Method and system for analyzing dietary behavior associated data of autistic patient

Abstract

The invention provides an autistic patient dietary behavior associated data analysis method and system, relates to the technical field of data analysis, and aims to effectively identify potential trigger events and establish a behavior stability benchmark by integrating multi-dimensional time sequence data, establishing a behavior stability benchmark and performing multi-dimensional difference comparison during behavior deviation. And potential triggering factors inconsistent with the safety mode are identified when the behavior deviates, so that the accuracy and the reliability of analysis are improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of data analysis, in particular, to a method and system for analyzing correlation data of dietary behaviors of autism patients. BACKGROUND

[0002] In a typical family unit, there is a child with autism spectrum disorder (ASD) and other family members. In order to alleviate certain specific behaviors of the child, the family may decide to implement a strict dietary intervention program, such as gluten-free, casein-free diet. For this purpose, the family usually introduces a data recording system, and the main caregiver is responsible for recording the types and amounts of food ingested by the child every day, and observing and recording the behavior of the child at fixed time points, such as emotional stability, frequency of repetitive behavior, willingness of social interaction, etc. The expectation of the family is to find the corresponding relationship between food and behavior by analyzing these records, so as to optimize the dietary program.

[0003] However, in actual application, after a period of data accumulation, the two-dimensional data analysis results based on food intake and behavior performance often show irregular and even contradictory phenomena. For example, the data in a certain period of time shows that there is a positive correlation between the intake of a certain food and the emotional stability of the child; but in another period of time, the child has behavior problems under the same condition of ingesting the food. This inconsistency makes the family doubt the effectiveness of the dietary intervention, and brings great confusion and pressure to the caregiver. The root of the problem is not in the recorded data itself, but in the implicit variables that the existing data recording system fails to capture in the family's common living environment.

[0004] These implicit variables may come from many aspects. For example, in the process of joint meal preparation, in order to improve efficiency, the caregiver may use the same set of kitchen utensils and operating table in the same kitchen space to process the special diet of the ASD child and the regular diet of the family at the same time. Even after cleaning, trace amounts of contaminants (such as gluten protein) may still be transferred to the child's diet that should be "safe" through the kitchen utensils or table. The amount of such trace contaminants may be below the regular detection standard and below the level that the caregiver can be aware of, so in the dietary log, the child's intake is accurately recorded as "safe food". However, for ASD children with highly sensitive digestive and immune systems, such trace intake is enough to trigger adverse physiological reactions in their bodies and manifest as behavior problems after a few hours. The information received by the data analysis system is that "safe food" corresponds to "behavior problems", thus leading to a wrong causal direction.

[0005] Moreover, the emotional state of the caregiver can also constitute a hidden variable. It is a heavy and stressful task to manage the special diet of an ASD child while also catering to the dietary needs of the entire family. The caregiver can be in a state of high tension or anxiety while preparing or eating meals, worrying about cross-contamination, worrying about the child's picky eating behavior, or being uncertain about the effectiveness of the intervention. ASD children are highly sensitive to the emotional state of the caregiver. During mealtime, the child can perceive the tension in the air through the caregiver's tone of voice, facial expressions, and body language, triggering their own anxiety and defensive behavior. This behavior problem, which is directly caused by the emotional stress in the environment, rather than the food itself, is objectively recorded in the data as a behavior event and is paired with the food consumed at that time. This creates a spurious association: the behavior problem caused by the transmission of the caregiver's emotional state is mistakenly attributed to the "safe" food consumed by the child.

[0006] These hidden variables, the trace cross-contamination during meal preparation, the sensory information interference in the dining environment, and the transmission of the caregiver's emotional stress, together form a complex, noisy data environment. They make it difficult for a simple two-dimensional data analysis method based on "food intake-behavior performance" to give stable and reliable conclusions, thus causing the entire dietary intervention practice to be in a dilemma. In the non-professional records of the caregiver and the emotional and environmental disturbances in the family data, the existing technology lacks an effective method that can first identify and establish a reliable "behavior stable period" as a reference benchmark from the mixed time series data, and on this basis, when a behavior problem occurs, it can accurately compare the data before the problem event with the benchmark in multiple dimensions, to locate the most direct potential trigger event combination that causes the behavior to deviate from the benchmark from multiple potential, closely related interference factors such as cross-contamination, sensory stimulation, and emotional transmission. SUMMARY

[0007] The purpose of the present application is to provide a method and system for analyzing the correlation data of the dietary behavior of autistic patients, which has the advantages of being able to integrate multi-dimensional data, establish a behavior stable benchmark, and identify potential trigger factors that are inconsistent with the safe mode when the behavior deviates, thereby improving the accuracy and reliability of the analysis.

[0008] In one aspect, the present application provides a method for analyzing the correlation data of the dietary behavior of autistic patients, comprising: obtaining time series data, the time series data including subject intake data, subject behavior data, and multi-dimensional context data related to the environment in which the subject is located; According to the subject behavior data, a candidate stable period in which the subject behavior is in a stable state is identified, and based on a received confirmation instruction for the candidate stable period, the candidate stable period is established as a reference stable period; Based on time series data in the reference stable period, a safety behavior pattern archive containing a set of safety intake and a set of safety context is generated; In a case where the subject behavior data indicates that the behavior deviates from the stable state, time series data before the deviation occurs is compared with the safety behavior pattern archive to identify difference items in the time series data before the deviation occurs that are inconsistent with the safety behavior pattern archive; Based on the difference items, an attribution report indicating a potential trigger event of the behavior deviation is generated.

[0009] Optionally, the step of identifying, according to the subject behavior data, a candidate stable period in which the subject behavior is in a stable state, and establishing, based on a received confirmation instruction for the candidate stable period, the candidate stable period as a reference stable period, comprises: Upon receiving a veto instruction for the candidate stable period, the candidate stable period is established as a temporary reference period; After the temporary reference period is established, a reference stable period is established based on a newly received confirmation instruction for another candidate stable period; The temporary reference period is replaced by the reference stable period, and analysis results generated based on the temporary reference period are updated using the reference stable period.

[0010] Optionally, the step of generating, based on time series data in the reference stable period, a safety behavior pattern archive containing a set of safety intake and a set of safety context, comprises: For the subject intake data and the multi-dimensional context data contained in the time series data in the reference stable period, historical occurrence records related to the subject intake data and the multi-dimensional context data in all historical time series data before the reference stable period are obtained; Based on the historical occurrence records, safety verification degrees of the subject intake data and the multi-dimensional context data are respectively determined; The subject intake data and the corresponding safety verification degree are recorded in the set of safety intake, and the multi-dimensional context data and the corresponding safety verification degree are recorded in the set of safety context, to generate the safety behavior pattern archive.

[0011] Optionally, the step of determining, based on the historical occurrence records, safety verification degrees of the subject intake data and the multi-dimensional context data respectively, comprises: For any data item in the subject intake data and the multi-dimensional context data in the historical occurrence record, searching for occurrence records related to the data item and associated with the historical stable period to obtain a statistical number of stable period associations for the data item; For the same data item in the historical occurrence records, searching for occurrence records related to the data item and associated with historical behavior deviation events, so as to obtain the number of deviation event associations for the data item; A composite index is generated based on the number of associations during the stable period and the number of associations during the deviation event, and the composite index is determined as the security verification degree of the data item.

[0012] Optionally, when the subject behavior data indicates that the behavior deviates from a stable state, the step of comparing the time series data before the deviation occurs with the safe behavior pattern file to identify difference items in the time series data before the deviation that are inconsistent with the safe behavior pattern file includes: Comparing the time series data before the deviation occurs with the safety behavior pattern file to identify initial discrepancies; For the initial difference item, retrieve historical time series data, count the number of historical behavior deviation events associated with the initial difference item, and determine the risk indicator value of the initial difference item based on the counted number; Based on the risk indicator value, discrepancies that are inconsistent with the safety behavior pattern profile are identified.

