Breast feeding evaluation index construction method and system

Abstract

The invention relates to the technical field of breast feeding, in particular to a breast feeding evaluation index construction method and system.The breast feeding evaluation index construction method comprises the following steps that feeding time and frequency are obtained, sucking and swallowing signal sequences are collected, a sucking continuous segment and a swallowing stable stage are recognized, a rhythm continuation section is extracted, and the fluctuation amplitude and time proportion is calculated; the method comprises the following steps: collecting time sequence signals of sucking and swallowing, constructing a continuous behavior data stream, extracting an overlapping region of a high-frequency sucking point and a swallowing concentrated section, measuring and calculating matching intensity and interval fluctuation, screening stable slowly-changing fragments, counting behavior types and frequencies, and generating a feeding evaluation result. According to the method, the matching intensity and the rhythm fluctuation level are quantified, stable and slowly-changing behavior characteristics are identified, multi-dimensional evaluation of milk supply, feeding rhythm and matching performance is achieved in combination with behavior classification and frequency statistics, and the accuracy of breast feeding state judgment and the data supporting capacity are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of breastfeeding, and particularly to a method and system for constructing breastfeeding evaluation indicators. Background Art

[0002] The technical field of breastfeeding includes related technologies for evaluating, guiding, and intervening in the entire process of a lactating mother providing nutrition to an infant. The core content of this technical field lies in collecting, recording, and analyzing quantitative and qualitative information in processes such as maternal milk secretion, infant sucking behavior, and physiological responses, to determine whether the feeding behavior meets the growth needs of the infant and to guide the mother to adjust the feeding method. Overall, the technical field of breastfeeding covers technical directions such as milk production monitoring, infant milk sucking behavior analysis, nipple posture detection, physiological signal recognition, physical sign change recording, and feeding cycle parameter management, and often relies on sensor detection, physiological data collection, behavioral quantification, and data comparison means to construct a scientific evaluation system.

[0003] Among them, the method for constructing breastfeeding evaluation indicators refers to a quantitative analysis system for evaluating feeding quality set for three aspects involved in breastfeeding behavior: feeding sufficiency, physiological state matching, and behavioral performance rationality. This patent theme covers forming an evaluation indicator combination based on specific physiological parameters such as maternal milk production, infant milk sucking frequency, milk sucking duration per unit time, infant weight change rate, and comparison of milk intake and excretion data, and completing the quantification of the indicators by collecting specific measurement values during the mother-infant interaction process and establishing an indicator scoring method. It uses means such as setting standard intervals, determining scoring weights, and defining threshold levels, and combines statistical distribution laws to construct the indicators, and finally forms an operable evaluation standard system.

[0004] In the prior art, when dealing with the breastfeeding process, it often focuses on static physiological data analysis, such as milk volume and weight changes, and fails to delve into the dynamic process of feeding behavior. The lack of real-time monitoring and detailed analysis of the rhythm of sucking and swallowing behaviors makes it impossible to effectively identify and adjust possible coordination problems and abnormal rhythms that infants may have at different feeding stages. In addition, the prior art does not take the continuity of feeding behavior and the mutual influence between behaviors as the focus of analysis, restricting the comprehensive evaluation of the quality of mother-infant interaction and weakening the timeliness and pertinence of intervention measures. The lack of dynamic and continuous behavioral data analysis may lead to the evaluation results being unable to fully reflect the feeding effect, affecting the scientificity and practicality of breastfeeding guidance. Summary of the Invention

[0005] The purpose of the present invention is to solve the deficiencies existing in the prior art, and to propose a method and system for constructing breastfeeding evaluation indicators.

[0006] To achieve the above object, the present invention adopts the following technical solutions: A method for constructing breastfeeding evaluation indicators, comprising the following steps:

[0007] S1: Obtain the feeding time and feeding frequency during breastfeeding. Through a vibration sensor deployed on the baby's chest, collect the number of sucking and swallowing times of the baby, integrate the data in chronological order, and obtain feeding record data;

[0008] S2: Based on the feeding record data, extract continuous segments of sucking signals and stable swallowing phases, identify rhythm continuation segments, record the start and end times, calculate the sucking fluctuation amplitude, and count the proportion of time periods to obtain feeding rhythm characteristic segments;

[0009] S3: Based on the feeding rhythm characteristic segments, extract high-frequency sucking time points and compare them with the concentrated time period of swallowing actions, identify overlapping segments and extract synchronous time intervals, calculate the matching density of sucking and swallowing actions within the synchronous segment, and measure the interval fluctuation to obtain the sucking and swallowing coordination performance;

[0010] S4: Based on the sucking and swallowing coordination performance, retrieve stable feeding segments, screen slow-changing segments, record the start and end times and frequencies, classify and count according to feeding behavior types to obtain the milk supply situation;

[0011] S5: Based on the milk supply situation, sucking and swallowing coordination performance, and feeding rhythm characteristic segments, classify the baby's behavior types, count the time and frequencies, integrate the summary of feeding behaviors, and generate a breastfeeding evaluation result.

[0012] As a further solution of the present invention, the feeding record data includes feeding time values, feeding frequency values, sucking times values, and swallowing times values. The feeding rhythm characteristic segments are specifically sucking continuous interval values, swallowing stable time period values, sucking fluctuation amplitude values, and rhythm time period proportion values. The sucking and swallowing coordination performance includes synchronous time intervals, action matching density values, and rhythm interval fluctuation values. The milk supply situation is specifically stable feeding frequency values, slow-changing segment quantity values, and behavior type classification results. The breastfeeding evaluation result specifically refers to the baby's behavior type classification result, behavior duration values, behavior occurrence frequencies, and behavior summary content.

[0013] As a further solution of the present invention, the specific steps of S1 are:

[0014] S101: Obtain the sucking and swallowing action signals collected by the vibration sensor on the baby's chest, record the corresponding time points, pair the actions with the time and integrate them in order into a time series to generate a sucking and swallowing action sequence;

[0015] S102: Based on the sucking and swallowing action sequence, identifying the corresponding time of the first sucking signal and the last swallowing signal, constructing the feeding start and end time interval, and marking the action timestamps during the period to obtain the feeding cycle time interval;

[0016] S103: Based on the sucking and swallowing action sequence and the feeding cycle time interval, filter and sort the events in the interval, count the number of events and the interval, generate a feeding frequency value and a feeding time value, and obtain feeding record data.

[0017] As a further solution of the present invention, the specific steps of S2 are:

[0018] S201: Based on the feeding record data, extract segments with continuous signals and intervals not exceeding the sucking interval value, filter out time segments that meet the conditions, and obtain sucking continuous segment interval values;

[0019] S202: Based on the time range of the sucking continuous segment interval value, extract the corresponding swallowing signal sequence, filter out the time period with stable amplitude and gentle slope, retain the signal segment with consistent rhythm, and generate the swallowing signal stable time period value;

[0020] S203: Based on the sucking continuous segment interval value and the swallowing signal stable time period value, extract the time intersection segment, call the amplitude sequence of the sucking segment in the intersection, calculate the fluctuation amplitude of the sucking action in the segment, and simultaneously calculate the ratio of the total duration of the intersection segment to the total duration of the sucking segment to obtain the feeding rhythm characteristic segment.

