Newborn breathing abnormity early warning system

Abstract

The invention relates to the technical field of neonatal respiration monitoring, and discloses a neonatal abnormal respiration early warning system. A physiological parameter acquisition module of the system acquires physiological parameter information of a target newborn in real time, and an analysis instruction is triggered after a physiological parameter abnormal signal is obtained; the breathing mode recognition module recognizes breathing mode characteristics and analyzes the breathing mode characteristics to obtain a breathing mode evaluation result; the abnormal feature extraction module extracts abnormal breathing features and analyzes the abnormal breathing features to obtain an abnormal feature evaluation value; the early warning grade evaluation module receives the result and the evaluation value, and obtains a low-risk early warning signal and a high-risk early warning signal through evaluation; the early warning signal generation module selects an early warning type according to the received signal, and obtains parameters of sound early warning and acousto-optic early warning which should be triggered. According to the system, dynamic monitoring and graded early warning of the neonatal breathing abnormity are achieved through multi-module cooperation, and support is provided for timely intervention of the neonatal breathing abnormity.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of neonatal respiratory monitoring, in particular to a neonatal respiratory abnormality early warning system. BACKGROUND

[0002] Newborns have not yet developed a perfect respiratory system, and their respiratory regulation function is weak. The frequency, rhythm and depth of their breath are prone to abnormalities. In clinical practice, respiratory abnormalities such as apnea, tachypnea and periodic breathing are common. If these conditions are not detected and addressed in a timely manner, they can lead to hypoxemia, brain damage, and even more serious consequences. In particular, premature babies have immature lungs and poor respiratory center regulation, so the incidence of respiratory abnormalities is higher, the duration is longer, and the threat to life and health is greater. The devices and methods used for neonatal respiratory monitoring in clinical practice have many limitations. Traditional monitoring devices mainly monitor a single parameter, such as blood oxygen saturation or respiratory rate. This approach cannot fully reflect the respiratory status of newborns. For example, when a newborn has shallow and slow breathing but the blood oxygen has not yet decreased significantly, blood oxygen monitoring alone cannot detect the abnormality in a timely manner. Simply relying on respiratory rate monitoring can miss key information such as respiratory rhythm disorders and sudden changes in respiratory depth, resulting in missed early abnormal signals.

[0003] Manual observation is an important supplement to existing monitoring methods, but it is difficult to achieve continuous and detailed monitoring of each newborn due to the work intensity and energy distribution of medical staff. In the neonatal intensive care unit, medical staff often need to care for multiple patients at the same time, and observation errors are more likely to occur at night or during busy periods. In addition, there are differences in experience and judgment criteria among different medical staff, which can lead to deviations in the identification and assessment of respiratory abnormalities, affecting the consistency and timeliness of intervention measures. The existing early warning system also has obvious shortcomings. Most systems use a unified warning method, regardless of the severity of the abnormality, and the same warning signal is issued, which can lead to two extreme situations: on the one hand, frequent warnings of minor abnormalities can cause medical staff to become desensitized, leading to "warning fatigue" and reducing sensitivity to real danger signals; on the other hand, the warning signal lacks sufficient differentiation when the abnormality is severe, and cannot quickly attract the priority attention of medical staff, which may delay the best intervention opportunity. There is a lack of effective linkage between physiological parameter collection and abnormality analysis in existing technology, and manual judgment is often required to determine whether to initiate in-depth analysis, which results in a time difference between the occurrence of an abnormal parameter and the start of detailed analysis, further increasing the risk of delay. These problems collectively result in suboptimal early intervention for neonatal respiratory abnormalities, and there is an urgent need for a system that can achieve real-time monitoring, accurate analysis and graded warning to improve this situation. SUMMARY

[0004] The present application aims to provide a new-born respiratory abnormality early warning system to solve the problems raised in the background art.

[0005] To achieve the above-mentioned purpose, the present application provides a new-born respiratory abnormality early warning system, which comprises: a physiological parameter acquisition module, which is used to acquire the physiological parameter information of the target new-born in real time, to obtain the physiological parameter information of the target new-born, to preliminarily judge the physiological state of the target new-born, to obtain a physiological parameter abnormality signal, to trigger an analysis instruction according to the obtained physiological parameter abnormality signal, and to execute a respiratory pattern recognition module and an abnormality feature extraction module according to the triggered analysis instruction; a respiratory pattern recognition module, which is used to recognize the respiratory pattern features of the target new-born, to analyze the respiratory pattern of the target new-born, to obtain a respiratory pattern evaluation result, and to send the result to an early warning level evaluation module; an abnormality feature extraction module, which is used to extract the abnormal respiratory features of the target new-born, to analyze the abnormality degree of the target new-born, to obtain an abnormality feature evaluation value, and to send the value to the early warning level evaluation module; an early warning level evaluation module, which is used to receive the respiratory pattern evaluation result and the abnormality feature evaluation value, to evaluate and analyze the respiratory abnormality level of the target new-born, to obtain a low-risk early warning signal and a high-risk early warning signal, and to send the signals to an early warning signal generation module; an early warning signal generation module, which is used to receive the low-risk early warning signal and the high-risk early warning signal, to select the type of early warning signal, and to obtain the parameters of the sound early warning to be triggered and the sound-light early warning to be triggered.

[0006] Preferably, the physiological parameter information of the target new-born is acquired in real time, and the specific acquisition process is as follows: The respiratory frequency, the blood oxygen saturation, the heart rate and the chest and abdominal movement amplitude of the target new-born in the current monitoring period are continuously acquired to obtain the respiratory frequency value, the blood oxygen saturation value, the heart rate value and the chest and abdominal movement amplitude value of the target new-born in the current monitoring period, the values of the four are extracted for standardized conversion to obtain a physiological parameter comprehensive value; By collecting respiratory waveform signals from the target newborn during the current monitoring period, the respiratory waveform signals at each sampling time within the current monitoring period are obtained and marked as waveform sampling values. A waveform amplitude threshold is set, and the waveform sampling values ​​are compared and analyzed with the waveform amplitude threshold. When the waveform sampling value is greater than or equal to the waveform amplitude threshold, the sampling time is determined as a valid breathing point; when the waveform sampling value is less than the waveform amplitude threshold, the sampling time is determined as an invalid breathing point. The number of times the target newborn is determined as a valid breathing point within the current monitoring period is counted, and the percentage is calculated by comparing it with the total number of sampling times to obtain the effective breathing percentage. At the same time, the respiratory interval time of the target newborn during the current monitoring period is collected to obtain the interval time of each breath within the current monitoring period, and these intervals are marked as interval duration values. The mean of each interval duration value within the current monitoring period is calculated to obtain the mean interval duration within the current monitoring period. The values ​​of the effective breathing percentage, the mean interval duration, and the comprehensive physiological parameter value within the current monitoring period are extracted, multiplied by their respective weighting coefficients, and then summed to obtain the physiological state assessment value. The physiological state assessment value is extracted and its range is determined. The physiological state assessment value is compared and analyzed with the preset physiological parameter abnormality threshold. When the physiological state assessment value is greater than or equal to the preset physiological parameter abnormality threshold, a physiological parameter abnormality signal is generated.

