System and method for providing child growth optimization curation service using causal correlation analysis of multi-modal data at each stage of development and caregiver behavior logs

The system analyzes multimodal growth data and caregiver behavior logs to provide personalized parenting missions, addressing the limitations of conventional services by scientifically identifying effective strategies and continuously improving recommendations.

KR103046682B1Active Publication Date: 2026-11-11FOX CONNECT CO LTD
View PDF 5 Cites 0 Cited by

Patent Information

Application Number
KR1020260055330
Authority / Receiving Office
KR · KR
Patent Type
Patents
Current Assignee / Owner
Filing Date
2026-03-27
Publication Date
2026-11-11
Estimated Expiration
2046-03-27

AI Technical Summary

Technical Problem

Conventional child growth management services fail to reflect individual children's characteristics and the impact of caregiver behaviors on their development, lacking sophisticated analysis of time-series data to identify effective parenting strategies.

Method used

A system and method that collects multimodal growth data and caregiver behavior logs, analyzes their correlation through cross-correlation analysis, and provides personalized parenting missions to support efficient parenting activities.

Benefits of technology

Provides customized parenting guides that scientifically identify the impact of caregiver behaviors on child development, preventing negative impacts and continuously refining recommendations through a feedback loop, while establishing a virtuous cycle of data expansion.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure 112026037373588-PAT00006_ABST
    Figure 112026037373588-PAT00006_ABST
Patent Text Reader

Abstract

The present invention relates to a system and method for providing a child growth optimization curation service using correlation analysis of multimodal data at developmental stages and caregiver behavior logs. The system is executed on a server including a control unit and communicates with a caregiver terminal to collect the child's multimodal growth data and the caregiver's behavior log data at different intervals, synchronize them using a scaling operation and a left outer join method, construct a correlation network based on positive (+) cross-correlation coefficients through cross-correlation analysis, identify vulnerable developmental areas, generate caregiver mission information effective for improving those areas by querying from a mission template storage area, and transmit it to the caregiver terminal via an encrypted communication channel.
Need to check novelty before this filing date? Find Prior Art

Description

Technology Field

[0001] The present invention relates to a system and method for providing a child growth optimization curation service, and more specifically, to a system and method for deriving a correlation between multimodal growth data and caregiver behavior log data according to developmental stages based on cross-correlation analysis, and providing personalized parenting missions based thereon. Background Technology

[0002] In modern society, the healthy growth and development of children is one of the most important concerns for caregivers. A child's growth occurs in a complex manner across physical, cognitive, and socio-emotional domains, and major developmental tasks and characteristics differ at each developmental stage.

[0003] Conventional child growth management services have mostly been limited to measuring fragmentary growth indicators (e.g., height, weight) or providing general developmental milestone information. This approach had limitations in that it failed to reflect the characteristics of individual children and differences in parenting environments, and it was difficult to identify the specific impact of caregiver behaviors on a child's growth.

[0004] In particular, there has been a lack of objective data-based analysis on how various parenting activities performed by caregivers (reading, play, conversation, etc.) affect which developmental areas of children. As a result, caregivers have struggled to establish effective parenting strategies.

[0005] In addition, since child development often occurs with a certain time delay rather than immediately, sophisticated analysis of time-series data is required to identify the relationship between specific parenting activities and developmental outcomes.

[0006] Accordingly, there is a need for the development of technology capable of providing parenting guides optimized for individual children by comprehensively collecting various growth indicators and analyzing their correlation with caregiver behavioral data. The problem to be solved

[0007] The technical problem that the present invention aims to solve is to provide a system and method for providing a child growth optimization curation service that collects multimodal growth data considering the characteristics of a child's developmental stage, analyzes the correlation with caregiver behavior log data, and provides customized parenting missions.

[0008] Another technical problem that the present invention aims to solve is to synchronize growth data and behavior log data collected at different intervals in a time series and to construct a correlation network that considers time lag effects through cross-correlation analysis.

[0009] Another technical problem that the present invention aims to solve is to support efficient parenting activities by automatically identifying vulnerable developmental areas and recommending parenting activities that contribute positively to the improvement of those areas.

[0010] Another technical problem that the present invention aims to solve is to continuously refine the correlation network by implementing a feedback loop based on the results of the nurturing mission.

[0011] Another technical problem that the present invention aims to solve is to establish a virtuous cycle structure that recursively expands the system's data assets by anonymizing the data of children who have successfully achieved growth goals and automatically converting it into role model data.

[0012] Another technical problem that the present invention aims to solve is to objectively certify a child's growth process by comprehensively evaluating the sincerity of data input, mission completion rate, and correlation clarity to calculate a quantified developmental reliability score, and issuing a digital growth certificate based on this. means of solving the problem

[0013] A method for providing a child growth optimization curation service according to an embodiment of the present invention for solving the above problem comprises: (a) a step in which the server receives child identification information and date of birth information from the caregiver terminal and stores them in a database;

[0014] (b) A step in which the server receives multimodal growth data measured at regular intervals from the caregiver terminal and stores it in the database, wherein the multimodal growth data includes at least two of weight data, height data, total sleep time data, number of syllables spoken, and number of peer interactions;

[0015] (c) the server receiving behavior log data including a parenting activity type code and interaction duration from the caregiver terminal at regular intervals and storing it in the database; and

[0016] (d) The server refers to the database and provides a child growth optimization curation service based on a plurality of growth data items and behavior log data according to the multimodal growth data.

[0017] (c) After step,

[0018] (e) a step in which the control unit calculates the current age of the child based on the date of birth information and classifies the child into one of the pre-set developmental stages according to the calculated age; and

[0019] (f) The control unit further includes the step of calculating the daily average of the number of times each parenting activity type code is performed and the daily average of the total interaction time for a certain period of the collected behavior log data, respectively, performing a scaling operation on the calculated daily average values ​​to convert them into weekly total data points, and then generating synchronized data by mapping the converted weekly total data points to the time series index of the multimodal growth data.

[0020] (f) After step,

[0021] (g) The control unit calculates a cross-correlation coefficient between a parenting activity type code and a plurality of growth data items based on synchronized data, and among the calculated cross-correlation coefficients, determines a value that shows a positive correlation and whose magnitude is greater than or equal to a preset threshold as a weight between the corresponding parenting activity type code and the corresponding growth data item and stores it in the database; further comprising the step of

[0022] (g) After step,

[0023] (h) A step in which the control unit compares each growth data item of the child’s current multimodal growth data with the average value of the corresponding item of a group of preceding children belonging to the same developmental stage within the database, and identifies a growth data item that differs from the average value in a negative direction by more than a preset ratio as a vulnerable developmental area; and

[0024] (i) The control unit extracts parenting activity type codes in descending order of weights for growth data items identified as vulnerable development areas, wherein the weights are greater than or equal to a preset threshold, and for each extracted parenting activity type code, generates parenting mission information by querying mission data records including a mission identifier, parenting activity type code, weight, mission level, mission title, mission description, goal execution time, goal execution frequency, and execution deadline from the mission template storage area of ​​the database, and transmits the generated parenting mission information to the caregiver terminal through a secure communication channel.

[0025] After the above (f) step,

[0026] (g1) A step in which the control unit generates a time series graph of the current child by arranging the behavior log data and the multimodal growth data used for synchronization processing in step (f) in parallel with respect to the time axis;

[0027] (g2) A step in which the control unit extracts role model data corresponding to a preset standard or higher from the data of a group of preceding children of the same developmental stage within the database and generates a time series graph for the role model data; and

[0028] (g3) The control unit further includes the step of generating comparison interface data in which the time series graph of the current child and the time series graph of the role model data are superimposed on the same time axis and transmitting it to the caregiver terminal.

[0029] In the above (h) step,

[0030] The method further includes the step of the control unit generating a notification message regarding the possibility of developmental delay and transmitting it to the caregiver terminal if the weekly change rate of the growth data item identified as the above vulnerable developmental area remains below a preset rate during a consecutive preset number of measurements.

[0031] The above (i) step is,

[0032] A step of calculating a priority score for each of the extracted parenting activity type codes based on the corresponding weight and the difference compared to the average value of the vulnerable development area;

[0033] A step of selecting the top 5 parenting activity type codes based on the above priority scores; and

[0034] It includes the step of converting the above-mentioned selected parenting activity type code into a three-stage hierarchical structure of daily detailed missions, weekly intermediate missions, and monthly comprehensive missions.

