Search

CN-122025095-A - Data processing method and system for medical data

CN122025095ACN 122025095 ACN122025095 ACN 122025095ACN-122025095-A

Abstract

The invention belongs to the technical field of data processing, and particularly relates to a data processing method and system of medical data, wherein the method comprises the steps of acquiring multidimensional physiological time sequence data from a front-end acquisition module of medical monitoring equipment, wherein the physiological time sequence data comprises main monitoring sequence data and reference sequence data, and carrying out normalization processing on the physiological time sequence data; the method comprises the steps of calculating local form dispersion by using normalized data, restraining extreme abnormal values by means of logarithmic weighting terms, calculating artifact confidence level by combining the local form dispersion of a main monitoring sequence and the energy amplitude of a reference sequence, and mixing original data with predicted data by using a self-adaptive weighting strategy based on the artifact confidence level to obtain finally output cleaned data. The invention can accurately filter the quasi-pathological artifact and retain the real pathological details.

Inventors

  • LI NA
  • ZHAO XIANPU
  • Hou Feibiao
  • LIU JIE

Assignees

  • 河南爱尔康科技有限公司

Dates

Publication Date
20260512
Application Date
20260128

Claims (10)

  1. 1. A data processing method of medical data, comprising: acquiring multidimensional physiological time sequence data from a front-end acquisition module of medical monitoring equipment, wherein the physiological time sequence data comprises main monitoring sequence data and reference sequence data, and carrying out normalization processing on the physiological time sequence data; Calculating local morphology dispersion by utilizing the normalized data and combining a sliding time window, wherein the local morphology dispersion is used for evaluating the confusion degree of waveforms and inhibiting extreme outliers through logarithmic weighting terms; Calculating artifact confidence coefficient by combining the local form dispersion of the main monitoring sequence and the energy amplitude of the reference sequence, wherein the artifact confidence coefficient reflects the probability that the current signal is non-physiological interference; Based on the artifact confidence level, mixing the original data and the predicted data by utilizing a self-adaptive weighting strategy to obtain finally output cleaned data so as to filter the artifact and retain pathological characteristics.
  2. 2. A data processing method of medical data according to claim 1, wherein said normalizing said physiological time series data comprises: and carrying out normalization processing on the acquired physiological time sequence data, mapping the signal values of all dimensions into a preset value interval, and converting the signal values into a dimensionless pure value sequence.
  3. 3. A data processing method of medical data according to claim 1, wherein the local morphology dispersion satisfies the relation: ; In the formula, Indicating time of day Is used to determine the local morphology dispersion of (c) in the model, Indicating the length of the sliding time window, Indicating time of day Is used to determine the first order difference value of (1), Indicating time of day The local standard deviation within the window in which it is located, The stability constant is indicated as such, Indicating time of day Is a second order difference value of (2).
  4. 4.A data processing method of medical data according to claim 3, wherein the acquiring means of the first order difference value and the second order difference value comprises: Calculating the difference value between the main monitoring sequence data at the current moment and the main monitoring sequence data at the previous moment to obtain the first-order difference value; And calculating the difference value between the first-order difference value at the current moment and the first-order difference value at the previous moment to obtain the second-order difference value.
  5. 5. The method for processing medical data according to claim 1, wherein the calculation formula of the artifact confidence is: ; In the formula, Indicating time of day Is used to determine the artifact confidence level of (1), Indicating that the primary monitoring sequence is at time Is used to determine the local morphology dispersion of (c) in the model, Representing the morphological dispersion of the reference sequence, Representing the constant of the balancing factor, Representing the energy magnitudes of the reference sequence within the current window, Representing the baseline energy constant of the reference sequence.
  6. 6. The method for processing medical data according to claim 1, wherein the reference sequence data is data capable of reflecting environmental interference or body movement state, and comprises a triaxial acceleration signal or an impedance respiratory signal, and the calculated artifact confidence level is verified by coupling between morphological characteristics of physiological signals and energy characteristics of physical signals.
  7. 7. A data processing method of medical data according to claim 1, wherein the final output post-cleaning data is obtained by: Determining a weighted proportion of the original data value and the estimated value according to the artifact confidence; and carrying out weighted summation on the original data value acquired at the current moment and the estimated value at the current moment by using the weighted proportion to obtain the cleaned data.
  8. 8. A data processing method of medical data according to claim 7, wherein the estimated value of the current time is obtained by linear extrapolation of the history of the cleaned data or by autoregressive prediction based on history window data.
  9. 9. The method of claim 7, wherein said utilizing an adaptive weighting strategy comprises: The difference between the high confidence coefficient and the low confidence coefficient is nonlinearly increased by squaring the artifact confidence coefficient; When the artifact confidence is higher, increasing the weight of the estimated value; and when the artifact confidence is lower, increasing the weight of the original data value.
  10. 10. A data processing system for medical data, comprising a processor and a memory, the memory storing computer program instructions which, when executed by the processor, implement a method of data processing for medical data according to any one of claims 1-9.