[0007] Preferably, the respiratory pattern characteristics of the target newborn are identified, and the specific identification process is as follows: By synchronously acquiring the respiratory waveform signal and chest and abdominal movement signal of the target newborn during the current monitoring period, the respiratory waveform signal and chest and abdominal movement signal of the target newborn during the current monitoring period are obtained. Based on this, the respiratory waveform diagram and chest and abdominal movement curve diagram of the target newborn during the current monitoring period are generated through signal processing methods. The system database extracts reference respiratory waveforms of normal newborns. The respiratory waveforms of the target newborn during the current monitoring period are compared with those of the reference respiratory waveforms of normal newborns to obtain the morphological similarity of the respiratory waveforms of the target newborn during the current monitoring period. The waveforms are then classified into different levels, and the level value is recorded as the morphological similarity level. Obtain the waveform period parameters of the reference respiratory waveform of a normal newborn and record them as the reference period values; The peak interval and trough interval are extracted from the respiratory waveform of the target newborn during the current monitoring period and are denoted as peak interval and trough interval, respectively. Reference peak interval and reference trough interval are extracted from the reference respiratory waveform of normal newborns and are denoted as reference peak interval and reference trough interval, respectively. Based on a comprehensive assessment of waveform morphology similarity, peak interval deviation, and trough interval deviation, a respiratory pattern consistency score for the target newborn during the current monitoring period is obtained. Reference chest and abdominal movement curves of normal newborns are extracted from the system database. The chest and abdominal movement curves of the target newborn during the current monitoring period are compared and analyzed with the reference chest and abdominal movement curves of normal newborns to obtain the synchronicity deviation of chest and abdominal movement in the target newborn during the current monitoring period. The corresponding synchronicity deviation segments are extracted from the chest and abdominal movement curves of the target newborn during the current monitoring period as the respiratory coordination abnormal segments in the current monitoring period of the target newborn. The number of respiratory coordination abnormal segments in the current monitoring period of the target newborn is counted and the value is recorded as the number of coordination abnormalities. Extract the duration of each abnormal respiratory coordination segment in the current monitoring period of the target newborn from the chest and abdominal motion curve graph of the target newborn in the current monitoring period, obtain the duration of each abnormal respiratory coordination segment in the current monitoring period of the target newborn, and take its value as the abnormal duration. The respiratory pattern assessment results are obtained by comprehensively calculating the respiratory pattern consistency score, the number of coordination abnormalities, and the duration of abnormalities.

[0008] Preferably, abnormal respiratory features of the target newborn are extracted. The specific extraction process is as follows: By detecting apnea events in the target newborn during the current monitoring period and recording the decrease in blood oxygen saturation during the current monitoring period, the maximum decrease in blood oxygen saturation is extracted and recorded as the blood oxygen decrease value. Then, the correlation between the duration of apnea events and the blood oxygen decrease value is analyzed to obtain the correlation value of the target newborn during the current monitoring period. By calculating the rate of change of respiratory rate of the target newborn during the current monitoring period, and extracting the fluctuation range of respiratory rate from the data of respiratory rate change of the target newborn during the current monitoring period, the fluctuation value of respiratory rate of the target newborn during the current monitoring period is obtained. At the same time, the number of times the respiratory rate exceeds the normal range is extracted from the data of respiratory rate change of the target newborn during the current monitoring period, thus obtaining the number of frequency abnormalities of the target newborn during the current monitoring period. By acquiring the heart rate variability index of the target newborn during the current monitoring period, the heart rate variability value of the target newborn during the current monitoring period is obtained. The relevant values, respiratory rate fluctuation values, frequency abnormality values, and heart rate variability values ​​of the target newborn during the current monitoring period are extracted and processed according to a multi-dimensional feature fusion algorithm to obtain abnormal feature evaluation values.

[0009] Preferably, the respiratory abnormality level of the target newborn is assessed and analyzed. The specific analysis process is as follows: Retrieve the physiological status assessment values ​​of the target newborn during the current monitoring period, set the influence coefficient of the physiological status assessment values ​​of the target newborn during the current monitoring period, and process them according to the influence coefficient conversion rules to obtain the converted physiological influence value; The respiratory pattern assessment results, abnormal feature assessment values, and converted physiological impact values ​​of the target newborn during the current monitoring period are extracted, and then normalized to obtain the comprehensive assessment value of respiratory abnormality. Set a grading threshold for the comprehensive assessment value of respiratory abnormalities. Compare and analyze the comprehensive assessment value of respiratory abnormalities with the grading threshold. When the comprehensive assessment value of respiratory abnormalities is less than the first grading threshold, a low-risk warning signal is generated. When the comprehensive assessment value of respiratory abnormalities is greater than or equal to the first grading threshold and less than the second grading threshold, a medium-risk warning signal is generated. When the comprehensive assessment value of respiratory abnormalities is greater than or equal to the second grading threshold, a high-risk warning signal is generated.

[0010] Preferably, the type of warning signal is selected, and the specific selection process is as follows: If a low-risk warning signal is detected, a Level 1 warning instruction is triggered. Based on the triggered Level 1 warning instruction, the warning method for the current monitoring period of the target newborn is selected, thereby determining the sound warning to be triggered. If a medium-risk warning signal is detected, a level-two warning instruction is triggered. Based on the triggered level-two warning instruction, the warning method for the target newborn within the current monitoring period is selected, thereby determining the appropriate sound and light combination warning to be triggered. If a high-risk warning signal is detected, a Level 3 warning instruction is triggered. Based on the triggered Level 3 warning instruction, the warning method and notification frequency for the target newborn within the current monitoring period are selected, thereby obtaining the parameters for triggering the combined sound and light warning and remote notification.

[0011] Preferably, the physiological parameter acquisition module further includes a data preprocessing submodule, used to preprocess the acquired raw physiological parameter information. The specific preprocessing process is as follows: By smoothing and filtering the raw respiratory rate data collected during the current monitoring period of the target newborn, high-frequency noise interference is removed, and filtered respiratory rate data is obtained. By detecting outliers in the blood oxygen saturation data collected during the current monitoring period of the target newborn, abnormal data points that exceed the physiological range are identified and removed, and corrected blood oxygen saturation data is obtained. By performing baseline drift correction on the chest and abdominal motion signals collected during the current monitoring period of the target newborn, the signal baseline shift during long-term monitoring is eliminated, and the corrected chest and abdominal motion signals are obtained. The filtered respiratory rate data, corrected blood oxygen saturation data, and corrected chest and abdominal motion signals are time-synchronized and aligned to ensure the consistency of each parameter in the time dimension, resulting in a preprocessed physiological parameter dataset.