[0035] The present invention comprises: a control unit; a communication module connected to the control unit for transmitting and receiving data with the caregiver terminal; and a database for storing child identification information, date of birth information, multimodal growth data, caregiver behavior log data, relationship analysis data, role model data, and mission templates.

[0036] The above control unit is,

[0037] A data collection module configured to store child identification information, date of birth information, multimodal growth data, and behavior log data received through the communication module in the database;

[0038] A preprocessing module configured to calculate the child's current age in months based on the above date of birth information and classify it into one of the preset developmental stages, calculate the daily average of the number of times parenting activity type codes are performed and the daily average of the total interaction time for a certain period of behavior log data collected at regular intervals, respectively, perform a scaling operation on each calculated daily average value to convert it into weekly total data points, and then map it to the time series index of the above multimodal growth data to perform synchronization processing;

[0039] A correlation analysis module configured to generate correlation analysis data by calculating cross-correlation coefficients between parenting activity type codes and growth data items for the above-mentioned synchronized data, and determining as weights values ​​among the cross-correlation coefficients that exhibit a positive correlation and whose magnitude is greater than or equal to a preset threshold;

[0040] A developmental comparison module configured to compare each item of the child's current growth data with the average value of the corresponding item of a group of preceding children at the same developmental stage within the database, and to identify items lower than a preset ratio as vulnerable developmental areas; and

[0041] It includes a curation generation module configured to extract, in descending order, codes among the parenting activity type codes corresponding to the growth data items identified as the vulnerable developmental areas, wherein the weight is greater than or equal to a preset threshold, generate parenting mission information by querying a corresponding mission template from the database for each extracted parenting activity type code, and then transmit it to the caregiver terminal through the communication module. Effects of the invention

[0042] According to the present invention, various growth indicators are comprehensively collected by considering the characteristics of a child's developmental stage, and a customized parenting guide can be provided by objectively analyzing the correlation with the caregiver's behavioral data.

[0043] Furthermore, according to the present invention, by synchronizing data collected at different intervals and deriving correlations that reflect time-delay effects through cross-correlation analysis, the impact of parenting activities on child development can be scientifically identified. In particular, by adopting only positive (+) cross-correlation coefficients as causal weights, activities that have a negative impact on child development are fundamentally prevented from being included in recommended missions.

[0044] In addition, according to the present invention, by automatically recommending effective parenting activities for vulnerable developmental areas, caregivers can efficiently utilize limited time and resources to support the balanced growth of children.

[0045] In addition, according to the present invention, since the correlation network is continuously updated through a feedback loop, more accurate and effective parenting missions can be provided as the service usage period increases.

[0046] In addition, according to the present invention, by de-identifying successful growth case data and automatically converting it into role model data, a virtuous cycle structure can be established in which the system's data assets are recursively expanded.

[0047] In addition, according to the present invention, by calculating a developmental reliability score and issuing a digital growth certificate, the child's growth process can be objectively certified and the effects of parenting activities can be quantitatively verified. Brief explanation of the drawing

[0048] FIG. 1 is a sequence diagram showing the process of transmitting and receiving data between a system and a database according to one embodiment of the present invention. FIG. 2 is a flowchart illustrating a data management, preprocessing, and time series synchronization process according to an embodiment of the present invention. FIG. 3 is a block diagram showing the hardware and software module configuration of a child growth optimization curation service providing system according to one embodiment of the present invention. FIG. 4 is a flowchart illustrating the correlation analysis and causal weight determination process using cross-correlation analysis according to one embodiment of the present invention. FIG. 5 is an example of a user interface including a radial chart for calculating the age of a target child in months, classifying developmental stages, and analyzing multimodal growth according to one embodiment of the present invention. FIG. 6 is an example of a user interface showing a notification function for checking detailed numerical values ​​by indicator and automatically identifying vulnerable development areas according to an embodiment of the present invention. FIG. 7 is an example of a user interface showing a time-series synchronization graph of behavior logs and growth indicators and the results of causal weight analysis by parenting activity according to one embodiment of the present invention. FIG. 8 is an example of a user interface illustrating the generation of personalized parenting mission information, card-shaped visualization, and encryption channel transmission functions according to one embodiment of the present invention. FIG. 9 is an example of a user interface for a child monitoring dashboard that allows for intuitive understanding of the child's overall developmental status according to one embodiment of the present invention. FIG. 10 is an example of a user interface providing curation results based on caregiver behavior log analysis and AI causal correlation according to one embodiment of the present invention. Specific details for implementing the invention

[0049] Hereinafter, preferred embodiments of the present invention will be described in detail with reference to the attached drawings. The advantages and features of the present invention, and the methods for achieving them, will become clear by referring to the embodiments described in detail below in conjunction with the attached drawings.

[0050] The server system (100) of the present invention includes a control unit (101), a communication module (110) that transmits and receives data to and from a caregiver terminal (200), and a database (120) that stores various data. The control unit (101) includes a data collection module (130), a preprocessing module (140), a correlation analysis module (150), a development comparison module (160), and a curation generation module (170), and each of the modules is a logical component that is implemented in software under the control of the control unit (101).

[0051] The above database (120) includes an area for storing child identification information, date of birth information, multimodal growth data, caregiver behavior log data, correlation network data, and role model data, and a mission template storage area for storing mission templates.

[0052] The data collection module (130) stores multimodal growth data received once a week in a first cycle and behavior log data received once a day in a second cycle through the communication module (110) in the database (120).

[0053] The above preprocessing module (140) calculates the child's current age in months based on the date of birth information and classifies the developmental stage, calculates the daily average of the number of times each parenting activity type code is performed and the daily average of the total interaction time for the 7-day values ​​of the behavior log data collected in the second cycle, performs a scaling operation by multiplying each calculated daily average value by 7 to convert it into a total data point on a weekly basis, and then performs synchronization processing by mapping the converted data point using a Left Outer Join method with the time series index of the multimodal growth data as the key to generate synchronized data. At this time, during the joining process, for weeks where behavior log data does not exist (missing values), the corresponding value is replaced with 0.

[0054] The correlation analysis module (150) calculates cross-correlation coefficients for each of multiple delay intervals of 1 week, 2 weeks, and 4 weeks for the synchronized data, and determines the value with the largest absolute value in the positive (+) direction among the cross-correlation coefficients as the causal weight to construct a correlation network between the parenting activity type code and the growth data item. The reason for adopting only the positive (+) direction is that parenting activities with high negative (-) cross-correlation coefficients are interpreted as having a negative effect on the corresponding growth data item and must therefore be excluded from the recommendation target.

[0055] The above development comparison module (160) compares each item of the child's current growth data with the average value of the corresponding item of a group of preceding children at the same developmental stage within the database (120) and identifies items that differ by more than 10 percent in the negative direction as vulnerable developmental areas.

[0056] The above curation generation module (170) is triggered and starts operating as soon as a vulnerable development area is identified by the above development comparison module (160). The above curation generation module (170) extracts parenting activity type codes with a causal weight of 0.3 or higher for growth data items identified as vulnerable development areas in the above correlation network, in descending order of causal weight.

[0057] Subsequently, for each extracted parenting activity type code, a mission template corresponding to the parenting activity type code is retrieved from the mission template storage area within the database (120). The mission template consists of a data record including fields for a mission identifier (mission_id), a parenting activity type code (activity_code), a causal weight (weight), a mission level (level: one of daily, weekly, or monthly), a mission title (title), a mission description (description), a target performance time (target_minutes), a target performance frequency (frequency), and a performance due date (due_date).

[0058] The above curation generation module (170) binds the causal weight of the corresponding parenting activity type code, the current date-based execution deadline, and vulnerable developmental area information to each field of the retrieved mission template, thereby serializing the parenting mission information into an array object of a pre-agreed data serialization format (JSON format in one embodiment).

[0059] The serialized parenting mission information is transmitted to the mission receiving API endpoint of the parent terminal (200) via an encrypted communication channel (POST method of the HTTPS protocol in one embodiment). If the transmission response code is not normal (HTTP 200), the curation generation module (170) attempts to retransmit by applying an exponential backoff algorithm (initial delay 5 seconds, exponential growth rate 2, maximum retries 3 times), and all transmission history is recorded in the log table of the database (120).

[0060] In the present invention, the developmental stages of a child are classified into three stages based on age in months as follows.

[0061] Table 1 is a classification of developmental stages (sample).