[0021] As a further solution of the present invention, the calculation formula of the fluctuation amplitude of the sucking action within the segment is specifically:

[0022]

[0023] Among them, W v Represents the fluctuation amplitude of the sucking action within the segment, V si Represents the i-th sucking signal amplitude sampling point, V s Represents the average value of the sucking signal amplitude sequence, max(V s ) represents the maximum value of the sucking amplitude sequence, min(V s ) represents the minimum value of the sucking amplitude sequence, n represents the total number of sucking signal samples in this segment, δ t Represents the ratio of the variance of the time intervals between adjacent sampling points to the average interval.

[0024] As a further solution of the present invention, the specific steps of S3 are:

[0025] S301: Based on the feeding rhythm feature segments, extract the sucking trajectory data within the corresponding time period, locate the frequently fluctuating time points in each sucking signal, screen the time segments with concentrated peak distributions as the high-frequency point sets of behaviors, and generate the high-frequency time segments of sucking.

[0026] S302: Based on the high-frequency time segments of sucking, match the vibration data in the corresponding time period in the swallowing record, extract the vibration-intensive intervals and identify the time ranges overlapping with the high-frequency time of sucking, mark the distribution positions and total occurrence times of the overlapping intervals in complete cycles, and generate the sucking-swallowing synchronization segments.

[0027] S303: Based on the sucking-swallowing synchronization segments, calculate the intensity of behaviors within the synchronization segments, measure the variation differences in the time intervals between adjacent synchronization segments, and integrate the behavioral coordination feature indicators in combination with the distribution frequency statistical values to obtain the coordination performance of sucking and swallowing.

[0028] As a further solution of the present invention, the specific calculation formula for the intensity of behaviors within the synchronization segments is as follows:

[0029]

[0030] where D i represents the intensity of behaviors within the i-th synchronization segment, n j represents the number of times the j-th behavior appears within the i-th segment, M represents the total number of all behaviors within the segment, T i represents the duration of the i-th synchronization segment, T avg represents the average value of the durations of all synchronization segments, and σ represents the standard deviation.

[0031] As a further solution of the present invention, the specific steps of S4 are as follows:

[0032] S401: Based on the coordination performance of sucking and swallowing, obtain the time series of sucking and swallowing signals in the feeding record data, extract the time intervals and action amplitude differences between adjacent actions, and screen the continuous paragraphs with rhythm fluctuation amplitudes conforming to the stability of sucking and swallowing rhythms to obtain the stable rhythm time period intervals.

[0033] S402: Based on the signal fluctuation sequence corresponding to the stable rhythm time period intervals, compare the amplitude changes at adjacent time points, retain the continuous segments with stable fluctuation amplitudes, and extract the time periods that meet the conditions to obtain the slow-changing action segment time zones.

[0034] S403: Based on the feeding cycle numbers to which the slow-changing action segment time zones belong, count the occurrence frequencies in the cycles, extract the signal amplitude values and rhythm interval values of the segments, classify them into sucking and swallowing behavior types accordingly, and accumulate the segment quantities by behavior groups to obtain the milk supply situation.

[0035] As a further solution of the present invention, the specific steps of S5 are as follows;

[0036] S501: Based on the milk supply situation, combined with the sucking and swallowing coordination performance and the feeding rhythm characteristic segments, extract the time segments, action intensity and flow rate types in the paragraphs, split the behavior data according to dimensions, and compare each item with the class performance characteristics in the behavior standard description to generate a behavior comparison mapping table;

[0037] S502: Based on the behavior comparison mapping table, extract the corresponding time segments of each type of standard behavior in the whole process, count the start and end intervals, occurrence frequencies and interval distribution characteristics of the behavior distribution in the recording period, construct the performance classification of multiple types of behaviors in the cycle, and generate a behavior performance distribution structure;

[0038] S503: Based on the behavior performance distribution structure, splice the time periods of multiple types of behaviors in sequence to form a time map, integrally describe the behaviors in different time periods according to the performance attribution, and summarize them into a complete behavior set within a one-time cycle to generate a breastfeeding evaluation result.

[0039] A construction system for breastfeeding evaluation indicators, comprising:

[0040] The feeding acquisition module installs a vibration sensor on the baby's chest to monitor the sucking and swallowing action signals in real time, associates the data with the feeding start and end times, performs unified time axis processing on the two action channels, and integrally generates a structured output including time sequence, amplitude change and frequency to obtain feeding record data;

[0041] The rhythm extraction module, based on the feeding record data, extracts the action segments in the sucking signal that are continuous and have an interval time meeting the threshold, records the start and end times and extracts the amplitude sequence, locates the segments in the swallowing signal with stability meeting the standard, performs intersection processing, calculates the fluctuation amplitude of the sucking action, and statistically calculates the duration ratio of the intersection segment in the sucking segment to obtain the feeding rhythm characteristic segment;

[0042] The coordination analysis module, based on the rhythm characteristic segment, locates the time nodes with frequent amplitude changes in the sucking signal, selects the time intersection with the significant change interval in the swallowing signal, calculates the ratio of the number of sucking times to the swallowing duration in the synchronous segment to generate the synchronous density, and analyzes the time interval fluctuation of adjacent synchronous segments to obtain the sucking and swallowing coordination performance;

[0043] The behavior classification module, based on the sucking and swallowing coordination performance, screens the slow-changing action segments with gentle amplitude and slope meeting the standard, extracts the average value of the sucking or swallowing signal amplitude and the rhythm interval value, classifies them according to the standard, and records the occurrence frequency and proportion of the behavior in each cycle to obtain the milk supply situation;

[0044] Based on the milk supply situation, the sucking and swallowing coordination performance, and the rhythm feature segments, the index generation module counts the duration and frequency of eligible segments according to the standard, arranges the behavior distribution within the cycle, and generates the breastfeeding evaluation result for a single cycle.

[0045] Compared with the prior art, the advantages and positive effects of the present invention are as follows:

[0046] In the present invention, by collecting the time series signals of sucking and swallowing, constructing a continuous behavior data stream, extracting rhythm segments and synchronous time intervals, quantifying the cooperation density and rhythm fluctuation level, identifying stable and slow-changing behavior characteristics, and combining behavior classification and frequency statistics, multi-dimensional evaluation of milk supply, feeding rhythm, and cooperation performance is realized, improving the accuracy of breastfeeding status judgment and the data support ability. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following will briefly introduce the drawings required for the description of the embodiments. Obviously, the following drawings are only some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0048] Figure 1 It is a schematic flow chart of the steps of the present invention;

[0049] Figure 2 It is a system module diagram of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0050] The following will describe the technical solutions in the present invention with reference to the drawings.

[0051] In the embodiments of the present invention, words such as "exemplarily" and "for example" are used to represent examples, illustrations, or explanations. Any embodiment or design solution described as "example" in the present invention should not be construed as being more preferred or having more advantages than other embodiments or design solutions. Exactly speaking, the use of the word "example" aims to present concepts in a specific way. In addition, in the embodiments of the present invention, the meaning expressed by "and / or" can be both, or either one of the two.

[0052] In the embodiments of the present invention, "image" and "picture" can sometimes be used interchangeably. It should be noted that when the difference is not emphasized, the meanings they express are the same. "(of)", "corresponding", and "corresponding" can sometimes be used interchangeably. It should be noted that when the difference is not emphasized, the meanings they express are the same.

[0053] In the embodiments of the present invention, sometimes a subscript such as W1 may be written in a non-subscript form such as W1. When the difference is not emphasized, the meanings they express are the same.

[0054] To make the technical problems, technical solutions, and advantages to be solved by the present invention clearer, the following will be described in detail with reference to the accompanying drawings and specific embodiments.