[0012] Preferably, it also includes a historical data comparison module, used to compare and analyze the current physiological parameter information with historical data. The specific comparison process is as follows: Historical physiological parameter data of the target newborn over the past 72 hours were extracted from the system storage unit, including mean respiratory rate, mean blood oxygen saturation, and respiratory pattern characteristic parameters, to establish a historical reference dataset; The preprocessed physiological parameter dataset of the target newborn during the current monitoring period is compared item by item with the historical reference dataset, and the percentage deviation between the current value and the historical mean of each parameter is calculated. The number of parameter items whose percentage of statistical deviation exceeds the preset deviation threshold is recorded as the number of abnormal parameters; When the number of abnormal parameters is greater than or equal to the preset threshold, the analysis sensitivity of the abnormal feature extraction module is enhanced; when the number of abnormal parameters is less than the preset threshold, the default analysis sensitivity of the abnormal feature extraction module is maintained.

[0013] Preferably, the early warning level assessment module further includes a multi-parameter fusion analysis submodule, which is used to comprehensively assess abnormal early warning based on multiple physiological parameters.

[0014] Preferably, the early warning level assessment module specifically comprises: The weight coefficients of respiratory pattern assessment results, abnormal feature assessment values, and converted physiological impact values ​​in the comprehensive assessment were determined by the analytic hierarchy process. Among them, the weight coefficient of respiratory pattern assessment results was the highest, followed by abnormal feature assessment values, and the weight coefficient of converted physiological impact values ​​was the lowest. Each evaluation value is multiplied by its corresponding weighting coefficient and then summed to obtain a preliminary integrated evaluation value. The preliminary fusion evaluation value is nonlinearly corrected by fuzzy logic algorithm to eliminate the dimensional differences and coupling interference between different parameters, and the final comprehensive evaluation value of respiratory abnormality is obtained. The final comprehensive assessment value of respiratory abnormalities is compared with the preset grading threshold to determine the corresponding warning level.

[0015] Compared with the prior art, the beneficial effects of the present invention are: This neonatal respiratory abnormality early warning system, through the coordinated operation of multiple modules, constructs a complete process from physiological parameter acquisition to early warning signal output, effectively solving many problems in existing neonatal respiratory monitoring. The physiological parameter acquisition module can capture the physiological parameters of the target newborn in real time. Once an abnormality is detected, it automatically triggers subsequent analysis instructions without relying on manual operation to start the analysis process. This real-time triggering mechanism shortens the interval from the occurrence of parameter abnormalities to the start of analysis, avoiding missed detection or delayed processing of abnormal signals due to human intervention. Compared with the disconnect between parameter acquisition and analysis in traditional monitoring equipment, this module makes the entire monitoring process more immediate, enabling the capture of early signs of respiratory abnormalities at the first moment. The breathing pattern recognition module focuses on identifying and analyzing the characteristics of newborns' breathing patterns. It not only considers basic data such as respiratory rate but also delves into pattern features such as changes in respiratory rhythm and depth, thus generating a comprehensive breathing pattern assessment. This overcomes the limitations of existing technologies that rely on a single parameter to determine respiratory status, enabling more accurate identification of special breathing patterns such as periodic breathing and apnea, providing richer information for subsequent risk assessment. The abnormal feature extraction module specifically extracts abnormal respiratory features. By analyzing detailed features such as apnea duration, respiratory rate fluctuation amplitude, and the correlation between blood oxygen saturation and respiratory rhythm, it generates specific abnormal feature assessment values. This refined analysis of the degree of abnormality compensates for the ambiguity in the judgment of the degree of abnormality in the existing system, allowing medical staff to clearly understand the severity of abnormalities and avoiding overreaction to minor abnormalities or underreaction to severe abnormalities. The early warning level assessment module combines respiratory pattern assessment results with abnormal feature assessment values ​​to comprehensively evaluate the level of respiratory abnormalities, classifying them into low-risk and high-risk warning levels. This tiered approach avoids the problems of existing early warning systems that rely on single warning signals and cannot differentiate between levels of urgency. It allows medical staff to quickly determine the priority of treatment based on the warning level, rationally allocate medical resources, and ensure that limited human and material resources are prioritized for handling high-risk situations. The warning signal generation module selects the corresponding warning type based on different warning levels: an audible warning is triggered for low-risk situations, while an audible and visual warning is triggered for high-risk situations. This differentiated warning method makes the warning signals more targeted, avoiding unnecessary stimulation to newborns from frequent bright lights or high-decibel sounds, thus protecting their physiological and psychological state. Furthermore, in high-risk situations, the combination of sound and light ensures that medical staff can detect the warning immediately and take timely intervention measures. Through the organic cooperation of its various modules, the entire system has improved in terms of real-time performance, analysis accuracy, and targeted early warning, enabling it to more effectively address the monitoring and early warning needs of abnormal breathing in newborns and providing more reliable technical support for the health monitoring of newborns. Attached Figure Description

[0016] Figure 1This is a schematic diagram illustrating the working principle of the neonatal respiratory abnormality early warning system described in this invention. Figure 2 A flowchart for real-time acquisition of physiological parameter information; Figure 3 Flowchart for respiratory pattern feature recognition; Figure 4 This is a schematic diagram illustrating the working principle of the multi-parameter fusion analysis submodule. Detailed Implementation

[0017] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0018] Please see Figure 1 This invention provides a neonatal respiratory abnormality early warning system, which includes a physiological parameter acquisition module, a breathing pattern recognition module, an abnormal feature extraction module, an early warning level assessment module, and an early warning signal generation module. The specific implementation steps are as follows: The physiological parameter acquisition module collects physiological parameter information of the target newborn in real time, obtains the physiological parameter information of the target newborn, makes a preliminary judgment on the physiological state of the target newborn, obtains abnormal physiological parameter signals, triggers analysis instructions based on the obtained abnormal physiological parameter signals, and executes the breathing pattern recognition module and abnormal feature extraction module based on the triggered analysis instructions.

[0019] The breathing pattern recognition module identifies the breathing pattern characteristics of the target newborn, analyzes the breathing pattern of the target newborn, obtains the breathing pattern assessment result, and sends it to the early warning level assessment module.

[0020] The abnormal feature extraction module extracts the abnormal breathing features of the target newborn, analyzes the degree of abnormality of the target newborn, obtains the abnormal feature assessment value, and sends it to the early warning level assessment module.

[0021] The early warning level assessment module receives the respiratory pattern assessment results and abnormal feature assessment values, thereby assessing and analyzing the respiratory abnormality level of the target newborn, obtaining low-risk and high-risk early warning signals, and sending them to the early warning signal generation module.

[0022] The warning signal generation module receives low-risk and high-risk warning signals, and selects the warning signal type accordingly to obtain parameters for triggering sound and light warnings.