[0062]

[0063] As shown in Table 1 above, the early childhood period is the time from birth to just before age 6 (72 months), during which physical growth and basic language and social development primarily take place. The transition period is the time from age 6 to just before age 7 (84 months), during which it exhibits transitional characteristics as one moves from early childhood to school age. The school age is the time from age 7 to just before age 18 (216 months), during which learning ability, emotional regulation, and forming peer relationships are key developmental tasks.

[0064] In the present invention, multimodal growth data is data collected to identify the growth status of a child from various aspects, and consists of a plurality of growth data items including at least two of weight data, height data, total sleep time data, number of syllables spoken data, and number of peer interactions data. Each of the plurality of growth data items has a time series index corresponding to the time of reception. Here, the number of syllables spoken data may include a value extracted by natural language processing of the child's voice data recorded by the caregiver through the microphone of the caregiver terminal (200), or an observation record value directly entered by the caregiver.

[0065] The above multimodal growth data is input through a caregiver terminal (200) in a first cycle once a week and transmitted to a server system (100). In the present invention, the first cycle is set to once a week, and the second cycle is set to once a day. This is intended to minimize the input burden on the caregiver while appropriately tracking changes in the child's growth.

[0066] Meanwhile, if the child is classified as being in the school age group of 84 months or older but less than 216 months, the multimodal growth data may additionally include data on the number of completed learning activities, evaluation score data, and emotional state score data entered by the child through the caregiver terminal, which are received via API from an external learning management system (LMS).

[0067] At this time, the raw data received from the external learning management system undergoes a parsing and normalization process according to a predefined data schema specification, is converted to match the format of the multimodal growth data, and then stored in the database (120).

[0068] In the present invention, behavior log data is a record of parenting activities performed by a caregiver, and includes a parenting activity type code and an interaction duration. The behavior log data is input through a caregiver terminal (200) in a second cycle once a day.

[0069] For each record of the above behavior log data, a first tagging attribute is assigned an activity type (e.g., reading, play, conversation), a second tagging attribute is assigned a stimulus type (e.g., visual, auditory, tactile), and a third tagging attribute is assigned a child response type (e.g., active, passive, unresponsive) according to a pre-stored parenting activity classification table, and the data is stored in a database (120) in a normalized form.

[0070] In the present invention, multimodal growth data collected in the first cycle (once a week) and behavior log data collected in the second cycle (once a day) have different temporal resolutions. Accordingly, in order to analyze the correlation between the two types of data, a process of synchronizing at the same time unit is required.

[0071] Resampling synchronization is performed as follows. First, for the 7-day values ​​of behavior log data collected in the second cycle (daily), the daily average of the number of times each parenting activity type code is performed and the daily average of the total interaction time are calculated, respectively. Then, a scaling operation is performed by multiplying each calculated daily average value by 7 to convert them into weekly total data points.

[0072] The above-described converted weekly total data points are mapped using a Left Outer Join method with the time series index of the above-described multimodal growth data as the primary key. During the joining process, missing values ​​(NULL) corresponding to weeks where behavior log data does not exist are replaced with 0.

[0073] In this invention, the correlation between parenting activities and growth data is derived through cross-correlation analysis.

[0074] Cross-correlation analysis is a method for analyzing correlations that take into account the time lag between two time series data, and it can capture the delayed effect of parenting activities on child development based on temporal precedence.

[0075] In this invention, based on this temporal precedence, the influence relationship between the parenting activity type code and the growth data item is defined as a 'correlation'.

[0076] The cross-correlation coefficient CCF(k) is calculated according to the following mathematical formula 1.

[0077]

[0078] (X t is the weekly average performance frequency of the parenting activity type code at time t, and Y t+k is the measured value of the growth data item after the k interval from time point t, and μ X and μ Y is the average of each time series, and σ X and σ Y is the respective standard deviation, and N is the total number of data points greater than or equal to a preset threshold value that is larger than the maximum value among the plurality of delay intervals.

[0079] That is, μ in the above mathematical formula 1 X , μ Y , σ X , σ Y All are calculated based on the same Nk data pairs valid in the delay interval k, and in this case, since Equation 1 above is equivalent to the Pearson Correlation Coefficient for the corresponding Nk data pairs, the calculated value has a value between -1 and +1.

[0080] In the present invention, the lag interval k is set to 1 week, 2 weeks, and 4 weeks to calculate the respective cross-correlation coefficients. This takes into account that the effects of parenting activities may appear after a short period (1 week), a medium period (2 weeks), or a long period (4 weeks). The setting of the lag intervals is based on the fact that the inventor applied the Akaike Information Criterion (AIC) to 52-week tracking time series data of 1,000 children, and the combination of k=1, 2, and 4 produced the minimum AIC value.

[0081] In this invention, the value with the largest absolute value in the positive (+) direction among the calculated cross-correlation coefficients is determined as the causal weight. Negative (-) cross-correlation coefficients are excluded from the determination of causal weights because they indicate that the corresponding parenting activity may have a negative impact on the growth data item.

[0082] For example, in the case where CCF(1 week)=0.42, CCF(2 weeks)=0.58, and CCF(4 weeks)=0.31 between parenting activity type code A001 (reading a book) and speech syllable count data, the absolute value of 0.58, which is the largest in the positive (+) direction, is determined as the causal weight between the corresponding parenting activity and growth data item.

[0083] In this way, causal weights between all parenting activity type codes and all growth data items are calculated, and the set of these forms a correlation network and is stored in a database (120).

[0084] In the present invention, a vulnerable developmental area is identified when a specific growth data item of the child differs by more than 10 percent in the negative direction compared to the average value of the corresponding item of a preceding group of children belonging to the same developmental stage.

[0085] The above 10 percent criterion was adopted based on the results of a logistic regression analysis using longitudinal cohort data of 2,500 children (followed from birth to 216 months), which confirmed that when the value is 10 percent or less in the negative direction relative to the mean, the probability of diagnosis of developmental delay within the following 12 months increases 3.2 times (p-value < 0.001, Odds Ratio = 3.2, 95% confidence interval).

[0086] Table 2 shows the results of deriving the vulnerability assessment criteria (analysis data: N=2,500, model: logistic regression, cross-validation: 5-Fold CV).

[0087]

[0088] For example, if the number of syllables spoken by a 36-month-old child is 100, and the average number of syllables spoken by preceding children in the infant stage (0-72 months) in the database (120) is 120, the number of syllables spoken by the child is about 16.7% lower than the average, so the language development area is identified as a vulnerable development area.

[0089] In addition, if the weekly change rate of the growth data item identified as the above vulnerable development area remains less than 2 percent during four consecutive measurements, the control unit (101) can generate a notification message regarding the possibility of developmental delay and send it to the caregiver terminal (200).

[0090] The above causal weight threshold of 0.3 was derived as the average lower limit corresponding to the statistical significance (p-value < 0.05) of the cross-correlation coefficient as a result of Monte Carlo simulation (10,000 iterations) using time series data (average 52 weeks follow-up) of 1,000 children by the inventor.

[0091] Table 3 shows the results of deriving the causal weight threshold (analysis data: N=1,000, model: cross-correlation + Bootstrap, confidence interval: 95%).

[0092]

[0093] In the present invention, the parenting mission is information that recommends parenting activities to the caregiver that contribute in a positive (+) direction to the improvement of vulnerable developmental areas. The curation generation module (170) extracts parenting activity type codes with a causal weight of 0.3 or higher for growth data items identified as vulnerable developmental areas in the correlation network, in descending order of causal weight.

[0094] The extracted parenting activity type code is converted into a three-level hierarchical structure of daily detailed missions, weekly intermediate missions, and monthly comprehensive missions. Specifically, for each extracted parenting activity type code, the curation generation module (170) retrieves a mission data record from the mission template storage area in the database (120) that includes fields for a mission identifier, parenting activity type code, causal weight, mission level, mission title, mission description, goal execution time, goal execution frequency, and execution deadline, and serializes the retrieved information into a mission data packet in a pre-agreed data serialization format (JSON format in one embodiment).

[0095] The mission data packet includes a priority score, which is calculated by multiplying the causal weight of the corresponding parenting activity type code by a Z-score-based vulnerability score calculated by dividing the difference from the average value of the vulnerable development area by the standard deviation of the item. The curation generation module (170) selects the top 5 parenting activity type codes based on the priority score to generate a mission data packet, and transmits the generated mission data packet to a caregiver terminal (200) through an encrypted communication channel (in one embodiment, the POST method of the HTTPS protocol).