[0013] Optionally, the step of retrieving historical time series data for the initial difference item, counting the number of historical behavior deviation events associated with the initial difference item, and determining the risk indication value of the initial difference item based on the counted number includes: For each difference point in the initial difference item, searching the historical time series data to identify a historical behavior deviation event associated with each difference point; Obtaining a severity level associated with the historical behavioral deviation event; Determining a risk contribution value of the historical behavior deviation event based on the severity, wherein the severity is positively correlated with the risk contribution value; Accumulate the risk contribution value of the historical behavior deviation events associated with each difference point, and use the accumulated result as the risk indicator value of each difference point.

[0014] Optionally, when the subject behavior data indicates that the behavior deviates from a stable state, the step of comparing the time series data before the deviation occurs with the safe behavior pattern file to identify difference items in the time series data before the deviation that are inconsistent with the safe behavior pattern file includes: determining whether any data item of the subject intake data or the multi-dimensional context data in the time series data before the deviation occurs is included in a corresponding set in the safe behavior pattern archive; if it is determined that the data item is included in the corresponding set in the safe behavior pattern archive, obtaining a corresponding safe verification degree of the data item in the safe behavior pattern archive, and identifying the data item as a difference item when the safe verification degree is lower than a preset threshold.

[0015] Optionally, the step of identifying a candidate stable period in which the subject behavior is in a stable state according to the subject behavior data comprises: sliding a preset length of time window on a time axis by a sliding window algorithm to identify a continuous time period that meets a preset condition, so as to identify a candidate stable period in which the subject behavior is in a stable state, the preset condition being that scores of the subject behavior data in the preset length of time window are all lower than a preset stable state threshold.

[0016] Optionally, after the step of comparing the time series data before the deviation occurs with the safe behavior pattern archive to identify difference items in the time series data before the deviation occurs that are inconsistent with the safe behavior pattern archive in the case where the subject behavior data indicates that the behavior deviates from the stable state, the system further comprises: determining whether the total number of the difference items exceeds an environment change threshold preset in the system, and stopping the generation of an attribution report indicating a potential trigger event of the behavior deviation when the total number of the difference items exceeds the environment change threshold.

[0017] On the other hand, the present application also provides a dietary behavior correlation data analysis system for autistic patients, which comprises: a data acquisition module configured to acquire time series data, the time series data comprising subject intake data, subject behavior data, and multi-dimensional context data related to an environment in which the subject is located; a stable period establishment module configured to identify a candidate stable period in which the subject behavior is in a stable state according to the subject behavior data, and to establish the candidate stable period as a reference stable period based on a received confirmation instruction for the candidate stable period; an archive generation module configured to generate a safe behavior pattern archive comprising a safe intake set and a safe context set based on the time series data in the reference stable period; a difference comparison module configured to compare the time series data before the deviation occurs with the safe behavior pattern archive to identify difference items in the time series data before the deviation occurs that are inconsistent with the safe behavior pattern archive in the case where the subject behavior data indicates that the behavior deviates from the stable state; A report generation module configured to generate an attribution report indicating the potential trigger event of the behavior deviation based on the difference item.

[0018] By the technical solution, the method and system for analyzing the meal behavior related data of the autism patient have at least the following advantages: By integrating multi-dimensional time sequence data, establishing a behavior stability benchmark, and comparing multi-dimensional differences when the behavior deviates, the potential trigger event is effectively identified, and the problems of insufficient data dimensions, lack of stable reference, and difficulty in positioning implicit trigger factors in the prior art are solved. BRIEF DESCRIPTION OF DRAWINGS

[0019] In order to more clearly illustrate the technical solutions of the present application, the following will briefly introduce the drawings needed in the embodiments. Obviously, for those skilled in the art, other drawings can also be obtained without creative labor.

[0020] Figure 1 Fig. 1 exemplarily shows a flow diagram of a method for analyzing meal behavior related data of an autism patient in an embodiment; Figure 2 Fig. 2 exemplarily shows a module configuration block diagram of a system for analyzing meal behavior related data of an autism patient in an embodiment.

[0021] Reference signs: 100, system for analyzing meal behavior related data of an autism patient; 10, data acquisition module; 20, stable period establishment module; 30, archive generation module; 40, difference comparison module; 50, report generation module. DETAILED DESCRIPTION

[0022] The technical solutions in the present application will be described in detail below with reference to the drawings in the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all embodiments. The components of the present application described and shown in the drawings herein can be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present application provided in the drawings is not intended to limit the scope of the claimed application, but only represents selected embodiments of the present application. Based on the embodiments of the present application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the present application.

[0023] It should be noted that similar reference numerals and letters refer to like items in the drawings and that once an item is defined in one drawing, that definition should be understood to be applicable to that item in the other drawings, unless explicitly stated otherwise. Also, in the description of the application, the terms "first", "second", etc. are used merely to identify corresponding items and do not connote or imply relative importance.

[0024] Traditional existing autism patient meal behavior correlation data analysis methods have problems in accurately identifying the correlation between behavior and meal when processing non-professional recorded data in a family environment. This is because the data is mixed with trace cross-contamination in the meal preparation process, sensory information interference in the dining environment, and the transmission of caregiver emotional stress, etc. implicit variables. These implicit variables cause the two-dimensional data analysis method based solely on food intake and behavior performance to fail, and cannot give stable and accurate conclusions. The core problem is how to autonomously identify and establish a stable behavior period as a reference benchmark from the mixed time series data, and on this basis, when subsequent behavior problems occur, the data before the problem event can be accurately compared with the benchmark in multiple dimensions, so as to locate the direct trigger event combination that causes the behavior to deviate from the benchmark from multiple potential correlation interference factors such as cross-contamination, sensory stimulation, and emotional transmission.

[0025] For example, assume that a user performs a gluten-free meal intervention for an autistic child in a family environment. The data recording system obtains the child's daily food intake, behavior performance, and environmental information. During meal preparation, the user uses the same operation table as the child's gluten-free food before handling it, although simple cleaning is performed, but still residual trace gluten. When eating, the family member eats food that emits a strong smell, and the child is exposed to this smell environment. At the same time, the user is nervous due to concern about the child's behavior or intervention effect, and transmits it to the child through non-verbal means. Hours later, the child exhibits behavior deviation. The data record shows that the child has ingested safe food, and the behavior log records the behavior deviation, but fails to capture implicit variables such as trace gluten exposure, odor stimulation, or user emotional state. The existing analysis method only compares safe food intake with behavior deviation, which may incorrectly attribute the behavior problem to safe food, leading to inaccurate intervention decisions.

[0026] If the above problems are not solved, the conclusions drawn based on inaccurate data analysis will mislead the intervention direction. Users may mistakenly remove foods harmless or beneficial to children from the recipe, or take ineffective intervention measures. This not only fails to improve the child's behavior problems, but also may lead to insufficient nutrient intake or increase the burden on the user. In the long run, such ineffective intervention practices will erode the user's confidence, leading to the interruption of the intervention program and delaying the child's rehabilitation process. At the same time, due to the inability to accurately identify the real trigger factors of behavior deviation, the system cannot provide relevant feedback and suggestions, and its application value will be limited.

[0027] As shown in Figure 1 A method for analyzing and optimizing data related to the dietary behavior of autism patients is shown. The method proposed in this application includes: S10, acquiring time series data, which includes subject intake data, subject behavior data, and multi-dimensional context data related to the environment in which the subject is located; S20, identifying a candidate stable period in which the subject's behavior is in a stable state based on the subject's behavior data, and establishing the candidate stable period as a reference stable period based on the received confirmation instruction for the candidate stable period; S30, generating a safe behavior pattern archive containing a set of safe intakes and a set of safe contexts based on the time series data in the reference stable period; S40, in the case where the subject's behavior deviates from the stable state, comparing the time series data before the deviation with the safe behavior pattern archive to identify the difference items in the time series data before the deviation that are inconsistent with the safe behavior pattern archive; S50, generating an attribution report indicating the potential trigger events of the behavior deviation based on the difference items.