[0023] Example 1: See Figure 2 This study continuously collects data on the respiratory rate, blood oxygen saturation, heart rate, and chest and abdominal movement amplitude of the target newborn during the current monitoring period. The respiratory rate is obtained by recording the number of breaths taken by the target newborn per unit time using dedicated monitoring equipment. Blood oxygen saturation is obtained by monitoring the proportion of oxygen-bound hemoglobin in the blood using sensors. Heart rate is calculated by monitoring the number of heartbeats. Chest and abdominal movement amplitude is obtained by capturing displacement changes during respiration using sensors attached to the chest and abdomen of the target newborn. These four values ​​are then extracted and standardized. During the standardization process, the original values ​​of each parameter are adjusted to the same data range according to a preset mapping rule, thus obtaining a comprehensive physiological parameter value.

[0024] By collecting respiratory waveform signals from the target newborn during the current monitoring period, the respiratory waveform signals at each sampling time within the current monitoring period are obtained and marked as waveform sample values. A waveform amplitude threshold is set, which is based on the respiratory physiological characteristics of newborns and reflects the lowest amplitude level of the waveform signal during normal breathing. The waveform sample values ​​are compared and analyzed with the waveform amplitude threshold. When the waveform sample value is greater than or equal to the waveform amplitude threshold, the sampling time is determined as a valid breathing point; when the waveform sample value is less than the waveform amplitude threshold, the sampling time is determined as an invalid breathing point. The number of times the target newborn is determined as a valid breathing point within the current monitoring period is counted, and this number is calculated as a percentage of the total number of samplings to obtain the percentage of valid breathings.

[0025] Simultaneously, respiratory intervals of the target newborns during the current monitoring period are collected to obtain the interval time of each breath during the current monitoring period, and these intervals are labeled as interval duration values. The mean of each interval duration value during the current monitoring period is calculated by summing all interval duration values ​​and dividing by the number of intervals.

[0026] Extract the values ​​of effective respiratory rate, average interval duration, and comprehensive physiological parameter value of the target newborn during the current monitoring period. Multiply these three values ​​by their respective weighting coefficients, which are pre-set based on the correlation between each parameter and its respiratory status. Then, add the products together to obtain the physiological status assessment value.

[0027] The physiological state assessment values ​​are extracted and their ranges are determined. These values ​​are then compared and analyzed with preset abnormal physiological parameter thresholds. These thresholds are determined based on the normal range of neonatal respiratory physiology, encompassing a comprehensive calculation of the normal fluctuation ranges of parameters such as respiratory rate, blood oxygen saturation, and heart rate. When a physiological state assessment value is greater than or equal to the preset abnormal physiological parameter threshold, an abnormal physiological parameter signal is generated.

[0028] Throughout the data acquisition process, the monitoring cycle can be set according to actual monitoring needs, such as every 5 minutes. Within each cycle, various parameters are continuously collected to ensure data continuity and timeliness. For respiratory rate, blood oxygen saturation, heart rate, and chest and abdominal movement amplitude, synchronous acquisition is used to ensure the correspondence of parameters over time and avoid data deviations due to time differences. During the standardization conversion process, different conversion methods are used for different parameters. For example, linear conversion is used for count-type parameters such as respiratory rate and heart rate, interval mapping is used for proportional parameters such as blood oxygen saturation, and normalization is used for physical quantity parameters such as chest and abdominal movement amplitude, ensuring that the converted values ​​accurately reflect the relative changes of each parameter.

[0029] The calculation of the effective respiratory rate must cover the entire monitoring period, with each sampling time determined independently and unaffected by results from other times. The calculation of the average interval duration must exclude significantly abnormal interval duration values. These abnormal values ​​are usually due to external interference or temporary sensor malfunctions, resulting in values ​​exceeding the normal respiratory interval range, to ensure the representativeness of the average. In the calculation of physiological state assessment values, the weighting coefficients must comprehensively consider the indicative role of each parameter in respiratory abnormalities. For example, blood oxygen saturation plays a significant role in reflecting respiratory function, and its corresponding weighting coefficient is relatively high.

[0030] The thresholds for abnormal physiological parameters are not fixed and are adjusted appropriately based on the gestational age, weight, and age of the target newborn. For premature and low-birth-weight infants, the threshold range is widened to accommodate their unique physiological conditions. Once an abnormal physiological parameter signal is generated, it serves as a trigger for subsequent modules, ensuring the system can promptly analyze any potential respiratory abnormalities. Throughout the acquisition process, all raw data and calculation results are stored in real time for subsequent traceability and verification, while also preventing data loss that could disrupt the continuity of the analysis.

[0031] Example 2: See Figure 3By simultaneously acquiring respiratory waveform signals and chest and abdominal motion signals of the target newborn during the current monitoring period, the respiratory waveform signals are obtained through an airflow sensor placed near the newborn's mouth and nose, while the chest and abdominal motion signals are acquired through a piezoelectric sensor attached to the chest and abdomen. The acquisition frequencies of both signals are kept consistent to ensure correspondence on the time axis, thus obtaining the respiratory waveform signals and chest and abdominal motion signals of the target newborn during the current monitoring period. Based on this, signal processing methods are used to amplify, filter, and digitize the original signals, generating respiratory waveform diagrams and chest and abdominal motion curves of the target newborn during the current monitoring period. The respiratory waveform diagram is plotted on the horizontal axis with time and airflow intensity on the vertical axis, while the chest and abdominal motion curve is plotted on the horizontal axis with time and motion amplitude on the vertical axis, thus obtaining the respiratory waveform diagrams and chest and abdominal motion curves of the target newborn during the current monitoring period.

[0032] Reference respiratory waveforms of normal newborns are extracted from the system database. These reference waveforms are derived from a large amount of monitoring data of healthy newborns, covering typical respiratory waveform characteristics of newborns of different gestational ages and weights. The respiratory waveforms of the target newborn during the current monitoring period are compared morphologically with those of the reference newborns. The comparison includes the rising slope, falling slope, peak shape, and trough depth of the waveforms. The morphological similarity of the respiratory waveforms of the target newborn during the current monitoring period is obtained by calculating the degree of overlap at each feature point. The morphological similarity is then graded based on the overlap range. For example, an overlap of 90% or more is grade one, 80%-90% is grade two, and so on. The grade value is recorded as the morphological similarity grade.

[0033] Obtain the waveform period parameter of the reference respiratory waveform of a normal newborn. This parameter is the average time interval between two adjacent peaks in the reference waveform, and is recorded as the reference period value. Extract the peak interval and trough interval from the respiratory waveform of the target newborn within the current monitoring period. The peak interval is the time difference between two adjacent peaks, and the trough interval is the time difference between two adjacent troughs, respectively recorded as peak interval and trough interval. Extract the reference peak interval and reference trough interval from the reference respiratory waveform of a normal newborn. The reference peak interval is the average time difference between two adjacent peaks in the reference waveform, and the reference trough interval is the average time difference between two adjacent troughs in the reference waveform, respectively recorded as reference peak interval and reference trough interval.

[0034] The respiratory pattern consistency score of the target newborn during the current monitoring period is obtained by comprehensively judging the similarity of waveform morphology, peak interval deviation, and trough interval deviation. Peak interval deviation is the absolute value of the difference between the peak interval of the target newborn and the reference peak interval, and trough interval deviation is the absolute value of the difference between the trough interval of the target newborn and the reference trough interval. By combining the three according to a preset ratio, the respiratory pattern consistency score of the target newborn during the current monitoring period is obtained.