[0096] In the present invention, the causal weights of the correlation network are continuously updated through a feedback loop based on the results of the parenting mission.

[0097] Specifically, when the server system (100) receives mission execution log data from the caregiver terminal (200), including whether the caregiving mission information has been completed and the duration of execution, the control unit (101) constructs a time series dataset of an updated window size that includes time series data accumulated up to the point in time when a measurement waiting period of 4 weeks has elapsed after receiving the mission execution log data, and recalculates the corresponding causal weight of the correlation network by reapplying the constructed time series dataset to Equation 1, and then updates and stores it in the database (120). Through this feedback loop, the correlation network becomes increasingly accurate by reflecting the actual results of the caregiving mission execution.

[0098] In the present invention, role model data refers to data corresponding to the top 10 percent for a specific growth data item identified as a vulnerable developmental area among data of a preceding group of children at the same developmental stage. The role model data is used as a comparison target with the current child's growth trajectory in a time-series comparison interface.

[0099] In addition, if the growth change amount based on the causal weights recalculated through the feedback loop is 15 percent or more in the positive direction relative to the average value, the control unit (101) generates an anonymized dataset by separating and processing the identification information of the child by applying a k-anonymity algorithm, and automatically stores the anonymized dataset in a role model data storage area provided in the database (120).

[0100] Anonymization processing is performed by applying the k-anonymity (k=5) algorithm. Specifically, in the child's identification information, the date of birth is generalized to the year and quarter of birth, the residential area is generalized to the city / province level, and the gender is kept as is so that there are at least 5 records with the same combination of attributes.

[0101] In this invention, the developmental reliability score is an indicator representing the reliability of the growth tracking and curation service for the child in question, and is calculated as a weighted sum of input sincerity, mission completion rate, and correlation clarity.

[0102] The input sincerity value is calculated as the ratio of the actual number of inputs to the number of scheduled inputs (12 weeks × once a week = 12 times) calculated according to the first cycle during a 12-week calculation period. The mission completion rate value is calculated as the ratio of the number of completed missions to the total number of transmitted parenting mission information. The correlation clarity value is calculated as the ratio of the number of connections with a causal weight of 0.3 or higher in the correlation network to the total number of connections.

[0103] The developmental reliability score is calculated as follows: the value obtained by multiplying the input sincerity value by the first coefficient, the value obtained by multiplying the mission performance rate value by the second coefficient, and the value obtained by multiplying the correlation clarity value by the third coefficient are summed. The first, second, and third coefficients are retrieved from the coefficient weight table for each developmental stage in Table 4 according to the child's developmental stage.

[0104]

[0105] As shown in Table 4 above, since the continuity of data input is important during the early childhood period, the first coefficient (input conscientiousness) is set high at 0.4. Since active participation through mission performance is important during the school-age period, the second coefficient (mission performance rate) is set high at 0.4.

[0106] The transition period reflects the intermediate characteristics of the infancy and school-age period, so the first coefficient and the second coefficient are each set equally to 0.35.

[0107] In the present invention, a growth certificate can be issued by synthesizing a child's growth history and records of parenting activities. The growth certificate data includes a developmental reliability score, a time-series change history derived by the time-series accumulation of the multimodal growth data, and a list of completed missions.

[0108] The control unit (101) structures the growth certificate data into a pre-agreed data serialization format (JSON format in one embodiment), applies a SHA-256 hash function to the structured growth certificate data to calculate a unique hash value, and then generates a digital growth certificate by combining the growth certificate data and the unique hash value into a single object. The generated digital growth certificate is transmitted to a caregiver terminal (200) or a pre-registered external linkage system through an encrypted communication channel.

[0109] The present invention provides a time-series comparison interface that allows a caregiver to intuitively understand the growth status of a child. The control unit (101) generates a time-series graph of the current child by arranging the behavior log data and the multimodal growth data used for synchronization processing in step (e) in parallel with respect to the time axis, and generates a time-series graph of the role model data by extracting the top 10 percent of specific growth data items identified as vulnerable developmental areas among the data of a group of preceding children of the same developmental stage in the database (120).

[0110] Subsequently, comparison interface data is generated by superimposing the time series graph of the current child and the time series graph of the role model data on the same time axis and transmitting it to the caregiver terminal (200). Through this, the caregiver can visually check what level the child's growth trajectory is at compared to the best case.

[0111] Meanwhile, as illustrated in FIG. 1, (a) the server receives the child's identification information and date of birth information from the caregiver terminal and stores it in a database;

[0112] (b) A step in which the server receives multimodal growth data measured at regular intervals from the caregiver terminal and stores it in the database, wherein the multimodal growth data is composed of a plurality of growth data items including at least two of weight data, height data, total sleep time data, number of syllables spoken, and number of peer interactions;

[0113] (c) A step in which the server receives behavior log data including a parenting activity type code and interaction duration from the caregiver terminal at regular intervals and stores it in the database;

[0114] (d) A step in which the control unit calculates the current age of the child based on the date of birth information, and classifies the child into one of the following developmental stages according to the calculated age: an infancy stage of 0 months or more and less than 72 months, a transitional stage of 72 months or more and less than 84 months, or a school age stage of 84 months or more and less than 216 months;

[0115] (e) A step in which the control unit calculates the daily average of the number of times each parenting activity type code is performed and the daily average of the total interaction time for a certain period of the collected behavior log data, respectively, performs a scaling operation by multiplying each calculated daily average value by 7 to convert it into a weekly total data point, and then performs synchronization processing by mapping the converted weekly total data point to the time series index of the multimodal growth data to generate synchronized data;

[0116] (f) The control unit analyzes the correlation between the parenting activity type code and the plurality of growth data items based on the synchronized data, and calculates a cross-correlation coefficient according to Equation 1 for each of the plurality of delay intervals where k is 1 week, 2 weeks, and 4 weeks, and determines the value with the largest absolute value in the positive (+) direction among the calculated cross-correlation coefficients as the causal weight between the corresponding parenting activity type code and the corresponding growth data item, and constructs a correlation network mapping the correlation between the parenting activity type code and the growth data item and stores it in the database;

[0117] (g) A step in which the control unit compares each growth data item of the child’s current multimodal growth data with the average value of the corresponding item of a preceding child group belonging to the same developmental stage within the database, and identifies a growth data item that differs by 10 percent or more in the negative direction from the average value as a vulnerable developmental area; and

[0118] (h) The control unit extracts parenting activity type codes with a causal weight of 0.3 or higher for growth data items of the identified vulnerable developmental area in the correlation network in descending order of causal weight, and for each extracted parenting activity type code, retrieves mission data records including a mission identifier, parenting activity type code, causal weight, mission level, mission title, mission description, goal execution time, goal execution frequency, and execution deadline from a mission template storage area in the database to generate parenting mission information in a pre-agreed data format, and transmits the generated parenting mission information to the caregiver terminal through an encrypted communication channel;

[0119] As illustrated in FIG. 2, the control unit (101) of the server system (100) can sequentially perform data management and preprocessing processes to integrally manage collected raw data and process it into an analyzable form.

[0120] The data collection module (130) of the above server system (100) can collect behavior log data accumulated on a daily basis over a certain period from the caregiver terminal (200) and safely store it in the database (120).

[0121] At the same time, the data collection module (130) can collect multimodal growth data measured on a weekly basis from the caregiver terminal (200) and store it together in the database (120).

[0122] The preprocessing module (140) within the control unit (101) can take over the collected behavior log data and initiate a data refinement operation to divide it according to parenting activity type codes that serve as criteria for statistical analysis.

[0123] The above preprocessing module (140) can perform a detailed parsing step to extract the specific number of times a particular activity is performed or to extract the total interaction time during which the interaction is maintained within the data structure divided by the parenting activity type code.

[0124] After the step of extracting the number of executions or the total interaction time, the control unit (101) can control the average calculation and scaling steps for matching the time series resolution overall.

[0125] The above preprocessing module (140) can calculate the daily average of the number of times each parenting activity type code is performed or the daily average total interaction time based on the extracted raw data to reduce noise in the short-term data and obtain a stable representative value.

[0126] The control unit (101) can control the preprocessing module (140) to perform a scaling operation of multiplying each calculated daily average value by a predetermined constant in order to match the periodicity of the behavior data collected on a daily basis with the periodicity of the multimodal growth data measured on a weekly basis.