[0028] In some embodiments, time series data is first acquired, which includes subject intake data, subject behavior data, and multi-dimensional context data related to the environment in which the subject is located. The time series data refers to a data set arranged in chronological order and can include different types of data, such as sensor-collected data, manually recorded data, and system log data, mainly to record the subject's state, behavior, and environment information over a period of time.

[0029] Subject intake data refers to information about food, drinks, or other substances ingested by the subject, which can take the form of text descriptions, classification labels, and nutritional component data, and is mainly used to record external inputs that may affect the subject's physiological and behavioral state.

[0030] The subject behavior data refers to the record of the behavior performance or state of the subject, which is in the form of behavior score, behavior event record, physiological index data, etc., to quantify or describe the behavior stability or deviation of the subject.

[0031] The multi-dimensional context data refers to various types of data related to the environment in which the subject is located, including physical environment information, social environment information, activity context information, caregiver state information, etc., which is mainly to record environmental factors that may affect the subject's behavior.

[0032] Based on the above-mentioned acquired time series data, the subject behavior data is used to identify the candidate stable period of the subject behavior, and the candidate stable period is established as the reference stable period based on the received confirmation instruction for the candidate stable period. The candidate stable period refers to a continuous period of time in which the subject behavior is relatively stable, which is identified by algorithm, manually marked, etc., and can provide a potential time window that can be used as a reference.

[0033] The reference stable period refers to a specific time period in which the subject behavior is stable, which is established by the confirmation instruction, and is mainly used to provide a reliable reference time window for establishing a safe behavior pattern archive.

[0034] The core innovation of the present application is to combine the acquisition of multi-dimensional time series data with the establishment of reference stable period based on user confirmation, and generate a multi-dimensional safe behavior pattern archive based on this, so that when the behavior deviates, the data before the deviation can be compared with the multi-dimensional archive to identify the difference items inconsistent with the stable pattern, achieving more accurate positioning of potential behavior trigger factors from complex environmental and behavior data.

[0035] The present application systematically analyzes the meal behavior correlation data of autistic patients through a series of steps.

[0036] Firstly, the system acquires time series data containing subject intake data, subject behavior data and multi-dimensional context data, which comprehensively records the key information of the subject within a period of time, laying a foundation for subsequent analysis.

[0037] Then, the system automatically identifies the candidate time period of the subject behavior performance based on the subject behavior data, and waits to receive the confirmation instruction from the user. Once the confirmation instruction for a certain candidate stable period is received, the candidate period is established as the reference stable period, which introduces manual judgment to ensure the reliability of the reference.

[0038] Subsequently, the system utilizes the time series data during the baseline stable period to construct a safe behavior pattern profile, which records the typical intake and context characteristics of the subject during the stable behavior period, forming a reference standard. Among them, the safe behavior pattern profile refers to a collection of data based on the data during the baseline stable period, which describes the typical intake and context characteristics of the subject in the stable state, which can be stored in the form of database records, data structures, etc., and its main purpose is to provide a reference model for subsequent comparison and difference identification.

[0039] When the subject behavior data indicates a deviation in behavior, the system extracts the time series data before the deviation occurs and compares it with the previously established safe behavior pattern profile. The purpose of comparison is to find the data items in the pre-deviation data that are inconsistent with the safe pattern. These inconsistent difference items are considered as potential trigger factors.

[0040] Among them, the difference item refers to the data point or data set in the time series data before the behavior deviation occurs, which is inconsistent with the pattern recorded in the safe behavior pattern profile, and its main purpose is to identify specific factors that may cause behavior deviation.

[0041] Finally, the system generates an attribution report based on the identified difference items, which clearly indicates the potential trigger events that may cause behavior deviation, helping users understand the reasons for behavior and take intervention measures. Among them, the attribution report refers to the analysis results generated based on the identified difference items, which indicate the potential reasons for behavior deviation, which can be in the form of text reports, chart displays, risk lists, etc., and its main purpose is to provide users with insights and suggestions about the trigger factors of behavior deviation. The whole process forms a closed loop from data collection, baseline establishment, pattern learning to deviation analysis and cause attribution, which can effectively cope with the analysis challenges in complex data environment.

[0042] In some embodiments, the timing data can be acquired in various ways. For example, the subject intake data can be acquired by the caregiver manually entering food names, quantities, or taking photos in the mobile application; the subject behavior data can be acquired by physiological indicators monitored by wearable devices combined with behavior scores recorded by the caregiver; the multi-dimensional context data can be acquired by environmental sensors, location information, and social interaction situations, self-emotional states recorded by the caregiver, etc. The system can use an algorithm based on the fluctuation of behavior scores to identify candidate stable periods, for example, set a time window length, and within the window, if the variance of behavior scores is lower than a pre-set threshold, mark the window as a candidate stable period. The user can see the list of identified candidate stable periods on the system interface and select one to confirm, for example, click a confirmation button. Once the reference stable period is established, the system will analyze all intake data and context data in that time period, record the intake and context features that appear frequently or are strongly associated with stable behavior into the safe behavior pattern archive, for example, establish two lists, one listing the foods that appear in the stable period and their appearance times, and the other listing the environmental features that appear in the stable period and their appearance times. When the subject behavior score suddenly rises, indicating that behavior deviation occurs, the system will backtrack to a period of time before the deviation occurs, extract the intake and context data in that time period, and compare it with the safe behavior pattern archive. The comparison can check whether the intake or context features that appear before the deviation are in the safe archive, or whether their appearance frequency is significantly lower than the frequency in the reference stable period. The identified difference items, for example, intake of food not recorded in the archive before the deviation, or the environment is a noisy public place that has never appeared in the archive, will be used to generate the attribution report. The attribution report can list these difference items and sort or mark them according to their degree of association with historical deviation events, prompting the caregiver that these may be potential causes of the current behavior deviation.

[0043] By the technical solution, the present application can effectively solve the problem of inaccurate meal behavior correlation analysis caused by implicit variable interference in the prior art. By obtaining multi-dimensional time series data, various factors affecting the subject behavior can be comprehensively captured, overcoming the limitation of relying on only a single data source. By introducing a user confirmation mechanism to establish a baseline stable period, the scheme can establish a reliable reference standard, avoiding the deviation that may be caused by analyzing based on unstable data. Based on the baseline period, a multi-dimensional safe behavior pattern archive is generated, providing an accurate basis for subsequent deviation analysis. When behavior deviation occurs, the data before the deviation is compared with the safe archive, which can accurately identify the difference items that do not conform to the stable pattern from the complex time series data. These difference items are more likely to be directly related to the triggering factors of behavior deviation. The final attribution report can clearly indicate the potential triggering event, helping the caregiver to more accurately understand the behavior reason, so as to develop more targeted and effective intervention strategies, improving the reliability and practicality of meal behavior correlation analysis.

[0044] In some embodiments, according to the subject behavior data, a candidate stable period in which the subject behavior is in a stable state is identified, and based on the received confirmation instruction for the candidate stable period, the candidate stable period is established as the baseline stable period, the step comprising: When the rejection instruction for the candidate stable period is received, the candidate stable period is established as a temporary baseline period; After the temporary baseline period is established, based on the newly received confirmation instruction for another candidate stable period, the baseline stable period is established; The temporary baseline period is replaced by the baseline stable period, and the analysis result generated based on the temporary baseline period is updated using the baseline stable period.