[0055] The embodiments of the present invention provide a method for constructing breastfeeding evaluation indicators, including the following steps:

[0056] S1: Obtain the feeding time and feeding frequency during breastfeeding. Through a vibration sensor deployed on the baby's chest, continuously collect the sucking and swallowing action signals generated by the baby in the feeding state, synchronously record the start and end times of feeding, construct a data sequence with time correspondence, integrate the data in chronological order to obtain feeding record data.

[0057] S2: Based on the feeding record data, extract the continuous segments of the sucking signal and the stable stage of the swallowing action shown by the baby during feeding, then match the time periods with concentrated vibrations in the swallowing data, identify the overlapping time period intersections, mark the start time and end time, calculate the fluctuation amplitude of the sucking action within the section, and count the proportion of the rhythm section to obtain the feeding rhythm characteristic segment.

[0058] S3: Based on the feeding rhythm characteristic segment, locate the high-frequency time points of the sucking behavior within the corresponding time period, then match the time periods with concentrated vibrations in the swallowing data, identify the overlapping time period intersections, extract the synchronous time period range and sort out the coverage frequency of the synchronous behavior in the whole cycle, calculate the matching density of the sucking and swallowing actions within the synchronous section, and measure the time interval fluctuation range between consecutive cooperation segments to obtain the sucking and swallowing cooperation performance.

[0059] S4: Based on the sucking and swallowing cooperation performance, retrieve the complete feeding record data, locate the stable feeding segments with stable rhythm and uniform action intensity during feeding, screen out the slow-varying segments with stable signal fluctuations and good action continuity, extract the corresponding start and end times and the frequencies of occurrence in the feeding cycle, classify the behavior types of the slow-varying segments, and conduct grouped statistics to obtain the milk supply situation.

[0060] S5: Based on the milk supply situation, combined with the analysis of the sucking and swallowing cooperation performance and the feeding rhythm characteristic segment, respectively compare with the descriptions of the behavior standards, summarize the corresponding types of the behavior patterns presented in the content within the scope of the standard performance, sort out the time range and repetition times of each type of behavior in the whole process, and integrate them into the behavior summary during a single feeding to generate the breastfeeding evaluation result.

[0061] The feeding record data includes feeding time value, feeding frequency value, number of sucking times value and number of swallowing times value. The feeding rhythm feature segments are specifically the continuous sucking interval value, stable swallowing time period value, sucking amplitude fluctuation value and rhythm period occupancy ratio value. The sucking and swallowing coordination performance includes synchronous time interval, action matching density value and rhythm interval fluctuation value. The milk supply situation is specifically the stable feeding frequency value, number of slow-varying segments value and behavior type classification result. The breastfeeding evaluation result specifically refers to the infant behavior type classification result, behavior duration value, behavior occurrence frequency value and behavior summary content

[0062] The specific steps of S1 are as follows:

[0063] S101: Obtain the sucking and swallowing action signals collected by the infant chest vibration sensor, record the corresponding time points, pair the actions with the time and integrate them in order into a time series to generate a sucking and swallowing action sequence;

[0064] First, determine the sampling area range of the sensor according to the sensor installation position, and set the sampling frequency value in combination with the frequency response parameters of the sensor. For example, use a vibration sensor with a sampling frequency of 1000Hz to continuously record the weak vibration data on the surface of the infant's chest to generate an original vibration signal stream. Then, according to the set timestamp synchronization mechanism, attach a corresponding millisecond-level time tag to each signal sample to form a list of signal samples with time sequence marks. Next, convert the original signal into a dual-channel signal data table with a unified structure through a data format conversion program. One channel corresponds to the vibration characteristics excited by the sucking action, and the other channel corresponds to the waveform of the swallowing action signal. Subsequently, sequentially call the data sequences of the sucking signal channel and the swallowing signal channel from the dual-channel data. In the sucking signal channel, extract the section with frequent amplitude fluctuations and a period interval close to the sucking average rhythm reference value. In the swallowing signal channel, extract the high-amplitude section with concentrated frequency and continuous duration. Judge the central time points of the sucking and swallowing feature segments, and label them as sucking points and swallowing points respectively in combination with the time tags. Then, according to the arrangement order of the time tags, recombine the sucking points and swallowing points in the order of sampling time from early to late. Each action event corresponds to a specific occurrence time point to form a time-action pair list. If an error exceeding the set maximum allowable interval threshold (for example, 200 milliseconds) is detected between two time points, it will not be included in the sequence. After that, organize the sucking points and swallowing point pairs that meet the time sequence continuity into a time series file in a standard format, such as storing it in CSV format. Each row record is in the form of a "timestamp + action type" combination, such as "12:03:04.123, sucking" or "12:03:06.457, swallowing", and establish a retrieval table of the action order through an indexing method to realize the structured tracking and analysis of subsequent action changes starting from any time point, and finally generate a sucking and swallowing action sequence.

[0065] S102: Based on the sucking and swallowing action sequence, identify the corresponding times of the first sucking signal and the last swallowing signal, construct the feeding start and end time intervals, and mark the action timestamps during this period to obtain the feeding cycle time interval;

[0066] First, read the record content line by line from the structured action sequence, and classify and extract "sucking" and "swallowing" according to the action type. For the sucking action, compare the timestamp fields item by item, and select the earliest occurring item as the first sucking time point. For example, if the first sucking action time in the record is "12:03:04.123", then this time is used as the feeding start time. Then, perform a traversal operation on all action records marked as "swallowing", compare the timestamps of each record, and extract the record with the largest time value as the last swallowing time point. For example, if the swallowing time at the end of the record is "12:18:32.456", then this time is set as the feeding end time. Next, construct the feeding start and end time intervals, that is, the start time is the first sucking time, and the end time is the last swallowing time, forming the time interval [12:03:04.123, 12:18:32.456]. Subsequently, judge the timestamps of all records in the action sequence. Any action record that falls within this time interval is marked as an action within the feeding cycle, and a time index list is reconstructed in the order of the timestamps from earliest to latest. If there is an action time interval in the record that exceeds the feeding interruption determination threshold (for example, 600 seconds), it is not included in this cycle interval to ensure the accuracy of the feeding continuity determination. For example, in a certain set of data, the sucking points are "12:04:11.005" and "12:05:23.019", and the swallowing points are "12:04:45.701" and "12:06:01.980". Through time series regularization and time comparison, the complete start and end action times and action process data are identified and combined, and finally the feeding cycle time interval is obtained.

[0067] S103: Based on the sucking and swallowing action sequence and the feeding cycle time interval, screen the events within the interval and sort them, count the number and intervals of the events, generate the feeding frequency value and the feeding time value, and obtain the feeding record data;

[0068] First, load the generated action sequence into memory in chronological order, read the timestamps of each record item by item, convert the time values to the standard time format, call the start and end times of the feeding cycle time interval, and perform a judgment operation on each record. If the action time value is not less than the start time and not greater than the end time, retain the action record; otherwise, discard it. For example, when the start time of the feeding cycle is "08:15:00.000" and the end time is "08:45:00.000", only retain the records of sucking or swallowing events that occur within this time period. Then, re-sort all the retained action records in ascending order of timestamps to construct the action stream within the complete interval, and further count the total number of action events in this stream. For example, if 80 sucking events and 25 swallowing events are identified within the interval, the total number of events is 105. Then, call the timestamp information to calculate the time difference between adjacent actions to form a set of event interval time. If there is a continuous interval in the set that exceeds the feeding rhythm break judgment threshold (such as 20 seconds), the corresponding section will be regarded as a rhythm-discontinuous section and recorded separately. This break threshold is set based on clinical experience as the longest time period during which the infant continuously loses sucking or swallowing actions. Subsequently, calculate the average frequency of action events by dividing the total number of events by the duration of the feeding cycle. The duration is the difference between the end time and the start time, and the unit is expressed in seconds or minutes. For example, if the total number of events is 105 and the total cycle time is 30 minutes, the feeding frequency value is 105÷30 = 3.5 times / minute. At the same time, record the duration itself as the feeding time value. Finally, summarize the event quantity, interval distribution, frequency value, and feeding duration to construct a complete record structure, and finally obtain the feeding record data.

