A method for recording characteristics of an infant rearing process

Abstract

The present application relates to the technical field of infant care record, and discloses a kind of infant care process feature recording method, comprising: collecting the multi-source basic data in the process of infant care, and obtaining the original record set of care;The original record set of care is split, and the event candidate fragment set is obtained;The source credibility, time credibility and consistency credibility of the event candidate fragment set are calculated, and the event credibility result is obtained;The front and rear relationship of the event credibility result is connected, and the care process chain is obtained;By comparing the crying intensity, activity intensity, sleep state and environment state before and after the care action, the care action response feature is obtained.The technical scheme of the present application uses the event front and rear relationship connection to form the care process chain, achieves the technical effect of continuous recording care process, realizes the care action, infant state change and care result associated expression.

Description

Technical Field

[0001] This invention relates to the field of infant and toddler care record technology, specifically a method for recording the characteristics of the infant and toddler care process. Background Technology

[0002] In the context of recording the daily care process of infants and toddlers during the handover between home and childcare, multiple caregivers create scattered records about feeding, sleeping, diaper changing, crying, soothing, environmental changes, and care handover. However, existing recording methods do not easily organize these scattered events into a process record with evidence sources, credibility, and before-and-after response relationships without over-reliance on manual recording or outputting medical diagnostic conclusions. This makes it difficult for different caregivers to understand the impact of care actions on changes in the infant's state based on the same objective record.

[0003] Currently, existing technologies typically employ infant recording applications, smart cameras, cry recognition devices, sleep monitoring devices, and wearable sensors for auxiliary recording. Caregivers manually input feeding times, sleep times, diaper change times, and body temperature information, or use images, sounds, and sensor data to identify infants crying, falling asleep, rolling over, or changes in posture. The manual recording results and the device recognition results are displayed on a timeline for caregivers to view.

[0004] In existing technologies, the recorded objects remain at the level of isolated events, mainly presenting the time when feeding, sleeping, crying, and diaper changing occur. There is a lack of a mechanism to connect the environmental changes before crying, caregiver actions, feeding behavior, soothing behavior, and state recovery process into a continuous process chain. As a result, the recorded data can only review events that have already occurred and cannot stably express the impact of the care process on changes in the infant's state. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention provides a method for recording the characteristics of infant and toddler care processes, thereby at least partially solving the problems mentioned in the background section.

[0006] To achieve the above objectives, the present invention provides the following technical solution: a method for recording the characteristics of infant and toddler care, comprising:

[0007] Collect multi-source basic data during the infant and toddler care process to obtain a set of original care records;

[0008] The original parenting record set is segmented to obtain a set of event candidate segments;

[0009] The source credibility, time credibility, and consistency credibility are calculated for the set of candidate event fragments to obtain the event credibility result;

[0010] By connecting the credibility results of the events sequentially, a parenting process chain can be obtained;

[0011] By comparing the intensity of crying, activity, sleep state, and environmental state before and after caregiving actions, the response characteristics of caregiving actions were obtained.

[0012] Conflict identification is performed on event records generated by different caregivers and different devices within the same time period to obtain conflict resolution results;

[0013] Integrate the parenting process chain, care action response characteristics, and conflict resolution results to output a record of infant and toddler parenting process characteristics.

[0014] Preferably, the original parenting record set is segmented to obtain a set of event candidate segments, including:

[0015] The original care record set is classified by data type to obtain sound records, status recognition records, care action records, and environmental change records;

[0016] The sound recording is segmented into silent intervals and adjacent segments are integrated to obtain crying candidate segments;

[0017] The state recognition record is segmented into state changes to obtain sleep candidate segments;

[0018] The action time is extracted from the care action record to obtain candidate care action segments;

[0019] The environmental change records are subjected to change threshold range extraction to obtain candidate environmental change segments;

[0020] The candidate segments of crying, sleeping, caregiving actions, and environmental changes are integrated to obtain a set of event candidate segments.

[0021] Preferably, the source credibility, time credibility, and consistency credibility are calculated for the set of candidate event fragments to obtain the event credibility result, including:

[0022] The source type of the candidate event fragment set is identified to obtain the source credibility.

[0023] The time reliability is obtained by verifying the difference between the acquisition time and the reception time of the event candidate fragment set.

[0024] The consistency confidence level is obtained by comparing the event support relationships within the same time period of the set of event candidate segments.

[0025] The source credibility, time credibility, and consistency credibility are weighted and fused to obtain the comprehensive credibility.