[0127] As shown in FIG. 2, the scaling operation can be performed by multiplying the daily average value by 7, which corresponds to the number of days in a week, to derive a cumulative estimate for one week.

[0128] The control unit (101) can perform a process of converting the data that has undergone the scaling operation into weekly total data points as a final step for time series synchronization.

[0129] The above preprocessing module (140) can obtain the converted weekly total amount data points and at the same time extract the unique time series index of the multimodal growth data from the database (120).

[0130] The above preprocessing module (140) can perform a synchronization mapping operation to mutually map the converted weekly total amount data points and the time series index of the extracted multimodal growth data based on the time axis.

[0131] Through the above mapping process, the control unit (101) can finally generate synchronized data in which the behavior log and growth indicator are accurately combined according to the temporal sequence.

[0132] The synchronized data generated above is updated and stored in the database (120) and can be used as core basic data for the correlation analysis module (150) or the development comparison module (160) to precisely analyze the time delay effect of the data.

[0133] Additionally, the curation generation module (170) can provide highly reliable personalized parenting missions to the caregiver terminal (200) through the communication module (110) based on the results of the time-series synchronized data analysis.

[0134] Therefore, the above server system (100) can achieve a technical effect that enables more objective and scientific correlation analysis by mathematically synchronizing heterogeneous modality data, which are difficult to compare directly due to different collection cycles, so that they have the same time series resolution.

[0135] As illustrated in FIG. 3, the server system (100) comprehensively implements a core data processing and computation infrastructure for providing child growth optimization curation services.

[0136] The above server system (100) is connected to the caregiver terminal (200) via a wired or wireless communication network and provides an environment in which data can be exchanged between them in real-time or non-real-time.

[0137] At the forefront of the above server system (100), a communication module (110) is positioned to handle connection with an external network and to form a secure channel to safely transmit and receive data with the caregiver terminal (200).

[0138] The identification information of the child, the behavior log of the caregiver, or various quantified measurement data received through the communication module (110) are immediately transmitted to the data collection module (130).

[0139] The data collection module (130) primarily collects raw data in an unstructured or semi-structured form received from the communication module (110) and temporarily stores it in a form that can be processed within the system.

[0140] The database (120) is a large-capacity storage that permanently stores and manages data collected from the data collection module (130) and can be constructed by dividing it into multiple logical partitions.

[0141] As illustrated in FIG. 3, the database (120) may be configured to include a first area for storing child information and growth data, a second area for recording behavior logs and causal data, or a third area for storing mission templates and related data for providing customized guides.

[0142] The above server system (100) is equipped with a control unit (101) that comprehensively directs the actual computation and processing of the collected data and the organic data pipeline flow between each sub-module.

[0143] In order to perform complex causal analysis and curation algorithms independently and in parallel, a preprocessing module (140), a correlation analysis module (150), a development comparison module (160), or a curation generation module (170) is mounted in the form of logical blocks within the control unit (101).

[0144] The preprocessing module (140) of the control unit (101) receives basic data from the data collection module (130) or calls historical data from the database (120) to perform a refinement operation that corrects missing values ​​and standardizes the specifications of the data.

[0145] In particular, the preprocessing module (140) intensively processes time-series synchronization operations to place the multimodal growth data and the behavior log data, which are input with different cycles, on the same time axis, and updates and stores the results in the database (120).

[0146] The above development comparison module (160) performs an operation to compare the development indicator data of a specific child, which has been refined through the above preprocessing module (140), with the statistical average data of a preceding child group accumulated in the above database (120).

[0147] The above development comparison module (160) derives a quantitative value for a vulnerable development area that falls below a standard value among the physical, cognitive, or emotional development items of the specific child based on the above comparison operation, and parameterizes it so that other modules within the control unit (101) can utilize it.

[0148] At the same time, the correlation analysis module (150) tracks the time-delay effect of a specific parenting activity on changes in a child's development indicators based on time-series data synchronized from the preprocessing module (140).

[0149] The correlation analysis module (150) calculates causal weights that quantify the correlation between the parenting activity and the development indicator by applying a sophisticated mathematical cross-correlation function, and structures and loads them into the second area of ​​the database (120).

[0150] The curation generation module (170), which is responsible for the lowest data flow of the control unit (101), finally fuses the vulnerable development area parameter received from the development comparison module (160) with the causal weight data derived by the correlation analysis module (150).

[0151] The above curation generation module (170) identifies a nurturing activity that can most efficiently improve the vulnerable development area based on the fused analysis data, and then queries the third area of ​​the database (120) to extract a specific original mission template mapped to the activity.

[0152] The above curation generation module (170) completes a curation operation to generate final parenting mission information by dynamically combining a difficulty level, execution deadline, or number of times that correspond to the developmental level of the target child with the extracted original mission template.

[0153] The control unit (101) controls the entire system to convert the parenting mission information completed by the curation generation module (170) into a serialized data packet and transmit it to the parent terminal (200) through the communication module (110).

[0154] Therefore, the server system (100) of the present invention illustrated in FIG. 3 can provide an automated child growth solution by seamlessly linking complex computing processes—such as precise preprocessing, causality analysis, objective developmental comparison, or customized mission combination—within a single architecture, going beyond simple data input and output.

[0155] FIG. 4 includes the step of (f) the control unit analyzing the correlation between the parenting activity type code and the plurality of growth data items based on the synchronized data, calculating the cross-correlation coefficient according to Equation 1 for each of the plurality of delay intervals where k is 1 week, 2 weeks, and 4 weeks, determining the value with the largest absolute value in the positive (+) direction among the calculated cross-correlation coefficients as the causal weight between the corresponding parenting activity type code and the corresponding growth data item, constructing a correlation network that maps the correlation between the parenting activity type code and the growth data item and storing it in the database;

[0156] As illustrated in FIG. 5, the server system (100) can collect and analyze multimodal data of the target child and provide a visualized interface to the caregiver terminal (200).

[0157] The control unit (101) of the above server system (100) can control the communication module (110) to monitor in real time whether the terminal communication with the caregiver terminal (200) is normal.

[0158] The control unit (101) can continuously check the synchronization status of the database (120) and control the terminal communication normal status or the database synchronization completion status to be displayed in the upper area of ​​the interface.

[0159] The data collection module (130) of the above server system (100) can collect identification information or a unique identification code of the target child from the caregiver terminal (200) and safely store it in the database (120).

[0160] The identification information of the above-mentioned target child may be configured to include the name Kim Ji-hoon and an identification code such as USR 2023 891, as shown in FIG. 5.

[0161] The preprocessing module (140) can quantitatively calculate the accurate age in months at the current time based on the date of birth information of the target child collected through the data collection module (130).

[0162] As illustrated in FIG. 5, the preprocessing module (140) can process the target child to be classified as an infant or toddler when the calculated age is forty-six months and visually displayed on the interface.

[0163] The control unit (101) can control each internal component to perform an overall multimodal growth analysis based on the multimodal growth data of the target child collected through the data collection module (130).

[0164] The development comparison module (160) can extract average data of a preceding group of children stored in the database (120) and perform a comparative analysis with the multimodal growth data of the target child.

[0165] The above multimodal growth data may consist of multiple growth data items, including height data, weight data, sleep time data, number of syllables spoken, or number of peer interactions.

[0166] The above development comparison module (160) can intuitively provide the caregiver with the child's overall developmental status by generating a radial chart that maps the plurality of growth data items to each axis.

[0167] As shown in Fig. 5, the radial chart has five axes extending from a center point and can display the average data of the preceding child group as a first polygonal area.

[0168] The above radial chart can clearly show the results of mutual comparison by visualizing the actual measurement values ​​of the subject child as a second polygonal area that overlaps with the first polygonal area.

[0169] According to the visual analysis results of the above development comparison module (160), the subject child may form a numerical distribution similar to the average of the preceding child group in terms of weight or sleep time.

[0170] On the other hand, the analysis results of the above development comparison module (160) can objectively be derived that the subject child shows a relatively lower distribution compared to the average of the preceding child group in terms of height, number of syllables spoken, or number of peer interactions.

[0171] The correlation analysis module (150) can be used as basic data for in-depth analysis of the association between the plurality of growth data items or the correlation with specific parenting activities based on the comparative analysis results derived from the development comparison module (160).