[0069] The specific steps of S2 are as follows:

[0070] S201: Based on the feeding record data, extract the segments with continuous signals and an interval time not exceeding the sucking interval value, screen the eligible time sections, and obtain the sucking continuous segment interval value;

[0071] First, call the sucking signal channel data in the record, read each signal data and its corresponding timestamp information in chronological order, and use the time difference between two adjacent signals as the basis for calculating the sucking interval. For example, in a certain sampling data, the timestamps are "10:00:00.100", "10:00:00.250", "10:00:00.400" in sequence, and the corresponding time differences are 150ms and 150ms. After obtaining all the time differences, set the sucking interval threshold, which is set according to the average sucking rhythm of the baby. If it is set to 200ms through clinical observation, then neither of the two intervals in the above sequence exceeds this value, and the signal is determined to be continuous. Then, further detect the signal amplitude. If the amplitude at each moment remains above the reference value of the sucking action response amplitude. For example, if the reference value is set to 0.05g and all amplitudes are above 0.06g, then this continuous time period meets the sucking action characteristics, and record the start and end times of this period as the candidate section. Then continue to re-detect the next time period starting from the next signal record. If the time difference between two adjacent signals is greater than the sucking interval threshold, it is regarded as the end of a sucking segment and a new judgment process starts. Traverse the entire data to screen out all signal sections that meet the interval requirements to form a preliminary section set. Then judge whether the length of each section meets the minimum duration requirement. For example, the minimum sucking section length is set to 1 second. If the length of a certain section is only 0.6 seconds, it is discarded. Finally, retain all the segments whose time differences meet the sucking interval threshold, whose amplitudes are stable, and whose section durations meet the reference duration, and record the start and end time points of each section. For example, "10:00:00.100~10:00:04.300", "10:00:05.500~10:00:09.000", etc. Finally, obtain the sucking continuous segment interval value.

[0072] S202: Based on the time range in the sucking continuous segment interval value, extract the corresponding swallowing signal sequence, screen out the time periods with stable amplitude and gentle slope, retain the signal segments with consistent rhythm, and generate the swallowing signal stable time period value;

[0073] First, the interval values are parsed in chronological order, and the start time and end time of each interval are used as query conditions respectively to call the corresponding swallowing signal channel data in the feeding record data. The signal amplitude sequence is read according to the millisecond-level time stamp. During the reading process, the amplitude change of each signal is recorded item by item. At the same time, the signal slope between adjacent two points is calculated, that is, the difference in amplitude change per unit time, and all slope values are judged against the set swallowing action slope threshold. If the slopes of consecutive signal points are all within the stable interval, for example, the set threshold is 0.005g / ms, only the signal segments with slope fluctuations within this threshold are retained. Further statistical analysis is carried out on the amplitudes within the retained segments, and the maximum amplitude, minimum amplitude, and average amplitude within the segment are called to calculate the amplitude fluctuation value, which is then compared with the swallowing amplitude stability reference value. For example, in a certain segment of data, the maximum amplitude is 0.084g, the minimum amplitude is 0.079g, and the average amplitude is 0.081g, then the amplitude fluctuation is 0.005g. If the fluctuation is less than the swallowing amplitude stability threshold of 0.008g, this segment is determined to be an amplitude stable segment. Then, the time segments that meet both the slope and amplitude judgment criteria are recorded as candidate swallowing signal segments, and the start and end times of this segment are used as label identifiers for retention. Subsequently, the rhythm consistency of multiple segments is judged. The average time interval between signal points within each segment is calculated, and the difference between the average interval values of all segments is judged. If the difference is within the rhythm consistency judgment threshold, for example, set to 20ms, this segment is retained as a rhythm consistent signal segment. If the average interval of a certain segment is 105ms and the average of other segments is 100ms, then the difference is 5ms, meeting the rhythm consistency requirement. Finally, all signal segments that meet the requirements are integrated into a signal segment data structure with a unified structure, and the start time and end time of each segment are recorded, such as "10:00:08.050~10:00:09.500", "10:00:10.200~10:00:11.700", etc., and finally the stable time period value of the swallowing signal is generated.

[0074] S203: Based on the sucking continuous segment interval value and the stable time period value of the swallowing signal, extract the time intersection segment, call the amplitude sequence of the sucking segment in the intersection, calculate the fluctuation amplitude of the sucking action within the segment, and at the same time count the proportion of the total duration of the intersection segment in the total duration of the sucking segment to obtain the feeding rhythm characteristic segment;

[0075] The specific calculation formula for the fluctuation amplitude of the sucking action within the segment is:

[0076]

[0077] Among them, W v represents the fluctuation amplitude of the sucking action within the segment, V si represents the i-th sucking signal amplitude sampling point, V srepresents the average value of the sucking signal amplitude sequence, max(V s ) represents the maximum value of the sucking amplitude sequence, min(V s ) represents the minimum value of the sucking amplitude sequence, n represents the total number of samples of the sucking signal within this segment, and δ t represents the ratio of the variance of the time intervals between adjacent sampling points to the average interval;

[0078] Suppose in a specific feeding monitoring, the following sucking amplitude sequence (unit: g) is recorded

[0079] V s = [0.075, 0.080, 0.085, 0.078, 0.082];

[0080] Based on these values, we perform the following calculations:

[0081] max(V s ) = 0.085;

[0082] min(V s ) = 0.075;

[0083]

[0084] Suppose the variability of the time intervals is 0.1, and this value is obtained from the statistical analysis of the actual time interval data. For example, the variance of the time intervals is 0.002 square seconds, and the average time interval is 0.02 seconds.

[0085] Substitute into the formula:

[0086]

[0087] The result shows that in the selected sucking segment, the amplitude fluctuation is very small, accounting for 0.0386% of the average amplitude, indicating that the sucking action is relatively stable within this section. This fluctuation amplitude is an important indicator to help identify and analyze the sucking stability of the baby during feeding, and can then be used to optimize the feeding strategy to ensure the comfort and efficiency of the baby during feeding.

[0088] The specific steps of S3 are as follows:

[0089] S301: Based on the feeding rhythm characteristic segment, extract the sucking trajectory data within the corresponding time period, locate the frequently fluctuating time points in each segment of the sucking signal, screen the time segments with concentrated peak distributions as the high-frequency point set of behaviors, and generate the high-frequency time segments of sucking;

[0090] First, extract the start and end time information recorded in each rhythm segment, and use this as an index to locate the corresponding pressure sequence interval from the original sucking pressure trajectory. In each interval, establish a curve structure with time as the horizontal axis and pressure value as the vertical axis, and perform peak extraction operations on this curve. By comparing the pressure values of each sampling point with the pressure values of its two adjacent time points, identify the time nodes that are local maxima. If the pressure value at this node is higher than the values on both sides, and the difference is not less than the sucking fluctuation difference benchmark (the set benchmark is 0.3 kPa), then this node is marked as a peak point. For example, if 17.6 kPa, 18.0 kPa, and 17.4 kPa are continuously recorded in a certain pressure sequence, then the middle 18.0 kPa is a peak point. Subsequently, collect the time positions of all peak points, sort them by the occurrence time. In adjacent time segments, if the number of peaks appearing within a unit time exceeds the peak density determination threshold (such as more than 3), then this time period is marked as a peak concentration segment. Perform the same screening operation in all rhythm segments, and record the identified peak concentration time periods. At the same time, establish a table record structure with parameters such as segment number, start and end times, number of peaks, average time interval, etc. as fields. For example: number T03, start time 42.5 seconds, end time 48.0 seconds, number of peaks 7, average interval 0.8 seconds. Finally, number and file all the recorded high-frequency segments, arrange them in the feeding order, and output them as sucking high-frequency time segments.