[0045] The rejection instruction is a signal sent by the user through the interface operation, indicating that the user does not accept the candidate stable period recommended by the current system or previously selected by the user as the final baseline.

[0046] The temporary baseline period refers to an intermediate state baseline period, which is set by the system after the user rejects a candidate stable period, in order to retain the candidate period and its preliminary analysis results, and provides a temporary analysis reference before the user reselects the baseline period.

[0047] The baseline stable period refers to a stable time period that is finally confirmed by the user and used to generate a safe behavior pattern archive and conduct subsequent behavior deviation analysis, and its purpose is to provide a reliable analysis reference; The present application solves the uncertainty and adjustment needs that users may encounter when establishing a reference stable period by introducing the concept of a temporary reference period and a reference period switching and result updating mechanism based on user instructions. When the user is not satisfied with the system-identified or previously selected candidate stable period and issues a veto instruction, the system does not directly abandon the candidate period, but temporarily establishes it as a temporary reference period. Doing so allows the system or user to conduct preliminary analysis or evaluation based on this temporary reference period.

[0048] Subsequently, if the user selects another more suitable candidate stable period through a new confirmation instruction, the system formally establishes this new candidate period as the reference stable period. The key is that the system replaces the previous temporary reference period with this new reference stable period, and most importantly, the system updates the analysis results previously generated based on the temporary reference period using this new reference stable period. This means that without re-executing all analysis steps, the system can efficiently adjust the analysis results to reflect the new reference. This mechanism allows the user to flexibly try different candidate stable periods and quickly obtain analysis results based on different references, thereby selecting the reference period that best reflects the subject's stable state, improving the accuracy of the analysis and the convenience of user operations. This refinement and optimization of the reference stable period establishment step makes the entire autism patient dietary behavior correlation data analysis method more robust and user-friendly in practical applications, better able to handle complex and variable data and user needs.

[0049] For example, a plurality of candidate stable periods are first identified from the subject behavior data, such as candidate stable period A and candidate stable period B. These candidate periods are displayed on the user interface for the user to select. Assume that the user initially selects candidate stable period A and issues a confirmation instruction. The system establishes candidate stable period A as the reference stable period and generates a preliminary analysis result, such as a safety behavior pattern profile A, based on the reference stable period A. After reviewing the safety behavior pattern profile A, the user may find that the profile fails to accurately reflect his / her cognition of the subject stable state, for example, the profile contains some unsafe intake or context that the user considers unsafe. At this time, the user can issue a rejection instruction for candidate stable period A through the system interface. After receiving the rejection instruction, the system downgrades the status of candidate stable period A from the formal reference period or marks it as a temporary reference period. Subsequently, the user may decide to try another candidate stable period B. The user selects candidate stable period B on the interface and issues a confirmation instruction. After receiving the confirmation instruction for another candidate stable period B, the system formally establishes candidate stable period B as the reference stable period. The system then performs an operation of replacing the temporary reference period A with the reference stable period B. This means that all subsequent analyses will be based on the reference stable period B. The system updates the analysis result (e.g., safety behavior pattern profile A) previously generated based on the temporary reference period A to generate a new analysis result (e.g., safety behavior pattern profile B) based on the reference stable period B. This updating process can include recalculating the safety verification degree, rebuilding the safety intake set and the safety context set, or re-performing the difference comparison and attribution analysis. In this way, the user can conveniently switch the reference stable period and quickly obtain the updated analysis result without starting the entire analysis process from scratch.

[0050] Through the above technical solution, the present application provides a mechanism for flexibly adjusting the reference stable period. When the user has doubts about the initially established reference stable period, he / she can set it to a temporary state through a rejection instruction and quickly select another candidate stable period as the new formal reference. The system can efficiently update the analysis result based on the new reference stable period, avoiding repeated performance of the entire analysis process. This allows the user to conveniently switch and optimize the reference stable period according to the actual situation and further understanding of the data, thereby obtaining more accurate and reliable analysis results and improving the efficiency of data analysis and the convenience of user operation.

[0051] In some embodiments, the step of generating a safety behavior pattern profile containing a safety intake set and a safety context set based on the time series data in the reference stable period includes: For the subject intake data and multi-dimensional context data contained in the time series data in the reference stable period, obtaining the historical occurrence records related to the subject intake data and multi-dimensional context data in all historical time series data before the reference stable period; Based on historical occurrence records, determine the safety verification degree of the subject intake data and the multi-dimensional context data respectively; Record the subject intake data and its corresponding safety verification degree in the safe intake set, and record the multi-dimensional context data and its corresponding safety verification degree in the safe context set, to generate a safe behavior pattern archive.

[0052] Among them, the historical occurrence record refers to the historical data entries that match or are related to the subject intake data and multi-dimensional context data in the reference stable period among all historical time series data before the reference stable period. Obtaining historical occurrence records can be achieved through database query or data retrieval, and the purpose is to provide historical reference for evaluating the safety of data in the reference period.

[0053] The safety verification degree refers to the quantitative evaluation of the safety of the subject intake data or multi-dimensional context data in the historical data. Specifically, it can be a numerical value or level calculated based on its frequency, correlation strength or impact degree in the historical stable period and / or historical behavior deviation events. Its purpose is to distinguish the safety risk level of different intakes or contexts.

[0054] The safe intake set is a data structure for storing subject intake data that occurs in the reference stable period, and is associated with their respective determined safety verification degree. Its purpose is to build an intake list containing quantitative safety information.

[0055] The safe context set is also a data structure for storing multi-dimensional context data that occurs in the reference stable period, and is associated with their respective determined safety verification degree. Its purpose is to build a context list containing quantitative safety information.

[0056] The safe behavior pattern archive is a comprehensive data structure that contains the safe intake set and the safe context set. It represents the subject's safe behavior pattern in the reference stable period, where each component is accompanied by a safety verification degree based on historical data evaluation. Its purpose is to provide a structured and quantitative reference for subsequent behavior deviation comparison analysis.

[0057] By obtaining the subject intake data and multi-dimensional context data that occur in the reference stable period, and tracing their occurrence in the historical time series data before the reference stable period, the safety of these data items is evaluated.

[0058] Specifically, by analyzing the historical occurrence records, the degree of association of a certain intake or context with stable behavior or deviated behavior in the past can be understood. Based on such historical association, the degree of safety verification of each intake and context is quantitatively determined. For example, a data item that frequently occurs in the stable period and rarely occurs in the deviated event will be determined to have a higher degree of safety verification; on the contrary, a data item that frequently occurs in the deviated event will be determined to have a lower degree of safety verification. Subsequently, these intakes and contexts and their corresponding degrees of safety verification are recorded into the safe intake set and the safe context set respectively, together constituting the safe behavior pattern archive. This archive not only lists the safe factors that occur in the benchmark period, but more importantly, it assigns a quantitative safety evaluation value to each factor. Thus, when subsequent behavior deviation analysis is performed, the data before the deviation occurs can be compared with this archive containing quantitative safety information, and those difference items with a lower degree of safety verification that are likely to be related to the deviated event can be more accurately identified. This archive generation method based on the quantification of safety based on historical data enables the subsequent comparison analysis to focus on potential high-risk factors, improving the accuracy and efficiency of the analysis, and thus more effectively locating the potential trigger event combination that leads to behavior deviation from the complex environment, solving the problem of information redundancy caused by simply listing data and ignoring the difference in safety degree.