[0035] Reference chest and abdominal movement curves of normal newborns are extracted from the system database. These curves record the synchronous changes in chest and abdominal movements under normal breathing conditions, including the proportional relationship of movement amplitude and phase difference. The chest and abdominal movement curves of the target newborn during the current monitoring period are compared and analyzed with those of the reference chest and abdominal movement curves of normal newborns. The comparison focuses on the relative trends of the chest and abdominal movement curves. When the trends deviate from the correspondence of the reference curves, a synchronization deviation is identified, thus determining the synchronization deviation of chest and abdominal movements in the target newborn during the current monitoring period. The corresponding synchronization deviation segments are extracted from the chest and abdominal movement curves of the target newborn during the current monitoring period. These deviation segments are continuous time periods where the synchronization deviation exceeds a preset range, and are considered as abnormal respiratory coordination segments in the target newborn during the current monitoring period. The number of abnormal respiratory coordination segments in the target newborn during the current monitoring period is counted, and their values ​​are recorded as the number of coordination abnormalities.

[0036] Extract the duration of each abnormal respiratory coordination segment from the chest and abdominal movement curve of the target newborn during the current monitoring period. The duration of each abnormal segment is the time difference between the start and end times of the segment. The duration of each abnormal respiratory coordination segment during the current monitoring period of the target newborn is obtained and its value is recorded as the abnormal duration.

[0037] The respiratory pattern assessment is calculated based on a comprehensive evaluation of the respiratory pattern consistency score, the number of coordination abnormalities, and the duration of these abnormalities. During the calculation, the number of coordination abnormalities and the duration of these abnormalities are converted into corresponding quantitative scores, which are then superimposed on the respiratory pattern consistency score according to a set weight to obtain the final respiratory pattern assessment result. Throughout the entire identification process, all extracted feature parameters and calculation results are stored in real time. At the end of each monitoring cycle, the comparison benchmarks for the respiratory waveform and chest and abdominal motion curves are automatically updated to adapt to the dynamic changes in the newborn's physiological state. For transient interferences occurring during signal acquisition, a time window is set for filtering, retaining only signal segments whose duration exceeds a minimum threshold for analysis, ensuring that the identification results reflect the true respiratory pattern characteristics of the target newborn.

[0038] Example 3: Apnea events are detected during the current monitoring period of the target newborn. Detection is based on changes in respiratory waveform signals and chest and abdominal movement signals. An apnea event is defined as a period where no valid breathing point is detected for multiple consecutive sampling times, and the duration exceeds a preset duration. Simultaneously, the decrease in blood oxygen saturation during the current monitoring period is recorded. The decrease in blood oxygen saturation is the difference between the blood oxygen saturation values ​​before and after the apnea event. The maximum decrease in blood oxygen saturation is extracted and recorded as the blood oxygen decrease value. The correlation between the duration of the apnea event and the blood oxygen decrease value is then analyzed. This correlation analysis is achieved by calculating the consistency of their changing trends. If the blood oxygen decrease value increases accordingly as the apnea duration lengthens, the correlation is strengthened, yielding the correlation values ​​for the current monitoring period of the target newborn.

[0039] The respiratory rate of change in the target newborn during the current monitoring period is calculated. This rate is the ratio of the difference in respiratory rate values ​​between two adjacent time points to the corresponding time interval. Simultaneously, the fluctuation range of respiratory rate is extracted from the data. This fluctuation range is the difference between the maximum and minimum respiratory rate values ​​within the monitoring period, thus obtaining the respiratory rate fluctuation value for the target newborn during the current monitoring period. Furthermore, the number of times the respiratory rate exceeds the normal range is extracted from the data. The normal range is pre-set based on the newborn's age and health condition, thus obtaining the number of abnormal respiratory rates during the current monitoring period.

[0040] The heart rate variability index of the target newborn during the current monitoring period is obtained by calculating the degree of variation in the interval between consecutive heartbeats, specifically the standard deviation of the time difference between two adjacent heartbeats.

[0041] The relevant values, respiratory rate fluctuation values, frequency of abnormal respiratory rates, and heart rate variability values ​​of the target newborn within the current monitoring period are extracted and processed using a multi-dimensional feature fusion algorithm to obtain an abnormal feature assessment value. In the multi-dimensional feature fusion algorithm, each value is first standardized to ensure they are of the same order of magnitude, and then fused using a weighted summation method to obtain the abnormal feature assessment value, as shown in the following formula: F = a × L + b × P + c × Q + d × R Where F represents the abnormal feature assessment value, L represents the correlation value, P represents the respiratory rate fluctuation value, Q represents the number of frequency abnormalities, R represents the heart rate variability value, and a, b, c, and d represent the weight coefficients corresponding to each value.

[0042] The respiratory abnormality level of the target newborns was assessed and analyzed. The specific analysis process is as follows: The physiological status assessment values ​​of the target newborn within the current monitoring period are retrieved. An influence coefficient is set for these physiological status assessment values, which are dynamically adjusted based on their magnitude; the higher the physiological status assessment value, the larger the influence coefficient. The converted physiological influence value is obtained by processing the values ​​according to the influence coefficient conversion rule, which is determined based on the product relationship between the physiological status assessment value and the influence coefficient.

[0043] The respiratory pattern assessment results, abnormal feature assessment values, and converted physiological impact values ​​of the target newborn within the current monitoring period are extracted and comprehensively normalized. This normalization process maps each value to a range of 0-1, preserving the relative magnitudes of the values. This yields the comprehensive respiratory abnormality assessment value.

[0044] A grading threshold is set for the comprehensive assessment value of respiratory abnormalities. This threshold includes a first grading threshold and a second grading threshold, with the first grading threshold being lower than the second grading threshold. The comprehensive assessment value of respiratory abnormalities is compared and analyzed against these grading thresholds. When the comprehensive assessment value of respiratory abnormalities is less than the first grading threshold, a low-risk warning signal is generated; when the comprehensive assessment value of respiratory abnormalities is greater than or equal to the first grading threshold but less than the second grading threshold, a medium-risk warning signal is generated; and when the comprehensive assessment value of respiratory abnormalities is greater than or equal to the second grading threshold, a high-risk warning signal is generated.

[0045] Throughout the process, data collection and processing were completed within a pre-set monitoring period. At the end of each monitoring period, the calculation benchmarks for each assessment indicator were automatically updated to adapt to the dynamic changes in the newborn's physiological state. Relationship values, respiratory rate fluctuation values, frequency of abnormal respiratory rates, and heart rate variability values ​​were all collected through continuous monitoring to ensure data integrity and continuity. In the comprehensive normalization process, different mapping functions were used for the characteristics of different parameters, ensuring that the normalized values ​​accurately reflected the actual degree of abnormality for each parameter. The grading thresholds were set with reference to the clinical diagnostic criteria for neonatal respiratory abnormalities, while also allowing for manual adjustment based on the specific usage environment and the individual circumstances of the target newborn to adapt to different monitoring needs.