[0091] S302: Based on the sucking high-frequency time segments, match the vibration data in the corresponding time period of the swallowing record, extract the vibration-intensive interval, identify the time range overlapping with the sucking high-frequency, and mark the distribution position and total occurrence times of the overlapping interval in complete cycles to generate the sucking-swallowing synchronization segment;

[0092] First, it is necessary to define the high-frequency sucking time period, which is usually accomplished by identifying the high-frequency components in the signal. For example, by setting a threshold to filter out the higher-frequency parts of the sucking signal. Specifically, when the frequency of the signal exceeds a certain set value, this time period is marked as the high-frequency sucking time period. For example, assume that the high-frequency time period of the sucking signal is identified as [10s, 15s], and the signal frequency within this time period is significantly higher. Next, extract the part of the swallowing record that overlaps with the sucking signal time period. This process can be achieved by comparing the timestamps of the swallowing signal and the sucking signal. Assume that the time period of the swallowing signal is [12s, 18s], which partially overlaps with the sucking signal time period [10s, 15s], then the overlapping interval is [12s, 15s]. After obtaining the overlapping interval, the next step is to extract the vibration-dense interval. This process is completed by evaluating the change intensity of the vibration signal within the time period. Usually, the intensity of the vibration signal can be determined by the amplitude change of the signal. Assume that the amplitude change of the vibration signal is significant within the interval [12.5s, 14s], so this interval is determined as the vibration-dense interval. Next, by analyzing the intersection of the high-frequency time period of the sucking signal and the vibration-dense interval, their overlapping interval can be determined. For example, if the vibration-dense interval is [12.5s, 14s], the overlapping part with the high-frequency sucking time period [10s, 15s] is [12.5s, 14s]. Finally, mark the distribution position and total occurrence times of the overlapping interval according to the complete cycle. This step is completed by calculating the number of vibration signal cycles within the overlapping time period. Assume that the vibration period within this overlapping interval is 2 seconds, then within the interval [12.5s, 14s], the number of occurrences of the vibration period is 0.75 cycles, and then mark the cycle for the overlapping interval. Finally, the generated sucking-swallowing synchronization section is [12.5s, 14s], and this interval is the time period when sucking and swallowing occur synchronously.

[0093] S303: Based on the sucking-swallowing synchronization section, calculate the density of behaviors within the synchronization section, measure the change difference in the time interval between adjacent synchronization sections, and combine the distribution frequency statistical values to integrate the behavior cooperation characteristic indicators to obtain the cooperation performance of sucking and swallowing;

[0094] The specific calculation formula for the density of behaviors within the synchronization section is as follows:

[0095]

[0096] Among them, D i represents the density of behaviors within the i-th synchronization section, n j represents the number of occurrences of the j-th behavior within the i-th section, M represents the total number of all behaviors within the section, T i represents the duration of the i-th synchronization section, T avgrepresents the average value of the durations of all synchronization sections, and σ represents the standard deviation;

[0097] Suppose the following data of the synchronization sections are collected:

[0098] Section 1: T1 = 10000 (10 seconds), Behavior statistics: Sucking behavior appears 5 times, swallowing behavior appears 2 times;

[0099] Section 2: T2 = 12000 (12 seconds), Behavior statistics: Sucking behavior appears 6 times, swallowing behavior appears 3 times;

[0100] Section 3: T3 = 8000 (8 seconds), Behavior statistics: Sucking behavior appears 4 times, swallowing behavior appears 1 time;

[0101] In addition, suppose the average value T of the durations of all sections avg = 10000 (10 seconds), and the calculated standard deviation σ = 1500 (1.5 seconds);

[0102] Calculate the density of each section:

[0103] Calculate the density D1 of Section 1:

[0104] n1 = 5 (number of sucking behavior times), n2 = 2 (number of swallowing behavior times);

[0105] M = 5 + 2 = 7;

[0106] T1 = 10000 (10 seconds);

[0107] Calculate using the formula:

[0108]

[0109]

[0110] Calculate the density D2 of Section 2:

[0111] n1 = 6 (number of sucking behavior times), n2 = 3 (number of swallowing behavior times);

[0112] M = 6 + 3 = 9;

[0113] T2 = 12000 (12 seconds);

[0114] Calculate using the formula:

[0115]

[0116] Calculate the density D3 of Section 3:

[0117] n1 = 4 (number of sucking behavior times), n2 = 1 (number of swallowing behavior times);

[0118] M = 4 + 1 = 5;

[0119] T3 = 8000 (8 seconds);

[0120] Calculate using the formula:

[0121]

[0122] The result shows that through the above calculations, the intensity of behaviors within each synchronization section is obtained: the intensity of section 1 is 0.3889%, the intensity of section 2 is 0.4373%, and the intensity of section 3 is 0.2857%. The result represents the intensity of behaviors within the differential synchronization sections. A higher intensity (section 2) indicates that sucking and swallowing behaviors occur more frequently within the section, while a lower intensity (section 3) indicates fewer behaviors in the section. Through calculation, the coordination of sucking and swallowing behaviors during different time periods can be analyzed, providing data support for the analysis of feeding rhythm.

[0123] The specific steps of S4 are as follows:

[0124] S401: Based on the coordination performance of sucking and swallowing, obtain the time series of sucking and swallowing signals in the feeding record data, extract the time intervals and action amplitude differences between adjacent actions, and screen out the continuous paragraphs whose rhythm fluctuation amplitude conforms to the stable rhythm of sucking and swallowing to obtain the stable rhythm time period interval;

[0125] First, extract the signal recording sequences of two channels from the feeding record data, corresponding to the sucking signal and the swallowing signal respectively. Each record contains a timestamp and a signal amplitude. Sort the data of the two channels according to the time tag to construct a complete time-series data stream. Then, perform a pairing operation on the times of the sucking and swallowing event points to obtain the time difference between each pair of adjacent events and calculate its interval duration. If the time interval between a pair of actions is greater than the rhythm continuity judgment threshold, for example, the threshold is set to 5 seconds, then this pair of records will not be processed as an object for rhythm analysis in the subsequent process. For the event point pairs that meet this condition, continue to call their corresponding sucking or swallowing amplitudes and calculate the difference between the amplitudes of the two points. Further screen whether the amplitude difference is within the allowable range of rhythm fluctuations. Set the sucking amplitude difference threshold to 0.02g and the swallowing amplitude difference threshold to 0.015g. If the amplitudes of two adjacent sucking events are 0.080g and 0.095g respectively, then the difference is 0.015g, which meets the sucking rhythm stability condition. Similarly, judge whether the difference between adjacent swallowing amplitudes is within the threshold. If it meets the condition, retain this event segment. Merge multiple consecutive action segments that meet the rhythm stability condition to form a candidate segment for stable rhythm. Subsequently, calculate the time intervals of all actions within this segment again to obtain the difference between the maximum and minimum values of the rhythm fluctuation range, and judge whether this value is less than the upper limit of the rhythm consistency determination. For example, the rhythm fluctuation is set not to exceed 0.5 seconds. If the maximum interval within a segment is 1.6 seconds, the minimum interval is 1.2 seconds, and the difference is 0.4 seconds, which is less than the set threshold, then this segment is considered to have stable rhythm fluctuations. Then record the start time and end time of this segment to form a time interval output in a unified format, such as "10:02:01.200~10:02:07.800", and finally obtain the stable rhythm time period interval.