[0059] For example, suppose a subject's consumption of "apple" and "milk" is recorded during a baseline stable period, and the contextual data includes "in the living room" and "with other family members present." The system retrieves all historical time-series data prior to the baseline stable period for the four data items: "apple," "milk," "in the living room," and "with other family members present." For example, the search results show that "apple" appeared 50 times in the past year, 48 of which were associated with stable behavior and 2 with slight deviations; "milk" appeared 30 times, 20 of which were associated with stable behavior and 10 with moderate deviations; "in the living room" appeared 100 times, 95 of which were associated with stable behavior and 5 with slight deviations; and "with other family members present" appeared 80 times, 50 of which were associated with stable behavior and 30 with deviations of varying degrees. Based on these historical occurrence records, the security verification level can be determined. For example, a simple calculation method can be used: security verification level = (number of associations during the stable period + 1) / (total number of occurrences + 2). Based on this calculation, the safety verification level for "apple" is (48+1) / (50+2) = 49 / 52 ≈ 0.94; for "milk," it's (20+1) / (30+2) = 21 / 32 ≈ 0.66; for "in the living room," it's (95+1) / (100+2) = 96 / 102 ≈ 0.94; and for "with other family members present," it's (50+1) / (80+2) = 51 / 82 ≈ 0.62. This information is then recorded in the safety behavior pattern profile. The safe ingestion set will include "apple: 0.94" and "milk: 0.66." The safe situation set will include "in the living room: 0.94" and "with other family members present: 0.62." This creates a safety behavior pattern profile containing quantitative safety information.

[0060] Through the above technical solution, when generating a safety behavior pattern profile based on time-series data from a baseline stability period, the data is no longer simply listed. Instead, a quantitative assessment of the data items during the baseline period is performed by obtaining historical occurrence records and determining the degree of safety verification. The resulting profile contains safety risk information for each ingested object and scenario, avoiding information redundancy, highlighting key safety factors, and taking into account the differences in safety levels of different data items. This allows for more accurate and effective identification of potential risk factors when subsequently comparing the data before the deviation occurred with the profile, improving the accuracy and efficiency of subsequent comparative analysis.

[0061] In some embodiments, the step of determining the security verification level of the subject's intake data and the multi-dimensional contextual data based on the historical occurrence records includes: For any data item in the subject intake data and the multi-dimensional context data in the historical occurrence records, retrieve occurrence records associated with the data item and the historical stable period to obtain the stable period association number of the data item; For the same data item in the historical occurrence records, retrieve occurrence records associated with the data item and the historical behavior deviation event to obtain the deviation event association number of the data item; Based on the stable period association number and the deviation event association number, generate a composite index, and determine the composite index as the safety verification degree of the data item.

[0062] Among them, the data item refers to a specific entry in the subject intake data or multi-dimensional context data in the historical occurrence records, such as a specific food or a specific environmental parameter value, which can be represented by a field or record in the data structure, and its purpose is to decompose the historical data into analyzable units.

[0063] Among them, the historical stable period refers to a period in which the subject behavior is in a stable state in the historical time series data, which can be identified in a similar way to the established baseline stable period, and is used to provide data reference in the normal state of the subject behavior.

[0064] Among them, the historical behavior deviation event refers to an event in which the subject behavior deviates from the stable state in the historical time series data, which can be identified by the behavior data score exceeding the threshold value, and its purpose is to provide data reference in the abnormal state of the subject behavior.

[0065] Among them, the composite index refers to a value calculated by combining the stable period association number and the deviation event association number, which can be implemented by various calculation formulas such as ratio, difference or weighted average, and its purpose is to comprehensively evaluate the historical occurrence pattern of the data item.

[0066] And the safety verification degree is the safety evaluation value of the data item determined based on the composite index, which can be represented by numerical score or grade division, and its purpose is to provide quantitative basis for generating the safety behavior pattern archive.

[0067] The application counts the number of occurrences of any data item in the historical occurrence record in the historical stable period (stable period correlation number) and the number of occurrences before the historical behavior deviation event occurs (deviation event correlation number) respectively. The stable period correlation number reflects the degree of association of the data item with the stable state of the subject behavior, and the deviation event correlation number reflects the degree of association of the data item with the unstable state of the subject behavior. By generating a composite index based on the two statistical numbers, the historical occurrence pattern of the data item can be comprehensively evaluated. Determining the composite index as the security verification degree of the data item makes the evaluation of the security of the data item more comprehensive and accurate. This method provides a reliable quantitative basis for generating a security behavior pattern archive containing a set of safe intakes and a set of safe contexts. The security verification degree determined in this way can more accurately reflect the reaction pattern of the subject to specific intakes or environmental factors in different contexts, thereby making subsequent behavior deviation analysis and attribution based on the archive more accurate and improving the effectiveness of the entire dietary behavior correlation data analysis method.

[0068] Through the above technical solution, based on the historical occurrence record, by counting the correlation numbers of the data item in the historical stable period and the historical behavior deviation event, a composite index is generated as the security verification degree, which can more comprehensively and accurately evaluate the security of the subject intake data and multi-dimensional context data. This more accurate security verification degree makes the generated security behavior pattern archive more reliable, providing a solid foundation for subsequent behavior deviation analysis and attribution, and improving the accuracy and effectiveness of the entire analysis method.

[0069] In the case where the subject behavior data indicates a deviation from the stable state, the step of comparing the time series data before the deviation occurs with the security behavior pattern archive to identify the difference items inconsistent with the security behavior pattern archive includes: Comparing the time series data before the deviation occurs with the security behavior pattern archive to identify the initial difference items; For the initial difference items, the historical time series data is retrieved, the number of historical behavior deviation events associated with the initial difference items is counted, and the number of statistical results is determined as the risk indication value of the initial difference items; Based on the risk indication value, the difference items inconsistent with the security behavior pattern archive are identified.

[0070] Among them, the initial difference items refer to all data items or events that are identified as deviating from the security behavior pattern through the preliminary comparison of the time series data before the deviation occurs with the security behavior pattern archive, which can be achieved by directly comparing the data content or the security verification degree. Its purpose is to preliminarily circumscribe all possible abnormal factors; The risk indication value is a value used to quantify the degree of association between the initial difference item and the historical behavior deviation event. It can be achieved by counting the number of historical associations or calculating a composite value based on the severity of the historical association events. The purpose is to assess the possibility or importance of the initial difference item causing the behavior deviation.

[0071] By preliminarily comparing the time series data before the deviation occurs with the security behavior pattern archive, all potential initial difference items are identified. Since these initial difference items are the result of preliminary screening, they may contain noise unrelated to the current behavior deviation. Therefore, further screening is needed.

[0072] Therefore, for each initial difference item, the application retrieves historical time series data, counts the number of times the difference item appears simultaneously with a behavior deviation event in history, and uses this number as the risk indication value of the difference item, thereby quantifying its relevance to the behavior deviation.

[0073] It is because of the introduction of the risk indication value based on historical data as the screening basis that subsequent can selectively identify those high-risk difference items as the final difference items inconsistent with the security behavior pattern archive, while excluding those low-risk accidental difference items. This secondary screening mechanism based on historical association effectively reduces noise interference, making the identified difference items more likely to have actual causal association with the current behavior deviation, thereby providing a reliable basis for subsequent generation of more accurate and targeted attribution reports.

[0074] For example, when the system detects that the subject behavior data indicates a behavior deviating from the stable state, the system obtains the time series data in a period of time before the deviation occurs. The system compares the intake and context data in the time series data with the pre-established safety behavior pattern profile. For example, if the profile specifies that a certain food should not be taken in a specific time period, and the time series data shows that the food is taken, the food is identified as an initial difference item. Similarly, if the profile specifies that the behavior should be stable in a specific context (e.g., a quiet environment), and the time series data shows that the environment is noisy, the noisy environment can be identified as another initial difference item. Next, the system searches the historical time series database for each identified initial difference item, such as "eating peanuts". The system finds all past behavior deviation events and checks whether "eating peanuts" also occurred in a period of time before the events occurred. The system counts the total number of times when "eating peanuts" occurs simultaneously with the historical behavior deviation events. This count, for example, 10 times, is determined as the risk indication value of the initial difference item "eating peanuts". The system repeats this process for all initial difference items to calculate their respective risk indication values. Finally, the system filters based on the risk indication values. For example, a threshold is set, and only the initial difference items with a risk indication value higher than the threshold are retained as the final identified difference items that are inconsistent with the safety behavior pattern profile. The initial difference items with a risk indication value lower than the threshold are considered as accidental factors and are not included in the final difference item set.

