CN-122025137-A - Physiological state monitoring method, device, equipment and medium based on multi-time scale constraint

CN122025137ACN 122025137 ACN122025137 ACN 122025137ACN-122025137-A

Abstract

The application discloses a physiological state monitoring method, device, equipment and medium based on multi-time scale constraint, which relate to the technical field of medical information processing and data analysis and are applied to a computer device and comprise the steps of determining target physiological baseline values of all modes; the method comprises the steps of obtaining current physiological signals of all modes in a current monitoring time period, extracting basic physiological characteristics under a multi-time scale sliding window from the current signals based on a consistency constraint mechanism to generate physiological change characteristics, analyzing cross-mode joint characteristics according to the basic physiological characteristics and the change characteristics, constructing state units containing all the characteristics, respectively organizing all the state units into state sequences of corresponding time scales to obtain a multi-scale physiological state sequence of the current monitoring time period, and adding the sequence to the end of a last sequence to obtain the latest physiological state monitoring result. The pertinence and the accuracy of the physiological state evaluation and the collaborative characterization of the multi-modal data are realized, and the continuity of the full-period data is ensured.

Inventors

BI YUFANG
NING GUANG
XU YU
LI MIAN
WANG BIN
XU YUCHEN

Assignees

上海交通大学医学院附属瑞金医院

Dates

Publication Date: 20260512
Application Date: 20260202

Claims (10)