[0046] Example 4: The selection process for early warning signal types is as follows: If a low-risk early warning signal is detected, a Level 1 early warning command is triggered. Based on the triggered Level 1 early warning command, the early warning method for the current monitoring period of the target newborn is selected, thus obtaining the triggerable sound warning. The sound warning uses a specific frequency sound wave signal, which is within the range of human hearing sensitivity and will not have a stimulating effect on the newborn. It is continuously output through a speaker device until the warning signal is lifted. If a medium-risk early warning signal is detected, a Level 2 early warning command is triggered. Based on the triggered Level 2 early warning command, the early warning method for the current monitoring period of the target newborn is selected, thus obtaining the triggerable sound and light combined early warning. The sound signal in the sound and light combined early warning differs from the sound signal in Level 1 early warning in frequency and interval. The light signal uses soft monochromatic light, continuously flashing through LEDs at a stable frequency to avoid interference with the surrounding environment due to excessively drastic light changes. If a high-risk early warning signal is detected, a Level 3 early warning command is triggered. Based on the triggered Level 3 early warning command, the early warning method and notification frequency for the current monitoring period of the target newborn are selected, thus obtaining the triggerable sound and light combined early warning and remote notification parameters. The combined sound and light warning in the Level 3 alert has a higher sound signal intensity than the Level 2 alert, and the light signal uses two different colors to flash alternately to enhance the warning effect. The remote notification parameters include the notification recipient, notification content, and sending interval. The notification recipient is the preset medical staff terminal, and the notification content includes the identification information of the target newborn and a summary of abnormal parameters in the current monitoring period. The sending interval is dynamically adjusted according to the duration of the high-risk warning signal. The initial interval is relatively short, and the interval length gradually increases as the warning duration extends, but it is always kept within the range that ensures that medical staff can respond in a timely manner.

[0047] The physiological parameter acquisition module also includes a data preprocessing submodule, used to preprocess the acquired raw physiological parameter information. The specific preprocessing process is as follows: The raw respiratory rate data acquired during the current monitoring period of the target newborn is smoothed and filtered using a sliding window technique. The respiratory rate data from multiple consecutive sampling points are averaged, with the window size determined based on the sampling frequency of the respiratory rate. This method removes high-frequency noise interference, resulting in filtered respiratory rate data. The blood oxygen saturation data acquired during the current monitoring period of the target newborn is then subjected to outlier detection. Outlier detection is based on a preset physiological range, set according to the normal fluctuation range of newborn blood oxygen saturation. When blood oxygen saturation data at a certain sampling time exceeds this range, it is identified as an outlier data point. After identifying and removing these outlier data points, the mean of adjacent normal data points is used to fill the gaps after removal, resulting in corrected blood oxygen saturation data. Baseline drift correction was performed on the chest and abdominal motion signals collected during the current monitoring period of the target newborns. Baseline drift correction involved extracting and separating the low-frequency trend components from the original signal. These low-frequency trend components were obtained by performing a long-period moving average on the original signal, eliminating baseline shift during long-term monitoring and resulting in corrected chest and abdominal motion signals. The filtered respiratory rate data, corrected blood oxygen saturation data, and corrected chest and abdominal motion signals were then time-synchronized and aligned using the system clock as a reference, with timestamps accurate to the millisecond level to ensure consistency of parameters over time, resulting in a preprocessed physiological parameter dataset. This preprocessed physiological parameter dataset will serve as the foundation for subsequent analyses in various modules. It is stored in chronological order of monitoring time, with each data point containing a corresponding time identifier and parameter type identifier for easy retrieval and recall. During preprocessing, for data loss due to temporary sensor malfunction, linear interpolation was used to supplement data within a preset duration if the loss was within the preset range; otherwise, it was marked as a data interruption and handled separately in subsequent analyses to avoid incomplete data affecting the analysis results.

[0048] Example 5: See Figure 4Historical physiological parameter data of the target newborn over the past 72 hours were extracted from the system storage unit. This data covers records from multiple continuous monitoring periods, including mean respiratory rate, mean blood oxygen saturation, and respiratory pattern characteristic parameters. The mean respiratory rate is the arithmetic mean of all valid respiratory rate data over the past 72 hours, the mean blood oxygen saturation is the arithmetic mean of all valid blood oxygen saturation data over the past 72 hours, and the respiratory pattern characteristic parameters include the morphological characteristics, periodic characteristics, and coordination characteristics of chest and abdominal movements of the respiratory waveform over the past 72 hours. Based on this data, a historical reference dataset was established, arranged in chronological order, with each time point corresponding to a complete set of parameter values.

[0049] The preprocessed physiological parameter dataset of the target newborn during the current monitoring period is compared item by item with the historical reference dataset, with each comparison performed separately for each parameter. For respiratory rate, the difference between the current respiratory rate value and the historical average respiratory rate is calculated, and then this difference is divided by the historical average respiratory rate to obtain the percentage deviation of respiratory rate. For blood oxygen saturation, the same calculation method is used to obtain the percentage deviation of blood oxygen saturation. For respiratory pattern characteristic parameters, appropriate comparison methods are used according to the characteristics of different parameters. For morphological features, the percentage deviation is calculated through morphological similarity; for periodic features, the percentage deviation is calculated through the ratio of the period difference to the historical average period; and for coordination features, the percentage deviation is calculated through the ratio of the difference in the proportion of abnormal segments to the historical proportion. In this way, the percentage deviation between the current value and the historical mean of each parameter is calculated.

[0050] The number of parameters whose percentage deviation exceeds a preset deviation threshold is recorded. The preset deviation threshold is set according to the physiological significance of different parameters. For example, the deviation threshold for respiratory rate is different from that for blood oxygen saturation. These are recorded as the number of abnormal parameters.

[0051] When the number of abnormal parameters is greater than or equal to the preset threshold, the analysis sensitivity of the abnormal feature extraction module is enhanced. Enhancement methods include narrowing the judgment range of abnormal features, increasing the sampling frequency of feature extraction, and lowering the trigger threshold of abnormal features. When the number of abnormal parameters is less than the preset threshold, the default analysis sensitivity of the abnormal feature extraction module is maintained. The default analysis sensitivity is the analysis standard initially set by the system, including the usual judgment range, sampling frequency, and trigger threshold.

[0052] The early warning level assessment module also includes a multi-parameter fusion analysis submodule, which is used to comprehensively assess abnormal early warnings based on multiple physiological parameters.

[0053] The warning level assessment module specifically involves determining the weight coefficients of the respiratory pattern assessment results, abnormal feature assessment values, and converted physiological impact values ​​in the comprehensive assessment using the analytic hierarchy process (AHP). The AHP constructs a judgment matrix to compare the importance of each parameter pairwise, and then obtains the weight coefficients through matrix operations. The respiratory pattern assessment results have the highest weight coefficient, followed by the abnormal feature assessment values, and the converted physiological impact values ​​have the lowest weight coefficient.