[0075] Through the above technical solutions, the application can perform risk assessment and screening on the initially identified difference items based on historical data, thereby excluding accidental factors with low correlation with behavior deviation, making the finally identified difference items more accurately reflect the potential trigger events that cause behavior deviation, and improving the accuracy and reference value of the attribution report.

[0076] In some embodiments, the step of retrieving historical time series data for the initial difference items, counting the number of historical behavior deviation events associated with the initial difference items, and determining the counted number as the risk indication value of the initial difference items includes: For each difference point in the initial difference items, the historical time series data is retrieved to identify historical behavior deviation events associated with each difference point. The severity associated with the historical behavior deviation events is obtained. Based on the severity, a risk contribution value of the historical behavior deviation events is determined, wherein the severity and the risk contribution value are positively correlated. The risk contribution values of the historical behavior deviation events associated with each difference point are accumulated, and the accumulated result is taken as the risk indication value of each difference point.

[0077] The difference point is a specific data item or data feature that is inconsistent with the safety behavior pattern archive after comparing the time series data before the deviation occurs with the safety behavior pattern archive, which can be implemented by using a data comparison algorithm, a pattern matching algorithm, etc.

[0078] The historical time series data is all time series data recorded by the system before the current behavior deviation event occurs, including subject intake data, subject behavior data, and multi-dimensional context data, which can be stored in a database, a data warehouse, or other data storage media.

[0079] The risk contribution value is a weight or numerical value provided by a specific historical behavior deviation event to the risk assessment of the difference point associated with it, which can be determined based on the severity by looking up a table, formula calculation, or machine learning model. The risk indication value refers to a quantitative measure of the potential risk of a difference point after considering the severity of all historical behavior deviation events associated with the difference point, which can be calculated by using accumulation, weighted average, or other aggregation algorithms.

[0080] The present application retrieves historical time series data for each of the initial difference items, identifies historical behavior deviation events associated with each of the difference points, and obtains the severity of these historical behavior deviation events. Based on the severity, the risk contribution value of the historical behavior deviation event is determined, and the higher the severity, the greater the risk contribution value. Then, the risk contribution values of the historical behavior deviation events associated with each of the difference points are accumulated, and the accumulated result is taken as the risk indication value of each of the difference points. Because the severity of the historical behavior deviation event is included in the risk assessment process of the difference point, the risk indication value can more accurately reflect the degree of association between the difference point and the high-risk behavior deviation event, thereby distinguishing difference items of different risk levels. This provides a more refined and differentiated risk assessment compared to the method of only counting the number of historical associations, effectively solving the problem of low-value information interference caused by risk assessment based only on the number of occurrences, so that subsequent attribution reports based on difference items can focus on difference items with high risk indication values, improving the accuracy and efficiency of attribution analysis.

[0081] For example, assume that an initial difference item is identified before a behavior deviation occurs, which contains a difference point, such as “intake: peanuts”. To evaluate the risk of the difference point “peanuts”, the system retrieves historical time-series data and identifies that “intake: peanuts” appeared before a behavior deviation event three times in the past. The first time, the behavior deviation was manifested as “mild agitation”, which was evaluated as 3 in severity; the second time, the behavior deviation was manifested as “moderate emotional outburst”, which was evaluated as 7 in severity; and the third time, the behavior deviation was manifested as “mild repetitive behavior”, which was evaluated as 2 in severity. The system determines a risk contribution value based on the severity, for example, simply taking the severity score as the risk contribution value. Thus, the risk contribution values of the three historical events are 3, 7, and 2, respectively. The system accumulates the risk contribution values of the three historical events: 3 + 7 + 2 = 12. The accumulated result 12 is taken as the risk indication value of the difference point “intake: peanuts”. In this way, even though “peanuts” only appeared three times, since it is associated with a moderate emotional outburst event, its risk indication value (12) will be higher than a difference point that appears more times (e.g., five times) but is only associated with mild behavior fluctuations (e.g., five mild fluctuations, each with a severity of 1, and an accumulated risk indication value of 5).

[0082] By the above technical solution, by taking the severity of historical behavior deviation events into account in the risk evaluation of a difference point, the application can more accurately quantify the associated risk of each difference point and behavior deviation event, thereby effectively distinguishing potential trigger factors of different risk levels, enabling the subsequent attribution analysis to focus on high-risk difference items, and improving the value and analysis efficiency of the attribution report.

[0083] In some embodiments, in the case where the subject behavior data indicates a stable state of behavior deviation, the step of comparing the time-series data before the deviation occurs with the safe behavior pattern archive to identify difference items in the time-series data before the deviation that are inconsistent with the safe behavior pattern archive comprises: determining whether any data item of the subject intake data or multi-dimensional context data in the time-series data before the deviation is contained in the corresponding set in the safe behavior pattern archive; if it is determined that the data item is contained in the corresponding set in the safe behavior pattern archive, obtaining the corresponding safe verification degree of the data item in the safe behavior pattern archive, and identifying the data item as a difference item when the safe verification degree is lower than a preset threshold.

[0084] wherein the time-series data before the deviation occurs refers to time-series data collected within a period of time before the subject behavior data indicates a stable state of behavior deviation, which can include subject intake data, subject behavior data, and multi-dimensional context data related to the environment in which the subject is located.

[0085] The preset threshold refers to a threshold for judging whether the security verification degree is low enough, which can be set according to actual application requirements and historical data analysis results, for example, can be set as a fixed value, or dynamically adjusted according to the data item type.

[0086] By comparing the time series data before the deviation occurs when the subject behavior data indicates that the behavior deviates from the stable state, the difference items inconsistent with the safe behavior pattern archive are identified.

[0087] Specifically, first, it is judged whether any data item of the subject intake data or multi-dimensional context data in the time series data before the deviation occurs is included in the corresponding set in the safe behavior pattern archive. This judgment can quickly filter out data items that have never appeared in the safe behavior pattern, and preliminarily narrow the range of potential difference items.

[0088] Further, for those data items included in the corresponding set of the safe behavior pattern archive, the security verification degree recorded in the safe behavior pattern archive is obtained. The security verification degree reflects the occurrence of the data item in the historical stable period and the deviation event, and is an important indicator for measuring its security. Only when the security verification degree of the data item is lower than the preset threshold, it is identified as a difference item. This means that even if a data item appears in the safe behavior pattern, if its historical record shows a high correlation with the deviation event or a low frequency of occurrence in the stable period, it will still be considered as a potential difference item. On the contrary, if a data item appears before the deviation, but its security verification degree is high, it will not be misjudged as a difference item.

[0089] It is precisely because of this fine comparison mechanism combining existence judgment and security verification degree evaluation that the application can more accurately identify the difference items that may truly cause the behavior deviation, avoiding the misjudgment and noise interference that may be caused by simple comparison. This fine identification process benefits from the establishment of the safe behavior pattern archive, which provides the security verification degree information of the data item, making the comparison process more intelligent and effective.

[0090] Through the above technical solutions, the application can perform more fine comparison on the time series data before the deviation occurs. By judging whether the data item is included in the corresponding set of the safe behavior pattern archive and further combining the security verification degree for judgment, it can avoid misjudging those items with high security although they have appeared as difference items, thereby improving the accuracy of difference item identification. This helps to more effectively filter out the key factors that may truly cause the behavior deviation from complex time series data, laying a foundation for generating more reliable attribution reports subsequently.

