parameter data according to a verification result; The knowledge fusion analysis module is used for storing a pre-constructed dynamic medical knowledge graph, extracting time sequence features from the effective physiological parameter data, matching corresponding clinical phenotype nodes in the dynamic medical knowledge graph based on the time sequence features, inputting the matched clinical phenotype nodes and covariate information of a user into a medical knowledge fusion reasoning model for calculation, and obtaining a posterior risk probability value of a target health abnormal condition; And the traceability feedback output module is used for generating a health monitoring report containing the posterior risk probability value, retrieving medical basis chain information associated with the clinical phenotype node from the dynamic medical knowledge graph and outputting the medical basis chain information and the health monitoring report in an associated mode. In an embodiment, the data confidence verification module is configured to calculate a deviation value between any two paths of signal output results in the multiple paths of original physiological signals in real time when executing dynamic consistency verification, judge whether the deviation value is smaller than a preset consistency threshold, identify corresponding data as the valid physiological parameter data if the deviation value of at least two independent sensor output results is recognized to be smaller than the consistency threshold, generate a confidence degradation prompt signal if the deviation value of all the sensor output results is recognized to be larger than the consistency threshold, and output prompt information to guide a user to adjust the contact state of the heterogeneous sensor array and the user or keep the static information state in response to the confidence degradation prompt signal. In an embodiment, the system further comprises an adaptive calibration module configured to: monitoring real-time physiological state data of a user, and identifying a time period in a stable physiological period as a reference calibration window; acquiring sensor measurement data within the reference calibration window to establish an individualized baseline mo