[0054] Each assessment value is multiplied by its corresponding weighting coefficient and then summed to obtain a preliminary fusion assessment value. A fuzzy logic algorithm is then used to nonlinearly correct this preliminary fusion assessment value. Based on preset fuzzy rules, the algorithm maps the preliminary fusion assessment value to a corresponding correction interval. During the correction process, the interrelationships between parameters are considered, eliminating dimensional differences and coupling interference between different parameters, resulting in a final comprehensive assessment value for respiratory abnormalities. This final comprehensive assessment value for respiratory abnormalities is compared with preset grading thresholds to determine the corresponding warning level. The comparison method for the grading thresholds is consistent with that in the respiratory abnormality level assessment and analysis; that is, by comparing the comprehensive assessment value with the first and second grading thresholds, a low-risk, medium-risk, or high-risk warning signal is generated.

[0055] Throughout the process, the historical data comparison module and the early warning level assessment module maintain data interaction. The results of the historical data comparison are fed back to the abnormal feature extraction module in real time, affecting its analysis process. The multi-parameter fusion analysis submodule of the early warning level assessment module continuously receives processing results from the respiratory pattern recognition module, the abnormal feature extraction module, and the physiological parameter acquisition module, ensuring that the comprehensive assessment can reflect the current overall physiological state of the target newborn. The weight coefficients and threshold settings of all parameters are stored in the system configuration file, which supports adjustments according to clinical needs. However, the adjustment process must follow the preset permission management mechanism to prevent unauthorized modifications.

[0056] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0057] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A neonatal respiratory abnormality early warning system, characterized in that, include: The physiological parameter acquisition module is used to collect physiological parameter information of the target newborn in real time, obtain the physiological parameter information of the target newborn, make a preliminary judgment on the physiological state of the target newborn, obtain abnormal physiological parameter signals, trigger analysis instructions based on the obtained abnormal physiological parameter signals, and execute the breathing pattern recognition module and abnormal feature extraction module based on the triggered analysis instructions. The breathing pattern recognition module is used to identify the breathing pattern characteristics of the target newborn, thereby analyzing the breathing pattern of the target newborn, obtaining the breathing pattern assessment result, and sending it to the early warning level assessment module. The abnormal feature extraction module is used to extract abnormal respiratory features of the target newborn, thereby analyzing the degree of abnormality of the target newborn, obtaining an abnormal feature assessment value, and sending it to the early warning level assessment module. The early warning level assessment module is used to receive the breathing pattern assessment results and abnormal feature assessment values, thereby assessing and analyzing the breathing abnormality level of the target newborn, obtaining low-risk and high-risk early warning signals, and sending them to the early warning signal generation module. The warning signal generation module is used to receive low-risk and high-risk warning signals, thereby selecting the warning signal type and obtaining the parameters for triggering sound warnings and sound and light warnings.

2. The neonatal respiratory abnormality early warning system according to claim 1, characterized in that, The physiological parameters of the target newborn are collected in real time. The specific collection process is as follows: By continuously collecting respiratory rate, blood oxygen saturation, heart rate, and chest and abdominal movement amplitude of the target newborn during the current monitoring period, the values ​​of respiratory rate, blood oxygen saturation, heart rate, and chest and abdominal movement amplitude of the target newborn during the current monitoring period are obtained. The values ​​of the four are extracted and standardized to obtain a comprehensive value of physiological parameters. By collecting respiratory waveform signals from the target newborn during the current monitoring period, the respiratory waveform signals at each sampling time within the current monitoring period are obtained and marked as waveform sampling values. A waveform amplitude threshold is set, and the waveform sampling values ​​are compared and analyzed with the waveform amplitude threshold. When the waveform sampling value is greater than or equal to the waveform amplitude threshold, the sampling time is determined as a valid breathing point. When the waveform sampling value is less than the waveform amplitude threshold, the sampling time is determined as an invalid breathing point. The number of times the target newborn is determined as a valid breathing point within the current monitoring period is counted, and the percentage is calculated by comparing it with the total number of samplings to obtain the percentage of valid breathings. Simultaneously, the respiratory interval time of the target newborn during the current monitoring period is collected to obtain the interval time of each breath during the current monitoring period of the target newborn, and these intervals are labeled as interval duration values. The mean of each interval duration value during the current monitoring period of the target newborn is calculated to obtain the mean interval duration during the current monitoring period of the target newborn. The effective breathing ratio, the mean interval duration, and the comprehensive value of physiological parameters during the current monitoring period of the target newborn are extracted, multiplied by the corresponding weighting coefficients, and then added together to obtain the physiological status assessment value. The physiological state assessment value is extracted and its range is determined. The physiological state assessment value is compared and analyzed with the preset physiological parameter abnormality threshold. When the physiological state assessment value is greater than or equal to the preset physiological parameter abnormality threshold, a physiological parameter abnormality signal is generated.

3. The neonatal respiratory abnormality early warning system according to claim 1, characterized in that, The respiratory pattern characteristics of the target newborn are identified, and the specific identification process is as follows: By synchronously acquiring the respiratory waveform signal and chest and abdominal movement signal of the target newborn during the current monitoring period, the respiratory waveform signal and chest and abdominal movement signal of the target newborn during the current monitoring period are obtained. Based on this, the respiratory waveform diagram and chest and abdominal movement curve diagram of the target newborn during the current monitoring period are generated through signal processing methods. The system database extracts reference respiratory waveforms of normal newborns. The respiratory waveforms of the target newborn during the current monitoring period are compared with those of the reference respiratory waveforms of normal newborns to obtain the morphological similarity of the respiratory waveforms of the target newborn during the current monitoring period. The waveforms are then classified into different levels, and the level value is recorded as the morphological similarity level. Obtain the waveform period parameters of the reference respiratory waveform of a normal newborn and record them as the reference period values; The peak interval and trough interval are extracted from the respiratory waveform of the target newborn during the current monitoring period and are denoted as peak interval and trough interval, respectively. Reference peak interval and reference trough interval are extracted from the reference respiratory waveform of normal newborns and are denoted as reference peak interval and reference trough interval, respectively. Based on a comprehensive assessment of waveform morphology similarity, peak interval deviation, and trough interval deviation, a respiratory pattern consistency score for the target newborn during the current monitoring period is obtained. Reference chest and abdominal movement curves of normal newborns are extracted from the system database. The chest and abdominal movement curves of the target newborn during the current monitoring period are compared and analyzed with the reference chest and abdominal movement curves of normal newborns to obtain the synchronicity deviation of chest and abdominal movement in the target newborn during the current monitoring period. The corresponding synchronicity deviation segments are extracted from the chest and abdominal movement curves of the target newborn during the current monitoring period as the respiratory coordination abnormal segments in the current monitoring period of the target newborn. The number of respiratory coordination abnormal segments in the current monitoring period of the target newborn is counted and the value is recorded as the number of coordination abnormalities. Extract the duration of each abnormal respiratory coordination segment in the current monitoring period of the target newborn from the chest and abdominal motion curve graph of the target newborn in the current monitoring period, obtain the duration of each abnormal respiratory coordination segment in the current monitoring period of the target newborn, and take its value as the abnormal duration. The respiratory pattern assessment results are obtained by comprehensively calculating the respiratory pattern consistency score, the number of coordination abnormalities, and the duration of abnormalities.