[0091] In some embodiments, according to the subject behavior data, the step of identifying a candidate stable period in which the subject behavior is in a stable state comprises: The continuous time period satisfying the preset condition is identified by sliding a preset length of time window on the time axis through a sliding window algorithm, so as to identify the candidate stable period in which the subject behavior is in a stable state, wherein the preset condition refers to that the scores of the subject behavior data are all lower than a preset stable state threshold in the preset length of time window.

[0092] The sliding window algorithm refers to a time series analysis technique, which moves a fixed size window on the data sequence, processes the data in the window, and can be implemented in a fixed step sliding, variable step sliding and the like, for local analysis of continuous data.

[0093] The preset length of time window refers to the size of the window in the sliding window algorithm, that is, the length of the time period analyzed each time, which can be set in a fixed time length or a fixed number of data points, for determining the time scale of analyzing behavior stability.

[0094] The preset condition refers to the standard for judging whether the behavior in a time window is stable, which can be set in an average value, a maximum value, a fluctuation range and the like of the behavior score in the window, for quantifying the stability of the behavior; the score of the subject behavior data is a numerical value obtained by quantitatively evaluating the subject behavior data, which can be generated in a behavior observation scale, sensor data analysis and the like, for converting complex behavior into a calculable numerical value.

[0095] The preset stable state threshold is a limit value for judging whether the behavior score meets the stable condition, which can be set in a historical data statistics, an expert experience setting, an individual baseline determination and the like, for distinguishing between stable behavior and unstable behavior.

[0096] The application identifies a continuous time period meeting the preset condition as a candidate stable period by adopting a sliding window algorithm to slide on a time window of a preset length on a time axis and checking whether the subject behavior data scores in each window are all below a preset stable state threshold during the sliding process. Specifically, the system obtains continuous subject behavior data and scores the same. A time window of a preset length is defined and moved forward on the time axis of the behavior data according to a certain step size. At each position of the window, the system evaluates the subject behavior data scores in the time period covered by the window. Only when the scores of all the behavior data in the window are below the preset stable state threshold, the time period is considered to meet the stability requirement. The system identifies these continuous time periods meeting the condition and marks the same as candidate stable periods. This method can flexibly adapt to the changes of the subject behavior data on different time scales and capture the local time period meeting the stable standard through the sliding window. In combination with the step of identifying the candidate stable period in the basic scheme, this specific identification method can more accurately extract the time period representing the real stable state of the subject from the complex behavior data stream, provides a basis for generating a more reliable safety behavior mode archive subsequently and improves the accuracy of the entire analysis method.

[0097] For example, assuming that the subject behavior data is a behavior score recorded every minute, the system can set the preset length of the time window to 30 minutes and the preset stable state threshold to 5 points. After the system receives the continuous subject behavior score data stream, the sliding window algorithm is started. A 30-minute window starts from the starting point of the data stream and slides backward at a step size of, for example, 1 minute. At each window position, the system checks all the behavior scores recorded in the 30 minutes. If all the scores in the 30 minutes are less than 5 points, the system marks the time period as a candidate stable period. The window continues to slide, and if the subsequent continuous windows all meet the condition, the system combines the continuous time periods covered by the windows and identifies a longer candidate stable period.

[0098] Through the above technical solution, the sliding window algorithm can flexibly search for a segment meeting the stable standard in the continuous behavior data stream, overcoming the limitations of simple fixed time period division. In combination with the preset condition, the requirement for stability can be adjusted according to the actual situation. This method can more accurately extract the time period representing the real stable state of the subject from the complex behavior data, improving the accuracy of identifying the stable period.

[0099] In some embodiments, after the step of comparing the time series data before the deviation occurs with the safety behavior mode archive to identify the difference items in the time series data before the deviation occurs that are inconsistent with the safety behavior mode archive in the case where the subject behavior data indicates that the behavior deviates from the stable state, the method further comprises: Determine whether the total number of difference items exceeds the environmental change threshold preset within the system. If the total number of difference items exceeds the environmental change threshold, the system stops generating attribution reports indicating potential triggering events of behavioral deviation.

[0100] The environment change threshold refers to a value preset within the system, which is used to determine whether the total number of difference items indicates that a significant change has occurred in the environment. It can be implemented using a fixed value or a value dynamically adjusted based on historical data.

[0101] The solution of the present application adds a judgment step after identifying the discrepancies in the time series data before the deviation that are inconsistent with the security behavior pattern archive. This judgment step counts the identified discrepancies to obtain a total number of discrepancies, and compares this total number with the environmental change threshold preset within the system. It is precisely because of the introduction of the environmental change threshold that the system can identify situations where the total number of discrepancies increases abnormally. This abnormal increase usually indicates that a major change has occurred in the environment, rather than being caused by a specific potential triggering event. By stopping the generation of attribution reports when the total number of discrepancies exceeds the environmental change threshold, this solution avoids the generation of redundant and potentially misleading attribution reports when a large number of irrelevant discrepancies appear due to drastic changes in the environment. This is different from the basic solution that generates attribution reports directly based on the discrepancies, which will generate invalid reports when the environment changes significantly. By adding a judgment step, this solution makes the generation of attribution reports more targeted, and only performs them when the number of discrepancies is within a reasonable range, thereby improving the effectiveness and practicality of the analysis.

[0102] Through the above technical solution, the present application can, after identifying the difference items, determine whether the total number of difference items indicates a significant change in the environment. When the total number of difference items exceeds the preset environmental change threshold, the system stops generating attribution reports. This avoids the generation of redundant reports when environmental changes lead to a large number of irrelevant difference items, improves the efficiency and accuracy of analysis, and prevents information overload from interfering with caregivers' judgment.

[0103] On the other hand, Figure 2 As shown, a system for analyzing dietary behavior-related data of autistic patients is exemplarily shown. This application further proposes a system for analyzing dietary behavior-related data of autistic patients 100, which includes: A data acquisition module 10 is used to acquire time series data, wherein the time series data includes the subject's ingested data, the subject's behavior data, and multi-dimensional contextual data related to the subject's environment; a stable period establishing module 20 for identifying, based on the subject behavior data, a candidate stable period in which the subject behavior is in a stable state, and establishing the candidate stable period as a reference stable period based on a received confirmation instruction for the candidate stable period; The archive generation module 30 is configured to generate a safe behavior pattern archive containing a safe intake set and a safe context set based on the time series data in the reference stable period. The difference comparison module 40 is configured to compare the time series data before the deviation occurs with the safe behavior pattern archive to identify the difference items inconsistent with the safe behavior pattern archive in the time series data before the deviation occurs when the subject behavior data indicates that the behavior deviates from the stable state. The report generation module 50 is configured to generate an attribution report indicating the potential trigger event of the behavior deviation based on the difference items.

[0104] The present application realizes the method of analyzing the dietary behavior correlation data of autistic patients by converting it into a modular system. The data acquisition module is responsible for uniformly collecting time series data containing subject intake data, subject behavior data, and multi-dimensional context data, providing a basis for subsequent analysis. The stable period establishment module receives subject behavior data, identifies candidate stable periods where the behavior is in a stable state through analysis, and establishes the candidate stable period as the reference stable period based on the received confirmation instruction. This process introduces a manual confirmation link to ensure the reliability of the reference stable period. The archive generation module uses the time series data in the reference stable period to construct a safe behavior pattern archive containing a safe intake set and a safe context set, which serves as a reference for subsequent comparison. When the subject behavior data indicates that the behavior deviates from the stable state, the difference comparison module starts comparing the time series data before the deviation occurs with the safe behavior pattern archive to identify the difference items inconsistent with the safe pattern. Finally, the report generation module generates an attribution report indicating the potential trigger event of the behavior deviation based on the identified difference items. The entire system realizes a complete analysis process from raw data collection to final attribution report generation through the sequential execution of these modules and data transmission. This design of converting method steps into specific system modules makes the responsibilities of each functional unit clear, easy to handle in parallel and optimize, thereby better handling large amounts of time series data and high computational complexity, improving the efficiency and stability of data analysis.