1. A physiological state monitoring method based on a multi-time scale constraint, applied to a computer device, comprising: Determining a target physiological baseline value of each mode according to the multi-mode historical physiological signals of the target object in the historical monitoring time period; Acquiring current physiological signals of all modes of the target object in a current monitoring time period, extracting target basic physiological characteristics of all modes under each sliding window of a multi-time scale from the current physiological signals based on a consistency constraint mechanism, and generating physiological change characteristics based on the current physiological signals and the target physiological baseline value; Analyzing cross-modal joint characteristics used for representing joint relations among different modalities under each sliding window according to the basic physiological characteristics and the physiological change characteristics; constructing state units under each sliding window containing the basic physiological characteristics, the physiological change characteristics and the cross-modal joint characteristics, and respectively organizing each state unit into a state sequence of a corresponding time scale; And carrying out structuring treatment on the state sequences with different time scales to obtain a multi-scale physiological state sequence of the current monitoring time period, and adding the multi-scale physiological state sequence of the current monitoring time period to the tail of the multi-scale physiological state sequence of the previous monitoring time period to obtain the latest physiological state monitoring result.
2. The method for monitoring physiological condition based on multi-time scale constraint according to claim 1, wherein determining a target physiological baseline value of each modality from a multi-modal historical physiological signal of the target subject during the historical monitoring period comprises: determining a historical monitoring time period based on the starting time of the current monitoring time period and a preset baseline analysis time period, and acquiring a multi-mode historical physiological signal of a target object in the historical monitoring time period; for non-rhythmic modes, determining any one or more characteristic values of a moving average value, a median value, a mode value and a peak value of the historical physiological signals as initial physiological baseline values of the modes; For a rhythmic mode, dividing the historical physiological signal according to the rhythmic mode, and determining a target characteristic value of the historical physiological signal in each period as an initial physiological baseline value in each period; And determining a target physiological baseline value of each mode according to the standard deviation of the historical physiological signals and the initial physiological baseline value.
3. The method for monitoring physiological condition based on multi-time scale constraint according to claim 2, wherein determining a target physiological baseline value for each modality based on the standard deviation of the historical physiological signal and the initial physiological baseline value comprises: Judging whether the initial physiological baseline value of the current mode meets a preset abnormal condition according to the standard deviation of the historical physiological signal of the current mode; if the initial physiological baseline value of the current mode does not meet the preset abnormal condition, determining the initial physiological baseline value as a target physiological baseline value of the current mode; And if the initial physiological baseline value of the current mode meets the preset abnormal condition, adjusting the initial physiological baseline value of the current mode of the target object based on the historical physiological signals of the current modes of other objects so as to obtain a target physiological baseline value of the current mode of the target object.
4. The physiological state monitoring method based on multi-time scale constraint according to claim 1, wherein the extracting, based on a consistency constraint mechanism, target basic physiological features of each modality under each sliding window of multi-time scale from the current physiological signal comprises: Determining the microscopic end time of the last sliding window of the preset microscopic time scale, the mesoscopic end time of the last sliding window of the preset mesoscopic time scale and the macroscopic end time of the last sliding window of the preset macroscopic time scale in the historical monitoring time period; Determining a first difference value between the ending time of the current monitoring time period and the macroscopic ending time, a second difference value between the current monitoring time period and the mesoscopic ending time, and a third difference value between the current monitoring time period and the microscopic ending time; if the first difference value is larger than the window size of the preset macroscopic time scale, dividing the current monitoring time period into microscopic sliding windows, mesoscopic sliding windows and macroscopic sliding windows according to the window size corresponding to the preset microscopic time scale, the window size corresponding to the preset mesoscopic time scale and the window size corresponding to the preset macroscopic time scale respectively; If the first difference value is not larger than the window size of the preset macroscopic time scale and the second difference value is larger than the window size of the preset mesoscopic time scale, dividing the current monitoring time period into microscopic sliding windows and mesoscopic sliding windows according to the window size corresponding to the preset microscopic time scale and the window size corresponding to the preset mesoscopic time scale respectively; if the second difference value is not larger than the window size of the preset mesoscopic time scale and the third difference value is larger than the window size of the preset microcosmic time scale, dividing the current monitoring time period into microcosmic sliding windows according to the window size corresponding to the preset microcosmic time scale; and extracting target basic physiological characteristics of each mode under each microscopic sliding window, each mesoscopic sliding window and each macroscopic sliding window from the current physiological signal based on a consistency constraint mechanism.
5. The physiological state monitoring method based on multi-time scale constraints according to claim 4, wherein the extracting target basic physiological features of each microscopic sliding window, each mesoscopic sliding window and each mode under each macroscopic sliding window from the current physiological signal based on a consistency constraint mechanism comprises: extracting target basic physiological characteristics of each mode under each microscopic sliding window from the current physiological signals; If each mesoscopic sliding window exists, aggregating target basic physiological characteristics of each mode under each microcosmic sliding window in the mesoscopic sliding window based on a consistency constraint mechanism to obtain target basic physiological characteristics of each mode under the mesoscopic sliding window; if each macroscopic sliding window exists, the target basic physiological characteristics of each mode under each microscopic sliding window in the macroscopic sliding window are aggregated based on a consistency constraint mechanism, so that the target basic physiological characteristics of each mode under the macroscopic sliding window are obtained.
6. The physiological state monitoring method based on multi-time scale constraint according to claim 5, wherein the evolution direction of the target basic physiological feature under the current macroscopic sliding window is consistent with the evolution direction of the target basic physiological feature under each microscopic sliding window in the current macroscopic sliding window, and the evolution direction of the target basic physiological feature under the current mesoscopic sliding window is consistent with the evolution direction of the target basic physiological feature under each microscopic sliding window in the current mesoscopic sliding window.
7. The physiological state monitoring method based on multi-time scale constraints according to any one of claims 1 to 6, wherein said structuring of said state sequences at different time scales to obtain a multi-scale physiological state sequence for said current monitored time period comprises: generating metadata of the state sequence with different time scales, wherein the metadata comprises the time scales, window parameters and generation time of the state sequence; and arranging the state sequence and the corresponding metadata according to a time sequence, and taking the generation time and the time scale as a first index and a second index of the state sequence to obtain the multi-scale physiological state sequence of the current monitoring time period.
8. A physiological condition monitoring device based on a multi-time scale constraint, for application to a computer device, comprising: The baseline value determining module is used for determining target physiological baseline values of all modes according to the multi-mode historical physiological signals of the target object in the historical monitoring time period; The first feature generation module is used for acquiring current physiological signals of all modes of the target object in a current monitoring time period, extracting target basic physiological features of all modes under each sliding window of a multi-time scale from the current physiological signals based on a consistency constraint mechanism, and generating physiological change features based on the current physiological signals and the target physiological baseline value; The second feature generation module is used for analyzing cross-mode joint features used for representing joint relations among different modes under each sliding window according to the basic physiological features and the physiological change features; The sequence organization module is used for constructing state units under each sliding window containing the basic physiological characteristics, the physiological change characteristics and the cross-modal joint characteristics, and respectively organizing each state unit into a state sequence with a corresponding time scale; The monitoring result acquisition module is used for carrying out structuring processing on the state sequences of different time scales to obtain a multi-scale physiological state sequence of the current monitoring time period, and adding the multi-scale physiological state sequence of the current monitoring time period to the tail end of the multi-scale physiological state sequence of the last monitoring time period to obtain the latest physiological state monitoring result.
9. An electronic device, comprising: A memory for storing a computer program; a processor for executing the computer program to implement the steps of the physiological condition monitoring method based on the multi-time scale constraint as claimed in any one of claims 1 to 7.
10. A computer readable storage medium for storing a computer program, wherein the computer program when executed by a processor implements the steps of the physiological state monitoring method based on multi-time scale constraints of any one of claims 1 to 7.