[0126] S402: Based on the signal fluctuation sequence corresponding to the stable rhythm time period interval, compare the amplitude changes at adjacent time points, retain the continuous segments with stable fluctuation amplitudes, extract the time periods that meet the conditions, and obtain the slow-changing action segment time zone;

[0127] First, call the sucking and swallowing signals in the feeding record data according to the start and end times of the stable rhythm period, extract the corresponding original signal amplitudes item by item within each segment, sort them in ascending order according to the timestamps, and construct a complete and continuous sequence of fluctuation values. On this basis, calculate the difference in amplitude between any two adjacent sampling points, record the magnitude of the instantaneous amplitude change in each continuous sequence, and compare it with the set fluctuation stability threshold. This threshold is set based on the sensor noise level and the range of the baby's normal physiological rhythm. For example, if the fluctuation stability threshold is 0.004g, then only when the amplitude change does not exceed this value is it judged that the fluctuation amplitude is stable. Traverse all pairs of sampling points and continuously mark the point sequences that meet the conditions, then merge adjacent stable points into continuous segments to form a set of candidate slow-varying action segments. For each candidate segment, continue to judge whether its total duration exceeds the minimum duration requirement of the slow-varying period. For example, the minimum duration is set to 1.5 seconds. If a segment lasts from "10:05:30.200" to "10:05:31.400", its duration is 1.2 seconds, which does not meet the requirement and is discarded. However, if a segment is 2.0 seconds long, it is retained. At the same time, summarize and count the maximum instantaneous change amplitude within each segment, and verify whether it is all lower than the upper limit of the slow-varying tolerance. If it meets the condition, the segment is confirmed as a valid slow-varying action segment. Finally, summarize all the eligible continuous periods and construct a unified output structure according to the start and end times, such as "10:03:05.100~10:03:07.200", "10:06:12.600~10:06:14.800", and finally obtain the time zone of the slow-varying action segment.

[0128] S403: Based on the feeding cycle number to which the time zone of the slow-varying action segment belongs, count the occurrence frequency in the cycle, extract the signal amplitude value and rhythm interval value of the segment, classify them into sucking and swallowing behavior types accordingly, and accumulate the number of segments by behavior group to obtain the milk supply situation;

[0129] First, the start and end time of each slowly changing action segment is compared with the existing feeding cycle interval to determine whether it falls within a specific feeding cycle. If the start and end time are completely included in a cycle, the segment is assigned to the corresponding feeding cycle number, and a corresponding relationship between the cycle number and the segment number is established. For example, if the time of a segment is "10:04:15.000~10:04:18.500", and its feeding cycle number is cycle 3, then the segment is marked as "Segment ID003-Cycle 3". Then, the segments contained in each cycle number are counted and the frequency of occurrence of slowly changing action segments in each feeding cycle is obtained. At the same time, the corresponding signal amplitude sequence in each segment is called and the maximum, minimum and average values in the segment are calculated to form a segment amplitude parameter set. Then, the time interval sequence of continuous signal points is extracted and its average rhythm interval value is calculated to construct the rhythm of each segment. For example, if the average amplitude of a segment is 0.081g and the average rhythm interval is 0.95 seconds, this group of values will be used to determine the behavior type. The amplitude range of sucking behavior is set to 0.07g-0.09g, the rhythm interval is 0.8-1.2 seconds, and the amplitude range of swallowing behavior is set to 0.09g-0.12g, and the rhythm interval is 1.3-2.0 seconds. If the amplitude and interval of the segment fall within the sucking behavior threshold range, the segment is marked as sucking type. Conversely, if it falls within the swallowing interval, it is marked as swallowing type. After classification, the number of each behavior type is accumulated according to the feeding cycle, and the number of sucking and swallowing segments in each cycle is counted. For example, in cycle 3, 5 sucking segments and 2 swallowing segments are identified, which indicates that the behavior composition feature of this cycle is dominated by sucking. Finally, the distribution of behavior types in each cycle is output as a behavior statistics table, and the milk supply situation is finally obtained.

[0130] The specific steps of S5 are:

[0131] S501: Based on the milk supply situation, combined with the sucking and swallowing coordination performance and feeding rhythm feature segments, the time segments, movement intensity and flow rate types in the paragraphs are extracted, the behavior data is split by dimension, and the class performance features in the behavior standard description are compared item by item to generate a behavior comparison mapping table;

[0132] First, it is necessary to combine the sucking and swallowing coordination performance with the feeding rhythm characteristic segments, extract the time intervals, action intensity, and flow rate types in the paragraphs. First, through the analysis of milk flow rate data and behavioral data, determine the occurrence time and intensity of sucking and swallowing behaviors in each time interval. Suppose in the interval [12s, 13s], the sucking behavior occurs at a higher frequency, with an intensity of 1.5 times per second, while the swallowing behavior is relatively sparse, with an intensity of 0.5 times per second. And the flow rate type shows a gradual increase in this interval. Suppose the flow rate increases from 0.5 mL / s to 1.0 mL / s. Next, split these behavioral data by dimension. Split them into different segments such as [12s, 13s], [14s, 15s] according to time intervals, split them into high-intensity segments (such as [12s, 13s]) and low-intensity segments (such as [14s, 15s]) according to action intensity, and split them into categories such as gradually increasing type, stable type, and gradually decreasing type according to flow rate type. After splitting, compare the class performance characteristics in the behavioral standard description item by item. This step is to compare through pre-set behavioral standards, such as whether the sucking intensity exceeds 1.0 times per second, whether the swallowing behavior is lower than 0.3 times per second, etc. Suppose the interval [12s, 13s] meets the standard of high-intensity sucking behavior, while [14s, 15s] meets the standard of stable flow rate and low-density swallowing behavior. Through these comparisons, finally generate a behavior comparison mapping table, match each time interval with its corresponding behavior type, intensity, flow rate type, etc. information to form a complete behavior comparison mapping table. Suppose the mapping table records behavior patterns such as [12s, 13s] corresponding to high-intensity sucking and gradually increasing flow rate, and [14s, 15s] corresponding to low-density swallowing and stable flow rate, thus realizing the precise mapping of behavior and milk flow rate.