[0105] Through the above technical solutions, the present application converts the method of analyzing the dietary behavior correlation data of autistic patients into a modular system, realizing the systematic deployment of the method. The modular design makes the responsibilities of each functional unit clear, easy to handle in parallel and optimize, thereby improving the efficiency of data analysis. The modules interact with each other through explicit interfaces, enhancing the stability and maintainability of the system. The system can effectively handle large amounts of time series data and quickly generate attribution reports, providing timely and targeted references for dietary intervention for autistic patients.

[0106] The above merely provides an example of the present application, and is not used to limit the protection scope of the present application. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application should be included in the protection scope of the present application.

Claims

1. A method for analyzing dietary behavior association data of autistic patients, characterized in that: include: Acquiring time series data, the time series data including subject ingestion data, subject behavior data, and multi-dimensional contextual data related to the subject's environment; identifying, based on the subject behavior data, a candidate stable period in which the subject behavior is in a stable state, and establishing the candidate stable period as a reference stable period based on a received confirmation instruction for the candidate stable period; generating a safety behavior pattern profile comprising a set of safe ingested objects and a set of safe situations based on the time series data during the baseline stable period; When the subject behavior data indicates that the behavior deviates from a stable state, comparing the time series data before the deviation occurs with the safe behavior pattern file to identify the difference items in the time series data before the deviation that are inconsistent with the safe behavior pattern file; Based on the difference terms, an attribution report is generated indicating a potential triggering event of the behavioral deviation.

2. The method for analyzing dietary behavior association data of autistic patients according to claim 1, characterized in that: The step of identifying, based on the subject behavior data, a candidate stable period in which the subject behavior is in a stable state, and establishing the candidate stable period as a reference stable period based on a received confirmation instruction for the candidate stable period includes: When a rejection instruction for the candidate stable period is received, establishing the candidate stable period as a temporary base period; After the temporary base period is established, a base stable period is established based on a newly received confirmation instruction for another candidate stable period; The temporary base period is replaced by the base stable period, and the analysis result generated based on the temporary base period is updated by the base stable period.

3. The method for analyzing dietary behavior association data of autistic patients according to claim 1, characterized in that: The step of generating a safe behavior pattern profile including a set of safe ingested objects and a set of safe situations based on the time series data during the baseline stable period includes: For the subject intake data and the multi-dimensional context data included in the time series data within the reference stable period, obtaining historical occurrence records related to the subject intake data and the multi-dimensional context data in all historical time series data before the reference stable period; Determining, based on the historical occurrence records, respective security verification levels of the subject intake data and the multi-dimensional contextual data; The subject intake data and its corresponding safety verification level are recorded in the safety intake set, and the multi-dimensional situation data and its corresponding safety verification level are recorded in the safety situation set to generate the safety behavior pattern file.

4. The method for analyzing dietary behavior association data of autistic patients according to claim 3, characterized in that: The step of determining the security verification levels of the subject intake data and the multi-dimensional context data based on the historical occurrence records includes: For any data item in the subject intake data and the multi-dimensional context data in the historical occurrence record, searching for occurrence records related to the data item and associated with the historical stable period to obtain a statistical number of stable period associations for the data item; For the same data item in the historical occurrence records, searching for occurrence records related to the data item and associated with historical behavior deviation events, so as to obtain the number of deviation event associations for the data item; A composite index is generated based on the number of associations during the stable period and the number of associations during the deviation event, and the composite index is determined as the security verification degree of the data item.

5. The method for analyzing dietary behavior association data of autistic patients according to claim 1, characterized in that: The step of comparing the time series data before the deviation occurs with the safe behavior pattern file to identify the difference items in the time series data before the deviation occurs that are inconsistent with the safe behavior pattern file when the subject behavior data indicates that the behavior deviates from the stable state includes: Comparing the time series data before the deviation occurs with the safety behavior pattern file to identify initial discrepancies; For the initial difference item, retrieve historical time series data, count the number of historical behavior deviation events associated with the initial difference item, and determine the risk indicator value of the initial difference item based on the counted number; Based on the risk indicator value, discrepancies that are inconsistent with the safety behavior pattern profile are identified.

6. The method for analyzing dietary behavior association data of autistic patients according to claim 5, characterized in that: The step of retrieving historical time series data for the initial difference item, counting the number of historical behavior deviation events associated with the initial difference item, and determining the risk indicator value of the initial difference item based on the counted number includes: For each difference point in the initial difference item, searching the historical time series data to identify a historical behavior deviation event associated with each difference point; Obtaining a severity level associated with the historical behavioral deviation event; Determining a risk contribution value of the historical behavior deviation event based on the severity, wherein the severity is positively correlated with the risk contribution value; Accumulate the risk contribution value of the historical behavior deviation events associated with each difference point, and use the accumulated result as the risk indicator value of each difference point.

7. The method for analyzing dietary behavior association data of autistic patients according to claim 1, characterized in that: When the subject behavior data indicates that the behavior deviates from a stable state, the step of comparing the time series data before the deviation occurs with the safe behavior pattern file to identify the difference items in the time series data before the deviation that are inconsistent with the safe behavior pattern file includes: Determining whether any data item of the subject's intake data or the multi-dimensional situational data in the time series data before the deviation occurs is included in a corresponding set in the safety behavior pattern file; If it is determined that the data item is included in the corresponding set in the security behavior pattern file, the security verification level corresponding to the data item in the security behavior pattern file is obtained, and when the security verification level is lower than a preset threshold, the data item is identified as a difference item.

8. The method for analyzing dietary behavior association data of autistic patients according to claim 1, characterized in that: The step of identifying, based on the subject behavior data, a candidate stable period in which the subject behavior is in a stable state comprises: By sliding a time window of preset length on the time axis through a sliding window algorithm, continuous time periods that meet preset conditions are identified, thereby identifying candidate stable periods in which the subject behavior is in a stable state. The preset condition means that within the time window of preset length, the scores of the subject behavior data are all lower than the preset stable state threshold.

9. The method for analyzing dietary behavior association data of autistic patients according to claim 1, characterized in that: When the subject behavior data indicates that the behavior deviates from a stable state, after comparing the time series data before the deviation occurs with the safe behavior pattern file to identify a difference item in the time series data before the deviation occurs that is inconsistent with the safe behavior pattern file, the method further includes: It is determined whether the total number of the difference items exceeds an environmental change threshold preset within the system. If the total number of the difference items exceeds the environmental change threshold, the system stops generating an attribution report indicating a potential triggering event of behavioral deviation.

10. A dietary behavior correlation data analysis system for autistic patients, characterized by: The system includes: A data acquisition module, configured to acquire time series data, wherein the time series data includes data on the subject's intake, data on the subject's behavior, and multi-dimensional contextual data related to the subject's environment; a stable period establishing module, configured to identify, based on the subject behavior data, a candidate stable period in which the subject behavior is in a stable state, and establish the candidate stable period as a reference stable period based on a received confirmation instruction for the candidate stable period; A profile generating module, configured to generate a safety behavior pattern profile comprising a set of safe ingested objects and a set of safe situations based on the time series data within the reference stable period; a difference comparison module, configured to compare the time series data before the deviation occurs with the safe behavior pattern file when the subject behavior data indicates that the behavior deviates from a stable state, so as to identify difference items in the time series data before the deviation that are inconsistent with the safe behavior pattern file; A report generating module is used to generate an attribution report indicating a potential triggering event of the behavioral deviation based on the difference item.

Images