[0172] The curation generation module (170) can immediately initiate a process of generating customized parenting mission information when an item that differs by more than a predetermined percentage from the average value of the preceding child group is automatically identified as a vulnerable developmental area by the development comparison module (160).

[0173] The above curation generation module (170) can extract an appropriate mission template from the mission template storage area within the database (120) to supplement the deficient developmental area of ​​the target child, process it, and transmit it to the caregiver terminal (200).

[0174] In this way, the control unit (101) can manage the organic data flow between the communication module (110), the database (120), the data collection module (130), the preprocessing module (140), the correlation analysis module (150), the development comparison module (160), or the curation generation module (170) within the server system (100) to derive multifaceted and objective growth analysis results for the target child.

[0175] Therefore, through the aforementioned multimodal growth analysis screen, in which various indicators are comprehensively synchronized rather than fragmentary information, caregivers can clearly recognize the child's current developmental stage and efficiently utilize it as basic judgment data to carry out future parenting missions.

[0176] As illustrated in FIG. 6, the server system (100) can analyze the multimodal data of the target child from various angles and provide a user interface capable of intuitive interaction to the caregiver terminal (200).

[0177] The control unit (101) of the above server system (100) can control the communication module (110) to output a multimodal growth analysis screen generated based on the calculation result of the development comparison module (160) to the display of the caregiver terminal (200).

[0178] A radial chart displaying the developmental indicators of the target child and the average indicators of the preceding child group simultaneously may be placed in the center of the above interface.

[0179] The control unit (101) can provide an information window containing detailed numerical information in the form of a popup on one side of the screen in response to the detection of a designated input event, such as a user's touch or mouse over, on a specific development indicator area constituting the radial chart, for example, a sleep time axis.

[0180] In the information window above, the name of the indicator, sleep time, is displayed alongside the standard value of the peer average group, 10.5 hours, and the actual value of the target child, 10.2 hours, arranged vertically to help the user accurately identify minute deviations between the two data points.

[0181] In addition to the detailed verification function of these individual indicators, the development comparison module (160) can internally perform an automated identification algorithm to select vulnerable areas requiring urgent management among the overall development indicators of the target child by utilizing big data accumulated in the database (120).

[0182] Specifically, the development comparison module (160) can continuously track indicators that show a difference of more than 10 percent in the negative direction relative to the average by comparing the numerical value of each item of the multimodal growth data of the target child with the numerical value of the average group of peers.

[0183] As shown in FIG. 6, if there is an item calculated to have delayed development exceeding the 10 percent standard, the control unit (101) can conspicuously display a warning interface titled "Automatic identification of vulnerable development area" in an area adjacent to the radial chart.

[0184] Within the above warning interface, detailed indicators finally determined as vulnerable areas by the above development comparison module (160) may be listed as individual rows.

[0185] For example, the development comparison module (160) can compare the target child's current figure of 210 and the average figure of 350 for the number of syllables spoken by the target child to calculate a quantitative difference rate of 40% and render this as the first identification item of the warning interface.

[0186] In addition, the above development comparison module (160) can calculate the difference rate by comparing the target child's current value of three times and the average value of eight times for the peer interaction frequency item, and then display this as a second identification item.

[0187] Raw data collected through the data collection module (130) and refined through the preprocessing module (140) can be converted into intuitive visual alert indicators that alert the caregiver through this deep computational process of the development comparison module (160).

[0188] Subsequently, the correlation analysis module (150) can perform a subsequent mapping analysis to determine which parenting activity has the highest causal weight to compensate for the identified percentage of speech syllable shortages or peer interaction shortages.

[0189] Finally, the curation generation module (170) can complete a series of processes to provide to the caregiver terminal (200) a personalized parenting mission to raise the development speed of the target child to a normal range based on the derived vulnerable developmental area and the difference rate data.

[0190] Therefore, the above server system (100) can create a technical effect that reduces the burden of data interpretation on the caregiver and induces timely developmental intervention by providing not only intuitive diagramming but also an integrated automatic vulnerability diagnosis function based on statistical criteria.

[0191] As illustrated in FIG. 7, the server system (100) can derive optimal parenting activities that contribute to the improvement of specific development indicators through mathematical modeling that considers time delay effects and provide them to the parent terminal (200).

[0192] The data collection module (130) of the above server system (100) can collect multimodal growth data measured on a weekly basis and behavior log data input on a daily basis, respectively, and store them in a database (120).

[0193] The preprocessing module (140) can apply a scaling operation to analyze the data having different collection cycles and perform a time series synchronization process that aligns them based on the same time axis.

[0194] The control unit (101) of the server system (100) can control the time series data synchronized by the preprocessing module (140) to be rendered in the form of a line graph at the top of the screen of the caregiver terminal (200) through the communication module (110).

[0195] The above line graph shows the weekly trend of change from a past point in time to the present, and can visualize the total amount of nurturing interactions and target growth indicators, such as the number of syllables spoken, as parallel linear data.

[0196] The control unit (101) can provide an information window containing detailed data of a specific point in time on the line graph when a user touch input is detected, for example, a point in time two weeks ago.

[0197] The information window displays the total number of parenting interactions for the corresponding week, eighteen, and the measured number of uttered syllables, one hundred and eighty, side by side, enabling the caregiver to intuitively check the time-series data trends.

[0198] The correlation analysis module (150) can calculate a cross-correlation function according to mathematical formulas to quantify the causal effect of the nurturing interaction on the growth indicator based on the synchronized time series data.

[0199] The above correlation analysis module (150) can calculate each cross-correlation coefficient by setting multiple delay intervals, such as one week, two weeks, or four weeks, to reflect the characteristic that the effect of parenting activities does not appear immediately but manifests after a certain period.

[0200] The correlation analysis module (150) can determine the value with the largest absolute value in the positive direction among the calculated cross-correlation coefficients as the final causal weight between a specific parenting activity type code and the corresponding growth data item.

[0201] The control unit (101) can convert and provide the causal weight for each parenting activity type code determined by the correlation analysis module (150) into a bar graph at the bottom of the screen of the parent terminal (200).

[0202] At the bottom of the screen above, multiple parenting activity type codes targeting improvement of the number of uttered syllables identified as vulnerable developmental areas by the development comparison module (160) can be arranged in descending order of causal weight size.

[0203] As shown in FIG. 7, the plurality of parenting activity type codes may be configured to include reading picture books, role-playing, singing nursery rhymes, or playing with blocks.

[0204] The above control unit (101) may apply a principle of extracting only parenting activity type codes with a calculated causal weight of at least zero point three as meaningful data in order to ensure the reliability of the curation.

[0205] The above control unit (101) can display a bar graph of picture book reading, role play, or nursery rhyme singing that satisfies the threshold value according to the extracted principle in an active state by highlighting it with a first color.

[0206] On the other hand, the control unit (101) can visually distinguish that the block play bar graph, which records a causal weight zero point one-five that is less than the threshold value, is in a deactivated state by processing it with a second color.

[0207] In addition, the control unit (101) may provide the delayed week information of the corresponding activity confirmed through the cross-correlation analysis along with the causal weight value of each parenting activity type code on the right side of the bar graph.

[0208] For example, in the case of reading picture books, text such as "delayed migration" is displayed along with a causal weight of 0.72, clearly conveying that continuing the activity has the greatest impact on improving the number of syllables spoken after migration.

[0209] The curation generation module (170) can combine personalized parenting missions by prioritizing parenting activity type codes verified above the threshold by the correlation analysis module (150) and transmit them to the parenter terminal (200).

[0210] Therefore, the above server system (100) can derive a scientific parenting guide based on causal grounds rather than simple statistics by synchronizing data with different time resolutions and performing cross-correlation analysis based on time delay.

[0211] As illustrated in FIG. 8, the control unit (101) of the server system (100) can render parenting mission information containing specific execution instructions in text and numerical data based on the meaningful parenting activity type code derived in the previous step and provide it to the parent terminal (200).

[0212] The above control unit (101) can control the curation creation module (170) to query the mission template storage area, which is logically divided and built in advance within the database (120), through a database query.

[0213] The above curation generation module (170) can extract original mission template data corresponding to the parenting activity type code determined to have the highest causal weight based on the query results from the above database (120).

[0214] The above curation generation module (170) can generate detailed parenting mission information by mapping dynamic variables of the target execution time, target execution frequency, or execution deadline based on the extracted original mission template data to match the child's current developmental level.