[0133] S502: Based on the behavior comparison mapping table, extract the time segments corresponding to each type of standard behavior in the whole process, count the start and end intervals, occurrence frequencies, and interval distribution characteristics of the behavior distribution in the recording period, construct the performance classification of multiple types of behaviors in the cycle, and generate the behavior performance distribution structure;

[0134] First, it is necessary to extract the time segments corresponding to each type of standard behavior throughout the whole process. This process is completed by looking up the time intervals of each type of behavior in the behavior comparison mapping table. Assume that in the mapping table, the sucking behavior appears in the time intervals of [12s, 13s], [17s, 18s], and [21s, 22s], and the swallowing behavior appears in the time intervals of [14s, 15s], [19s, 20s], and [23s, 24s]. Then, these time segments can be extracted. Next, count the start and end intervals, occurrence frequencies, and interval distribution characteristics of the behavior distribution in the recording period. First, record the start and end intervals of each behavior. Assume that the intervals of the sucking behavior are [12s, 13s], [17s, 18s], and [21s, 22s], and the intervals of the swallowing behavior are [14s, 15s], [19s, 20s], and [23s, 24s]. Then, the occurrence frequency of each behavior is 3 times. For the interval distribution of each behavior, it can be obtained by calculating the time intervals between adjacent segments. Assume that the intervals of the sucking behavior are [13s, 14s], [18s, 19s], and [22s, 23s], and the interval of each segment is 1 second. Then, construct the performance classification of multiple types of behaviors in the cycle. This step is to classify the performance of each behavior in different time periods. Assume that in the time period of [12s, 13s], the sucking behavior shows high-frequency and relatively intense actions, and in the time period of [14s, 15s], the swallowing behavior shows relatively low-frequency actions. Classify according to these characteristics. By analyzing the occurrence frequency, interval distribution, and behavior intensity of the behavior, construct a structured behavior performance distribution. Assume that the distribution of the sucking behavior shows high frequency concentrated in the early stage and low frequency concentrated in the later stage, while the swallowing behavior appears relatively evenly throughout the whole process. The finally generated behavior performance distribution structure can clearly reflect the time characteristics, frequencies, and interval characteristics of different behaviors.

[0135] S503: Based on the behavior performance distribution structure, splice the time periods of multiple types of behaviors in sequence and form a time map. Integrate and describe the behaviors in different time periods according to the performance attribution, summarize them into a complete set of behaviors within a one-time cycle, and generate the breast-feeding evaluation result;

[0136] First, it is necessary to sequentially splice multiple types of behavior time periods to form a time map. This process is achieved by connecting the time periods of different behaviors in sequence to form a complete time series. For example, in the aforementioned behavior mapping table, the time periods of the sucking behavior are [12s, 13s], [17s, 18s], [21s, 22s], and the time periods of the swallowing behavior are [14s, 15s], [19s, 20s], [23s, 24s]. These time periods are spliced in sequence in the time map to form a complete behavior time series from [12s] to [24s]. Next, the behaviors in the differential time periods are integrated and described according to the performance attribution. This process is to classify according to the performance characteristics of different behaviors in each time period. Suppose the sucking behavior shows a high frequency and a relatively large intensity in the [12s, 13s] section, while the swallowing behavior shows a low frequency and a relatively small intensity in the [14s, 15s] section. Then, the behaviors in these two time periods can be described and classified as high-intensity sucking and low-frequency swallowing. Further, these classified time periods are combined into a complete behavior set within a single cycle. For example, assume that the entire time cycle is [12s, 24s]. The alternating appearance of the sucking and swallowing behaviors and their performance characteristics constitute the complete behavior pattern of this cycle. Finally, the breastfeeding evaluation result is generated. By integrating and analyzing these behavior time periods and their performance characteristics, the breastfeeding evaluation result can be obtained. For example, this process may yield an evaluation such as "the coordination between high-frequency sucking and low-frequency swallowing is good, and the flow rate changes smoothly", indicating that the coordination between sucking and swallowing within the entire cycle is relatively good and meets the healthy feeding standard.

[0137] Please refer to Figure 2 , a construction system for breastfeeding evaluation indicators, comprising:

[0138] The feeding acquisition module installs a vibration sensor on the baby's chest to real-time monitor the sucking and swallowing action signals, associates the data with the start and end times of feeding, performs unified time-axis processing on the two action channels, and integrates and generates a structured output including time sequence, amplitude change, and frequency to obtain feeding record data;

[0139] The rhythm extraction module, based on the feeding record data, extracts the action sections in the sucking signal that are continuous and have an interval time meeting the threshold, records the start and end times and extracts the amplitude sequence, locates the sections in the swallowing signal with stability meeting the standard, performs intersection processing, calculates the fluctuation amplitude of the sucking action, and counts the duration ratio of the intersection segments in the sucking segments to obtain the feeding rhythm characteristic segments;

[0140] The cooperation analysis module locates the time nodes where the amplitude of the sucking signal changes frequently based on the rhythm feature segments, selects the time intersection with the significantly changing interval in the swallowing signal, calculates the ratio of the number of sucking times to the swallowing duration within the synchronous section, generates the synchronous density, and analyzes the time interval fluctuation of adjacent synchronous segments to obtain the cooperation performance of sucking and swallowing;

[0141] The behavior classification module filters out the slow-changing action segments with gentle amplitude and standard slope based on the cooperation performance of sucking and swallowing, extracts the average amplitude value and rhythm interval value of the sucking or swallowing signal, classifies them according to the standard, records the occurrence frequency and proportion of behaviors in each cycle, and obtains the milk supply situation;

[0142] The index generation module, based on the milk supply situation, the cooperation performance of sucking and swallowing, and the rhythm feature segments, statistically counts the duration and number of segments that meet the conditions according to the standard, arranges the behavior distribution within the cycle, and generates the breastfeeding evaluation result for a single cycle.

[0143] As described above, it is only the specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art within the technical scope disclosed by the present invention can easily think of changes or substitutions, which should all be covered within the protection scope of the present invention. Therefore, the protection scope of the present invention should be subject to the protection scope of the claimed rights.

Claims

1. A method for constructing an evaluation index for breastfeeding, characterized in that, It includes the following steps: S1: Obtain the feeding time and frequency during breast-feeding. Through a vibration sensor deployed on the baby's chest, collect the number of sucking and swallowing times of the baby, integrate the data in chronological order, and obtain feeding record data. S2: Based on the feeding record data, extract continuous segments of sucking signals and stable swallowing phases, identify rhythm continuation segments, record the start and end times, calculate the sucking fluctuation amplitude, and count the proportion of time periods to obtain feeding rhythm characteristic segments. S3: Based on the feeding rhythm characteristic segments, extract high-frequency sucking time points and compare them with the concentrated time periods of swallowing actions, identify overlapping segments and extract synchronous time intervals, calculate the matching density of sucking and swallowing actions within the synchronous segments, and measure the interval fluctuation to obtain the sucking and swallowing coordination performance. S4: Based on the sucking and swallowing coordination performance, retrieve stable feeding segments, screen out slowly changing segments, record the start and end times and frequencies, classify and count according to feeding behavior types, and obtain the milk supply situation. S5: Based on the milk supply situation, sucking and swallowing coordination performance, and feeding rhythm characteristic segments, classify the baby's behavior types, count the time and frequencies, integrate the summary of feeding behaviors, and generate a breast-feeding evaluation result.

2. The construction method of the breastfeeding evaluation index according to claim 1, characterized in that: The feeding record data includes feeding time values, feeding frequency values, sucking times values, and swallowing times values. The feeding rhythm characteristic segments are specifically sucking continuous interval values, swallowing stable time period values, sucking fluctuation amplitude values, and rhythm time period proportion values. The sucking and swallowing coordination performance includes synchronous time intervals, action matching density values, and rhythm interval fluctuation values. The milk supply situation is specifically stable feeding frequency values, slowly changing segment quantity values, and behavior type classification results. The breast-feeding evaluation result specifically refers to the baby's behavior type classification result, behavior duration values, behavior occurrence frequencies, and behavior summary content.