4. A neonatal respiratory abnormality early warning system according to claim 1, characterized in that, The abnormal respiratory features of the target newborn are extracted. The specific extraction process is as follows: By detecting apnea events in the target newborn during the current monitoring period and recording the decrease in blood oxygen saturation during the current monitoring period, the maximum decrease in blood oxygen saturation is extracted and recorded as the blood oxygen decrease value. Then, the correlation between the duration of apnea events and the blood oxygen decrease value is analyzed to obtain the correlation value of the target newborn during the current monitoring period. By calculating the rate of change of respiratory rate of the target newborn during the current monitoring period, and extracting the fluctuation range of respiratory rate from the data of respiratory rate change of the target newborn during the current monitoring period, the fluctuation value of respiratory rate of the target newborn during the current monitoring period is obtained. At the same time, the number of times the respiratory rate exceeds the normal range is extracted from the data of respiratory rate change of the target newborn during the current monitoring period, thus obtaining the number of frequency abnormalities of the target newborn during the current monitoring period. By acquiring the heart rate variability index of the target newborn during the current monitoring period, the heart rate variability value of the target newborn during the current monitoring period is obtained. The relevant values, respiratory rate fluctuation values, frequency abnormality values, and heart rate variability values ​​of the target newborn during the current monitoring period are extracted and processed according to a multi-dimensional feature fusion algorithm to obtain abnormal feature evaluation values.

5. A neonatal respiratory abnormality early warning system according to claim 1, characterized in that, The respiratory abnormality level of the target newborns was assessed and analyzed. The specific analysis process is as follows: Retrieve the physiological status assessment values ​​of the target newborn during the current monitoring period, set the influence coefficient of the physiological status assessment values ​​of the target newborn during the current monitoring period, and process them according to the influence coefficient conversion rules to obtain the converted physiological influence value; The respiratory pattern assessment results, abnormal feature assessment values, and converted physiological impact values ​​of the target newborn during the current monitoring period are extracted, and then normalized to obtain the comprehensive assessment value of respiratory abnormality. Set a grading threshold for the comprehensive assessment value of respiratory abnormalities. Compare and analyze the comprehensive assessment value of respiratory abnormalities with the grading threshold. When the comprehensive assessment value of respiratory abnormalities is less than the first grading threshold, a low-risk warning signal is generated. When the comprehensive assessment value of respiratory abnormalities is greater than or equal to the first grading threshold and less than the second grading threshold, a medium-risk warning signal is generated. When the comprehensive assessment value of respiratory abnormalities is greater than or equal to the second grading threshold, a high-risk warning signal is generated.

6. A neonatal respiratory abnormality early warning system according to claim 1, characterized in that, The specific selection process for the warning signal type is as follows: If a low-risk warning signal is detected, a Level 1 warning instruction is triggered. Based on the triggered Level 1 warning instruction, the warning method for the current monitoring period of the target newborn is selected, thereby determining the sound warning to be triggered. If a medium-risk warning signal is detected, a level-two warning instruction is triggered. Based on the triggered level-two warning instruction, the warning method for the target newborn within the current monitoring period is selected, thereby determining the appropriate sound and light combination warning to be triggered. If a high-risk warning signal is detected, a Level 3 warning instruction is triggered. Based on the triggered Level 3 warning instruction, the warning method and notification frequency for the target newborn within the current monitoring period are selected, thereby obtaining the parameters for triggering the combined sound and light warning and remote notification.

7. A neonatal respiratory abnormality early warning system according to claim 1, characterized in that, The physiological parameter acquisition module also includes a data preprocessing submodule, which is used to preprocess the acquired raw physiological parameter information. The specific preprocessing process is as follows: By smoothing and filtering the raw respiratory rate data collected during the current monitoring period of the target newborn, high-frequency noise interference is removed, and filtered respiratory rate data is obtained. By detecting outliers in the blood oxygen saturation data collected during the current monitoring period of the target newborn, abnormal data points that exceed the physiological range are identified and removed, and corrected blood oxygen saturation data is obtained. By performing baseline drift correction on the chest and abdominal motion signals collected during the current monitoring period of the target newborn, the signal baseline shift during long-term monitoring is eliminated, and the corrected chest and abdominal motion signals are obtained. The filtered respiratory rate data, corrected blood oxygen saturation data, and corrected chest and abdominal motion signals are time-synchronized and aligned to ensure the consistency of each parameter in the time dimension, resulting in a preprocessed physiological parameter dataset.

8. A neonatal respiratory abnormality early warning system according to claim 1, characterized in that, It also includes a historical data comparison module, which is used to compare and analyze current physiological parameter information with historical data. The specific comparison process is as follows: Historical physiological parameter data of the target newborn over the past 72 hours were extracted from the system storage unit, including mean respiratory rate, mean blood oxygen saturation, and respiratory pattern characteristic parameters, to establish a historical reference dataset; The preprocessed physiological parameter dataset of the target newborn during the current monitoring period is compared item by item with the historical reference dataset, and the percentage deviation between the current value and the historical mean of each parameter is calculated. The number of parameter items whose percentage of statistical deviation exceeds the preset deviation threshold is recorded as the number of abnormal parameters; When the number of abnormal parameters is greater than or equal to the preset threshold, the analysis sensitivity of the abnormal feature extraction module is enhanced; when the number of abnormal parameters is less than the preset threshold, the default analysis sensitivity of the abnormal feature extraction module is maintained.

9. A neonatal respiratory abnormality early warning system according to claim 1, characterized in that, The early warning level assessment module also includes a multi-parameter fusion analysis submodule, which is used to comprehensively assess abnormal early warnings by integrating multiple physiological parameters.

10. A neonatal respiratory abnormality early warning system according to claim 9, characterized in that, The warning level assessment module is specifically as follows: The weight coefficients of respiratory pattern assessment results, abnormal feature assessment values, and converted physiological impact values ​​in the comprehensive assessment were determined by the analytic hierarchy process. Among them, the weight coefficient of respiratory pattern assessment results was the highest, followed by abnormal feature assessment values, and the weight coefficient of converted physiological impact values ​​was the lowest. Each evaluation value is multiplied by its corresponding weighting coefficient and then summed to obtain a preliminary integrated evaluation value. The preliminary fusion evaluation value is nonlinearly corrected by fuzzy logic algorithm to eliminate the dimensional differences and coupling interference between different parameters, and the final comprehensive evaluation value of respiratory abnormality is obtained. The final comprehensive assessment value of respiratory abnormalities is compared with the preset grading threshold to determine the corresponding warning level.

Images