[0215] The control unit (101) can complete a process of converting the detailed parenting mission information into a pre-agreed data serialization format, such as a JSON file format, so that the curation generation module (170) can smoothly transmit and receive the detailed parenting mission information to and from the parent terminal (200).

[0216] As shown in FIG. 8, the nurturing mission information generated in the data serialization format can be visualized and displayed as a plurality of independent mission card layouts on the user interface.

[0217] The first mission card placed on the left among the above parenting mission information is equipped with a parenting activity type code and may have a level of difficulty and a unique identification reference.

[0218] The above-mentioned first mission card may display the intuitive mission title "Reading Animal Picture Books Aloud" in bold at the top and provide detailed practice guide text at the bottom that encourages exaggerated expression of onomatopoeia or mimetic words while looking at animal picture books with a child.

[0219] In addition, the first mission card can quantify the one-time goal performance time of the activity to 15 minutes and set the weekly goal performance frequency to 4 times a week.

[0220] The second mission card placed in parallel on the right among the above parenting mission information is equipped with a different parenting activity type code and may have a relatively low difficulty level of Level 1 and a unique identification reference.

[0221] The above-mentioned second mission card can promote the simultaneous improvement of language development and social skills by providing play-based detailed mission description text that sets a doll as a patient under the mission title of "Doll Hospital Play" and encourages the child to become a doctor and speak to it on their own.

[0222] The above second mission card can set the target performance time to twenty minutes and the target performance frequency to twice a week, reflecting the characteristics of the game, and can display the performance deadline calculated in the same way as the above first mission card on the display.

[0223] The control unit (101) can wait for an explicit approval command from a caregiver or manager by creating a trigger button interface for executing an encrypted channel transmission in the upper area of ​​the rendered mission cards.

[0224] The control unit (101) can control the communication module (110) to activate an encrypted communication channel with an HTTPS protocol applied when a click or touch event for the trigger button is received.

[0225] The control unit (101) can fundamentally prevent the leakage of personal development data by securely transmitting a parenting mission information packet in the data serialization format to the endpoint of the parent terminal (200) via the activated encrypted communication channel.

[0226] Therefore, the above server system (100) can induce the actual optimization of a child's development by converting the complex correlation analysis results into parenting missions in the form of specific behavioral guides that can be immediately practiced in daily life without the need for the caregiver to interpret them.

[0227] In one embodiment of the present invention, the control unit (101) of the server system (100) may further include a growth trajectory prediction model based on Long Short-Term Memory (LSTM).

[0228] The above growth trajectory prediction model receives the above synchronized data and outputs predicted values ​​for each growth data item for the next 4 to 12 weeks, and the above development comparison module (160) is used to preemptively identify vulnerable development areas.

[0229] Training data:

[0230] The training data used for training the growth trajectory prediction model consists of time series pairs of children’s multimodal growth data and synchronized behavior log data accumulated in the database (120). Each training sample consists of a sequence of inputs of 24 consecutive weeks (24 time series points) and a pair of actual value labels for each growth data item of the following 4 weeks (4 time series points).

[0231] Each time series point of the above input sequence is composed of a multivariate vector containing five growth data item values ​​of weight, height, total sleep time, number of syllables spoken, and number of peer interactions, and a weekly total amount data point by parenting activity type code. The learning data is constructed by extracting cases with continuous records of 24 weeks or more from a group of children in the database (120), and is stratified by developmental stage and included in the ratio of 60% for infants and toddlers, 10% for the transitional stage, and 30% for the school-age stage.

[0232] Data Preprocessing:

[0233] Preprocessing of the above training data is performed in the following order. First, Z-score normalization is performed on each growth data item using the mean and standard deviation of a group of children at the same developmental stage. Linear interpolation is applied first to missing values ​​(NULL), and for intervals of 3 weeks or more of consecutive missing values, they are replaced with the group mean value of the corresponding developmental stage. The training samples are augmented by partitioning the normalized and missing-value-processed multivariate time series using a sliding window technique.

[0234] Learning Model:

[0235] The above growth trajectory prediction model consists of a two-layer bidirectional LSTM structure. The first LSTM layer has 128 hidden units, and the second LSTM layer has 64 hidden units, with a dropout rate of 0.3 applied between each layer. The output of the LSTM layer is connected to a fully connected layer (number of output units: number of growth data items × prediction period) to output predicted values ​​for each item at future time points. The predicted values ​​are restored to numerical values ​​in the original unit system through an inverse Z-score transform.

[0236] FIG. 9 is an example of a user interface for a child monitoring dashboard that allows for intuitive understanding of the child's overall developmental status according to one embodiment of the present invention.

[0237] Referring to FIG. 9, the control unit (101) of the server system (100) can fuse and compute the collected multimodal growth data and behavior log data to render a child monitoring dashboard screen on the display of the caregiver terminal (200).

[0238] In the top area of ​​the dashboard screen, key performance indicators (KPIs) representing the current developmental status of a child (e.g., Kim Ji-hoon) are arranged in parallel in the form of multiple summary cards. Specifically, the summary cards include a physical growth score (e.g., 94 / 100), average sleep time (e.g., 11.5 hours), the number of syllables spoken (e.g., 320 syllables), which is a language development level extracted through natural language processing, and caregiver interaction time based on behavioral log data (e.g., 110 minutes / day).

[0239] The above control unit (101) displays the rate of increase / decrease compared to the previous month (e.g., +2.4%) or the absolute value of increase / decrease (e.g., -0.2 hours, +15 minutes) at the bottom along with the current value of each key indicator, thereby supporting the caregiver to immediately recognize the trend of short-term developmental changes.

[0240] In the lower left area of ​​the dashboard screen above, a 'multimodal growth trend' interface is provided that visualizes the mid-to-long-term changes of specific growth indicators. The control unit (101) maps height (cm) and weight (kg) data collected periodically over the past six months (e.g., October to March) onto the same X-axis based on a time series index and renders each indicator as a multi-line graph with different identification colors (e.g., height is light blue, weight is purple).

[0241] Additionally, when the control unit (101) detects a user input event for a specific point in time on the graph (e.g., January), it pops up the exact measurement value for that point in time (e.g., height 93.8 cm, weight 14.1 kg) in the form of a tooltip to improve accessibility to detailed data.

[0242] In the lower right area of ​​the dashboard screen above, a 'developmental balance' interface for evaluating the child's overall developmental balance is provided in the form of a radar chart. The developmental comparison module (160) sets five core developmental areas—language, social, motor, cognitive, and emotional—as axes, and visualizes the indicator values ​​for this month and last month as two overlapping polygonal areas.

[0243] Through this, caregivers can visually confirm the overall expansion of development this month compared to the previous month, while simultaneously intuitively comparing and analyzing notable growth trends in specific areas (e.g., language) or vulnerable areas requiring improvement.

[0244] FIG. 10 is an example of a user interface providing curation results based on caregiver behavior log analysis and AI causal correlation according to one embodiment of the present invention.

[0245] Referring to FIG. 10, the control unit (101) of the server system (100) renders a 'caregiver behavior log' area based on data synchronized through a preprocessing module (140) at the top left of the screen of the caregiver terminal (200). The control unit (101) plots the interaction time by collected childcare activity type as a time-series graph, and provides it in the form of a multi-area graph (Area Chart) by mapping activity type codes such as reading (purple), physical play (blue), and music / rhythm (green) to different identification colors. In addition, when a user input event for a specific point in time (e.g., January) on the graph is detected, the detailed interaction time for each childcare activity at that point in time (e.g., reading 30 minutes, physical play 25 minutes, music / rhythm 18 minutes) is displayed as a popup in the form of a tooltip to improve the readability of the data.

[0246] At the top right of the screen, an 'activity distribution' area is provided in the form of a donut chart to intuitively display the ratio of childcare activities for the current month. The control unit (101) collects behavior log data entered during a certain period (e.g., this month) to calculate the proportion of each childcare activity, such as reading, physical play, music / rhythm, and art / creation, within the total activities, and visually displays this in segments to support the caregiver in checking whether balanced interactions are being performed without being biased toward specific activities.

[0247] At the bottom of the screen, an 'AI Causal Correlation Analysis and Curation' area, which visualizes the combined computational results of the correlation analysis module (150) and the curation generation module (170), is arranged in parallel in the form of multiple insight cards. The insight cards include specific implementation guidelines derived by analyzing the causal impact of parenting activities on a child's development indicators.