3. The construction method of the breastfeeding evaluation index according to claim 1, wherein: The specific steps of S1 are: S101: Obtain the sucking and swallowing action signals collected by the vibration sensor on the baby's chest, record the corresponding time points, pair the actions with the time and integrate them in order into a time series to generate a sucking and swallowing action sequence. S102: Based on the sucking and swallowing action sequence, identify the times corresponding to the first sucking signal and the last swallowing signal, construct a feeding start and end time interval, and mark the action timestamps during this period to obtain a feeding cycle time interval. S103: Based on the sucking and swallowing action sequence and the feeding cycle time interval, screen out the events within the interval and sort them, count the number of events and intervals, generate feeding frequency values and feeding time values, and obtain feeding record data.

4. The method for constructing the breastfeeding evaluation index according to claim 1, wherein The specific steps of S2 are: S201: Based on the feeding record data, extract segments with continuous signals and an interval time not exceeding the sucking interval value, screen out the time periods that meet the conditions, and obtain sucking continuous segment interval values. S202: Based on the time range in the sucking continuous segment interval values, extract the corresponding swallowing signal sequences, screen out the time periods with stable amplitudes and gentle slopes, retain the signal segments with consistent rhythms, and generate swallowing signal stable time period values. S203: Based on the sucking continuous segment interval value and the swallowing signal stable time period value, extract the time intersection segment, call the amplitude sequence of the sucking segments in the intersection, calculate the fluctuation amplitude of the sucking action within the segment, and at the same time count the proportion of the total duration of the intersection segment in the total duration of the sucking segment to obtain the feeding rhythm characteristic segment.

5. The construction method of the breastfeeding evaluation index according to claim 1, wherein: The specific formula for calculating the fluctuation amplitude of the sucking action within the segment is as follows: Among them, W v represents the fluctuation amplitude of the sucking action within the section, represents the sampling point of the amplitude of the i-th sucking signal, V s represents the average value of the amplitude sequence of the sucking signal, max(V s ) represents the maximum value of the sucking amplitude sequence, min(V s ) represents the minimum value of the sucking amplitude sequence, n represents the total number of samples of the sucking signal within this segment, δ t represents the ratio of the variance of the time interval between adjacent sampling points to the average interval.

6. The construction method of the breastfeeding evaluation index according to claim 1, characterized in that, The specific steps of S3 are as follows: S301: Based on the feeding rhythm characteristic segment, extract the sucking trajectory data within the corresponding time period, locate the frequently fluctuating time points in each sucking signal, and screen the time segments with concentrated peak distributions as the behavior high-frequency point set to generate the sucking high-frequency time segment; S302: Based on the sucking high-frequency time segment, match the vibration data in the corresponding time period in the swallowing record, extract the vibration dense interval and identify the time range overlapping with the sucking high-frequency time, and mark the distribution position and total occurrence times of the overlapping interval according to the complete cycle to generate the sucking-swallowing synchronization segment; S303: Based on the sucking-swallowing synchronization segment, calculate the density of the behavior within the synchronization segment, measure the change difference of the time interval between adjacent synchronization segments, and combine the distribution frequency statistical value to integrate the behavior cooperation characteristic index to obtain the sucking and swallowing cooperation performance.

7. The construction method of the breastfeeding evaluation index according to claim 1, characterized in that: The specific formula for calculating the density of the behavior within the synchronization segment is as follows: Among them, D i represents the density of behaviors in the i-th synchronization section, n j represents the number of occurrences of the j-th behavior in the i-th section, M represents the total number of all behaviors in this section, T i represents the duration of the i-th synchronization section, T avg represents the average value of the durations of all synchronization sections, and σ represents the standard deviation.

8. The construction method of the breastfeeding evaluation index according to claim 1, characterized in that, The specific steps of S4 are as follows: S401: Based on the sucking and swallowing cooperation performance, obtain the time series of the sucking and swallowing signals in the feeding record data, extract the time interval and action amplitude difference between adjacent actions, and screen the continuous paragraphs with the rhythm fluctuation amplitude conforming to the stability of the sucking and swallowing rhythms to obtain the stable rhythm time period interval; S402: Based on the signal fluctuation sequence corresponding to the stable rhythm time period interval, compare the amplitude changes at adjacent time points, retain the continuous segments with stable fluctuation amplitude, and extract the time periods that meet the conditions to obtain the slow-changing action segment time zone; S403: Based on the feeding cycle number to which the slow-changing action segment time zone belongs, count the occurrence frequency in the cycle, extract the signal amplitude value and rhythm interval value of the segment, classify them into the sucking and swallowing behavior types accordingly, and accumulate the segment quantities according to the behavior groups to obtain the milk supply situation.

9. The construction method of the breastfeeding evaluation index according to claim 1, characterized in that, The specific steps of S5 are as follows: S501: Based on the milk supply situation, combine the sucking and swallowing cooperation performance and the feeding rhythm characteristic segment, extract the time segments, action density and flow rate types in the paragraph, split the behavior data according to the dimensions, and compare each item with the class performance characteristics in the behavior standard description to generate the behavior comparison mapping table; S502: Based on the behavior comparison mapping table, extract the corresponding time segments of each standard behavior in the whole process, count the start and end intervals, occurrence frequencies and interval distribution characteristics of the behavior distribution in the recording period, construct the performance classification of multiple behaviors in the cycle, and generate the behavior performance distribution structure; S503: Based on the behavior performance distribution structure, splice the time segments of multiple behaviors in sequence to form a time map, integrally describe the behaviors in the differentiated time segments according to the performance attribution, and summarize them into a complete behavior set within one-time cycle to generate the breastfeeding evaluation result.

10. A construction system for breastfeeding evaluation indicators, characterized in that, The system is used to implement the construction method of the breastfeeding evaluation index according to any one of claims 1-9. The system includes: The feeding acquisition module installs a vibration sensor on the baby's chest to monitor the sucking and swallowing action signals in real time, associates the data with the start and end times of feeding, performs unified time-axis processing on the two action channels, and integrates and generates a structured output including time series, amplitude change, and frequency to obtain feeding record data; The rhythm extraction module, based on the feeding record data, extracts action segments in the sucking signal that are continuous and have an interval time meeting the threshold, records the start and end times and extracts the amplitude sequence, locates the segments in the swallowing signal with stability meeting the standard, performs intersection processing, calculates the fluctuation amplitude of the sucking action, and statistically calculates the duration ratio of the intersection segment in the sucking segment to obtain the feeding rhythm characteristic segment. The coordination analysis module, based on the rhythm characteristic segment, locates the time nodes with frequent amplitude changes in the sucking signal, selects the time intersection with the significant change interval in the swallowing signal, calculates the ratio of the number of sucking times to the swallowing duration in the synchronous segment, generates the synchronous density, and analyzes the time interval fluctuation of adjacent synchronous segments to obtain the coordination performance of sucking and swallowing. The behavior classification module, based on the coordination performance of sucking and swallowing, screens out slow-varying action segments with gentle amplitude and slope meeting the standard, extracts the amplitude mean value and rhythm interval value of the sucking or swallowing signal, classifies them according to the standard, and records the occurrence frequency and proportion of behaviors in each cycle to obtain the milk supply situation. The index generation module, based on the milk supply situation, the coordination performance of sucking and swallowing, and the rhythm characteristic segment, statistically calculates the duration and number of segments meeting the conditions according to the standard, arranges the behavior distribution in the cycle, and generates the breastfeeding evaluation result for a single cycle.

Images