Description

Physiological state monitoring method, device, equipment and medium based on multi-time scale constraint Technical Field The invention relates to the technical field of medical information processing and data analysis, in particular to a physiological state monitoring method, device, equipment and medium based on multi-time scale constraint. Background With the rapid development of wearable devices, remote monitoring systems and electronic health records, multi-modal physiological data (including blood pressure, heart rate, electrocardio, blood oxygen, activity level, sleep parameters and the like) of an individual can be continuously or intermittently collected in a daily life scene, the data cover different sampling frequencies, different data sources (wearable devices, home monitoring devices, hospital information systems and the like) and different data types (continuous waveforms, discrete measurement values, event marks and the like), and how to uniformly represent the multi-modal heterogeneous physiological data in a time dimension is an important technical problem faced by the current medical information processing field, and an effective data representation method can provide a standardized data interface for an upper application system (such as health management, remote monitoring, auxiliary decision making and the like) and support long-term tracking and systematic management of the health state of the individual. In summary, how to realize pertinence and accuracy of physiological state evaluation and collaborative characterization of multi-modal data and ensure continuity of full-period data is a problem to be solved in the field. Disclosure of Invention In view of the above, the present invention aims to provide a physiological state monitoring method, device, equipment and medium based on multi-time scale constraint, which realize pertinence and accuracy of physiological state evaluation, collaborative characterization of multi-modal data, and ensure continuity of full-period data. The specific scheme is as follows: In a first aspect, the application discloses a physiological state monitoring method based on multi-time scale constraint, which is applied to a computer device and comprises the following steps: Determining a target physiological baseline value of each mode according to the multi-mode historical physiological signals of the target object in the historical monitoring time period; Acquiring current physiological signals of all modes of the target object in a current monitoring time period, extracting target basic physiological characteristics of all modes under each sliding window of a multi-time scale from the current physiological signals based on a consistency constraint mechanism, and generating physiological change characteristics based on the current physiological signals and the target physiological baseline value; Analyzing cross-modal joint characteristics used for representing joint relations among different modalities under each sliding window according to the basic physiological characteristics and the physiological change characteristics; constructing state units under each sliding window containing the basic physiological characteristics, the physiological change characteristics and the cross-modal joint characteristics, and respectively organizing each state unit into a state sequence of a corresponding time scale; And carrying out structuring treatment on the state sequences with different time scales to obtain a multi-scale physiological state sequence of the current monitoring time period, and adding the multi-scale physiological state sequence of the current monitoring time period to the tail of the multi-scale physiological state sequence of the previous monitoring time period to obtain the latest physiological state monitoring result. Optionally, the determining the target physiological baseline value of each mode according to the multi-mode historical physiological signals of the target object in the historical monitoring time period includes: determining a historical monitoring time period based on the starting time of the current monitoring time period and a preset baseline analysis time period, and acquiring a multi-mode historical physiological signal of a target object in the historical monitoring time period; for non-rhythmic modes, determining any one or more characteristic values of a moving average value, a median value, a mode value and a peak value of the historical physiological signals as initial physiological baseline values of the modes; For a rhythmic mode, dividing the historical physiological signal according to the rhythmic mode, and determining a target characteristic value of the historical physiological signal in each period as an initial physiological baseline value in each period; And determining a target physiological baseline value of each mode according to the standard deviation of the historical physiological signals and the initial