[0248] The first card is an area that guides positive developmental changes and includes a mission title of 'rapid growth in language development' and a 'high correlation' status tag. The correlation analysis module (150) calculates and displays a high causal correlation score (e.g., 0.89) between the increase in 'reading' interaction time and the surge in 'number of syllables spoken' as text, and the curation generation module (170) provides an advanced recommendation solution based on this to add open-ended questions while maintaining the current routine.

[0249] The second card is an area warning of vulnerable developmental areas or negative time-delay effects, and includes the title 'irregular sleep patterns' and a 'caution needed' status tag. The control unit (101) derives the effect of a specific activity (e.g., physical play in the late afternoon) on the reduction of sleep time among multiple growth data items through cross-correlation analysis, and presents a solution that recommends adjusting the activity time (e.g., completing before 4 p.m.) and switching to static activities (music, reading) as a correction guide.

[0250] The third card is an area for monitoring stabilized developmental areas, displaying the title 'Social Indicator Stabilization' and the 'Observation' status tag. The developmental comparison module (160) specifies the result of comparison with the average data of a preceding group of children (e.g., reaching the top 20% of peer average), and the curation generation module (170) provides the next stage of personalized parenting missions that encourage increasing the frequency of group play activities from once a week to twice a week to deepen the corresponding developmental level. Explanation of the symbols

[0251] 100: Server System 101: Control unit 110: Communication module 120: Database 130: Data Collection Module 140: Preprocessing Module 150: Correlation Analysis Module 160: Development Comparison Module 170: Curation creation module 200: Caregiver terminal

Claims

Claim 1 A method for providing a child growth optimization curation service performed on a server including a control unit and communicating with a caregiver terminal, comprising: (a) a step in which the server receives child identification information and date of birth information from the caregiver terminal and stores them in a database; (b) a step in which the server receives multimodal growth data measured at regular intervals from the caregiver terminal and stores it in the database, wherein the multimodal growth data includes at least two of weight data, height data, total sleep time data, number of syllables spoken, and number of peer interactions; (c) a step in which the server receives behavior log data including a caregiver activity type code and interaction duration at regular intervals from the caregiver terminal and stores it in the database; and (d) a step in which the server refers to the database and provides a child growth optimization curation service based on a plurality of growth data items and behavior log data according to the multimodal growth data; and (e) a step in which, after step (c), the control unit calculates the current age of the child in months based on the date of birth information and classifies the child into one of a pre-set developmental stage according to the calculated age in months.and (f) the control unit calculates the daily average of the number of times each parenting activity type code is performed and the daily average of the total interaction time for a certain period of the collected behavior log data, respectively, performs a scaling operation on the calculated daily average values ​​to convert them into weekly total data points, and then maps the converted weekly total data points to the time series index of the multimodal growth data to generate synchronized data; and (f) after the step, (g) the control unit calculates the cross-correlation coefficient between the parenting activity type code and the plurality of growth data items based on the synchronized data, and among the cross-correlation coefficients CCF(k) calculated according to Equation 1, the value that indicates a positive correlation and whose magnitude is greater than or equal to a preset threshold is determined as the weight between the corresponding parenting activity type code and the corresponding growth data item and stored in the database; further comprising the method of providing a child growth optimization curation service using relationship analysis of multimodal data by developmental stage and caregiver behavior logs. [Equation 1]; (X t is the weekly average performance frequency of the parenting activity type code at time t, and Y t+k is the measured value of the growth data item after the k interval from time point t, and μ X and μ Y is the average of each time series, and σ X and σ Y is the respective standard deviation, and N is the total number of data points greater than or equal to a preset threshold value that is larger than the maximum value among the plurality of delay intervals. Claim 2 delete Claim 3 delete Claim 4 A method for providing a child growth optimization curation service using relationship analysis between developmental stage multimodal data and a caregiver behavior log, characterized in that, in claim 1, after step (g), (h) the control unit compares each growth data item of the child's current multimodal growth data with the average value of the corresponding item of a preceding child group belonging to the same developmental stage within the database, and identifies a growth data item that differs by more than a preset ratio in the negative direction relative to the average value as a vulnerable developmental area; and (i) for the growth data item identified as a vulnerable developmental area, the control unit extracts a parenting activity type code in descending order of the weight, wherein the weight is greater than or equal to a preset threshold, and for each extracted parenting activity type code, retrieves a mission data record including a mission identifier, parenting activity type code, weight, mission level, mission title, mission description, goal execution time, goal execution frequency, and execution deadline from the mission template storage area of ​​the database to generate parenting mission information, and transmits the generated parenting mission information to the caregiver terminal through a secure communication channel. Claim 5 A method for providing a child growth optimization curation service using relationship analysis of developmental stage multimodal data and caregiver behavior logs, wherein, after step (f), the control unit generates a time series graph of the current child by arranging the behavior log data and the multimodal growth data used for synchronization processing in step (f) in parallel with respect to a time axis; (g2) the control unit generates a time series graph of the role model data by extracting role model data corresponding to a preset standard or higher from the data of a group of preceding children of the same developmental stage within the database; and (g3) the control unit generates comparison interface data in which the time series graph of the current child and the time series graph of the role model data are superimposed on the same time axis and transmits it to the caregiver terminal. Claim 6 A method for providing a child growth optimization curation service using relationship analysis between multimodal data of developmental stages and caregiver behavior logs, characterized in that, in step (h) above, if the weekly change rate of the growth data item identified as the vulnerable developmental area is maintained below a preset rate during a consecutive preset number of measurements, the control unit generates a notification message regarding the possibility of developmental delay and transmits it to the caregiver terminal. Claim 7 A method for providing a child growth optimization curation service using relationship analysis of multimodal data by developmental stage and caregiver behavior logs, wherein, in claim 4, the above step (i) comprises: a step of calculating a priority score for each of the extracted parenting activity type codes based on the difference between the corresponding weight and the average value of the vulnerable developmental area; a step of selecting the top 5 parenting activity type codes based on the priority score; and a step of converting the selected parenting activity type codes into a three-stage hierarchical structure of daily detailed missions, weekly intermediate missions, and monthly comprehensive missions. Claim 8 A system using a method for providing a child growth optimization curation service using the relationship analysis between multimodal data of developmental stages and caregiver behavior logs according to claim 1, comprising: a control unit; and a communication module connected to the control unit and transmitting and receiving data with the caregiver terminal. A database storing child identification information, date of birth information, multimodal growth data, caregiver behavior log data, relationship analysis data, role model data, and mission templates; wherein the control unit comprises: a data collection module configured to store child identification information, date of birth information, multimodal growth data, and behavior log data received through the communication module in the database; a preprocessing module configured to calculate the child's current age in months based on the date of birth information and classify it into one of the preset developmental stages, calculate the daily average of the number of times performed by each parenting activity type code and the daily average of the total interaction time for a certain period value of the behavior log data collected at a certain period, perform a scaling operation on each calculated daily average value to convert it into weekly total data points, and then map it to the time series index of the multimodal growth data to perform synchronization processing; and a correlation analysis module configured to calculate the cross-correlation coefficient between the parenting activity type code and the growth data item for the synchronized data, and to generate correlation analysis data by determining as a weight a value among the cross-correlation coefficients that shows a positive correlation and whose magnitude is greater than or equal to a preset threshold. A developmental comparison module configured to identify items lower than a preset ratio as vulnerable developmental areas by comparing each item of a child's current growth data with the average value of the corresponding item of a group of preceding children at the same developmental stage within the aforementioned database;A child growth optimization curation service provision system utilizing relationship analysis between developmental stage multimodal data and caregiver behavior logs, characterized by including: a curation generation module configured to extract codes in descending order among the parenting activity type codes corresponding to growth data items identified as vulnerable developmental areas, wherein the weight is greater than or equal to a preset threshold, and for each extracted parenting activity type code, query a corresponding mission template from the database to generate parenting mission information, and then transmit it to the caregiver terminal through the communication module.

Citation Information

Patent Citations

  • Child-rearing support system and program

    JP2024094201A

  • Apparatus and method for analyzing child development

    KR1020210026999A

  • A system to provide child-rearing information for each developmental stage based on big data

    KR1020230108813A

  • System for providing child care plan consulting information and method thereof

    KR102185313B1

  • System and method for providing developmental indicator inference and assessment based on infant behavior to teachers and parents

    KR102912962B1