Description

Data processing method and system for medical data Technical Field The invention relates to the technical field of data processing. More particularly, the present invention relates to a data processing method and system for medical data. Background In the field of modern intelligent medical treatment and intensive care, real-time and accurate physiological parameters of patients including electrocardio, blood oxygen, respiratory rate and the like are the basis for guaranteeing life safety, however, the working environment of medical monitoring equipment such as an intelligent monitoring mattress, a wearable monitor and the like is extremely complex, and the acquired signals are often interfered by the external environment, so that the data acquired by the medical monitoring equipment are generally processed by adopting frequency filtering or a statistical algorithm in the prior art. However, high-frequency electromagnetic noise or large-amplitude motion artifacts can be better processed by a frequency filtering or statistical algorithm mode, but when a patient is subjected to shivering or rhythmic beating action, a noise with frequency similar to that of a real bioelectric signal but with disordered waveform morphology is superimposed on a signal acquired by a sensor, and the existing equipment is difficult to distinguish the pathological artifact from the real pathological mutation, so that the false alarm rate of the equipment is extremely high. Therefore, how to accurately identify and filter the quasi-pathological artifacts caused by the micro body movement by mining the morphological logic characteristics of the data, and simultaneously, to keep the real pathological mutation details to the greatest extent is a technical problem to be solved by the technicians in the field. Disclosure of Invention In order to solve the problem that the prior art is difficult to accurately filter the quasi-pathological artifact, the invention provides a data processing method and a data processing system for medical data, which can accurately filter the quasi-pathological artifact and retain real pathological details. In a first aspect, the invention provides a data processing method of medical data, which comprises the steps of acquiring multidimensional physiological time series data from a front-end acquisition module of medical monitoring equipment, carrying out normalization processing on the physiological time series data, calculating local form dispersion by combining sliding time windows by utilizing the normalized data, wherein the local form dispersion is used for evaluating the confusion degree of waveforms, inhibiting extreme abnormal values by logarithmic weighting terms, calculating artifact confidence by combining the local form dispersion of a main monitoring sequence and the energy amplitude of a reference sequence, wherein the artifact confidence reflects the probability that a current signal is non-physiological interference, and mixing original data with predicted data by utilizing a self-adaptive weighting strategy based on the artifact confidence to obtain finally output cleaned data so as to filter artifacts and retain pathological characteristics. By adopting the technical scheme, a set of linear logic operation system is constructed, and the abnormal mechanical tremors and regular pathological mutations are sharply distinguished by introducing the logarithmic weighted terms of second-order difference through the local form dispersion index, so that the technical problem that the ventricular tremors are misjudged as noise points or the beating is misjudged as ventricular tremors in the traditional algorithm is solved, and the real pathological details can be accurately reserved while the quasi-pathological artifacts are filtered. Preferably, the normalizing the physiological time series data comprises normalizing the acquired physiological time series data, mapping the signal values of all dimensions into a preset value interval, and converting the signal values into a dimensionless pure value sequence. By adopting the technical scheme, the acquired physiological time sequence data is normalized, the signal values of all dimensions are mapped into the preset value interval and converted into the dimensionless pure value sequence, so that the dimensional difference of different sensor data can be eliminated, the following algorithm can fuse multidimensional data characteristics, the weight deviation caused by different factor value ranges is avoided, and a unified data base is laid for the following coupling analysis based on morphology and energy. Preferably, the local morphology dispersion satisfies the relation: ; In the formula, Indicating time of dayIs used to determine the local morphology dispersion of (c) in the model,Indicating the length of the sliding time window,Indicating time of dayIs used to determine the first order difference value of (1),Indicating time of dayThe local standard d