[0026] The overall credibility score is correlated with the corresponding event candidate fragments to obtain the event credibility score result.

[0027] Preferably, the event credibility results are linked sequentially to obtain a parenting process chain, including:

[0028] The event credibility results are filtered by event type to obtain the central event record;

[0029] Extract the preceding time range constraint from the event credibility results prior to the central event record to obtain the preceding event record;

[0030] The response time range constraint is extracted from the event credibility results after the central event record to obtain the subsequent event record;

[0031] The central event record, preceding event record, and subsequent event record are sequentially connected in time to obtain a single process record;

[0032] The individual process records are summarized to obtain the parenting process chain.

[0033] Preferably, the response characteristics of caregiving actions are obtained by comparing the intensity of crying, activity, sleep state, and environmental state before and after the caregiving action, including:

[0034] Care actions are located within the parenting process chain to obtain care action records;

[0035] The intensity of crying, activity, sleep state, and environmental state before the care action was recorded were extracted to obtain the pre-action state record;

[0036] The intensity of crying, activity, sleep state, and environmental state after the care actions are recorded are extracted to obtain the post-action state record;

[0037] The changes in the state records before and after the action are compared to obtain the state change results.

[0038] The response duration and state stability are extracted from the state change results to obtain the care action response characteristics.

[0039] Preferably, conflict identification is performed on event records generated by different caregivers and different devices within the same time period to obtain conflict resolution results, including:

[0040] Time overlap identification is performed on event records generated by different caregivers and different devices within the same time period to obtain overlapping event records;

[0041] The overlapping event records are subjected to state mutual exclusion identification to obtain conflict event records;

[0042] The credibility results corresponding to the conflict event records are compared to obtain the credibility difference results.

[0043] Preset difference conditions, retain the conflict event records that meet the difference conditions for the credibility difference results, and obtain the retained event records; mark the conflict event records that do not meet the difference conditions for the credibility difference results as pending confirmation, and obtain the pending confirmation event records.

[0044] The retained event records and unconfirmed event records are integrated to obtain the conflict resolution result.

[0045] Preferably, the time reliability is obtained by verifying the difference between the acquisition time and the reception time of the event candidate fragment set, including:

[0046] The collection time is extracted from the set of candidate event segments to obtain the collection time record;

[0047] The reception time is extracted from the set of candidate event segments to obtain the reception time record;

[0048] The time difference is calculated by analyzing the acquisition time record and the reception time record to obtain the time difference result;

[0049] A preset time difference condition is used to mark the time reliability of candidate event segments whose time difference results meet the time difference condition, thereby obtaining normal time records.

[0050] Time correction marks are applied to candidate event segments whose time difference results do not meet the time difference conditions to obtain time records to be corrected.

[0051] The credibility values ​​of the normal time records and the time records to be corrected are assigned to obtain the time credibility.

[0052] Preferably, the event candidate fragment set is compared with the event support relationship within the same time period to obtain the consistency confidence level, including:

[0053] Time segment matching is performed on the set of candidate event segments to obtain event records within the same segment;

[0054] The event type is identified from the same segment of event records to obtain the event records to be compared;

[0055] The corresponding relationships between the crying intensity, activity intensity, sleep state, and environmental state in the event records to be compared are compared to obtain the state correspondence results;

[0056] Support relationship tags are applied to event records that match the same event type in the state correspondence results to obtain support event records;

[0057] The event records that conflict with the same event type in the state correspondence results are marked with mutual exclusion relationships to obtain mutually exclusive event records;

[0058] The consistency reliability is obtained by assigning consistent values ​​to the supporting event records and the mutual exclusion event records.

[0059] Preferably, the response duration and state stability are extracted from the state change results to obtain care action response features, including:

[0060] Extract the end node of the care action from the state change result to obtain the response start point;

[0061] The continuous decrease in crying intensity and activity intensity after the response starting point is identified to obtain the state relief node;

[0062] The response duration is obtained by calculating the time difference between the response start point and the state relief node;

[0063] The sleep state and environmental state after the state relief node are continuously maintained to obtain a state stability result;

[0064] The response duration and state stability results are evaluated to determine the response effect, and the response effect results are obtained.

[0065] The response duration, state stability result, and response effect result are correlated to obtain the care action response characteristics.

[0066] Preferably, time overlap identification is performed on event records generated by different caregivers and different devices within the same time period to obtain overlapping event records, including:

[0067] By distinguishing the sources of event records generated by different caregivers and different devices within the same time period, event records from different sources can be obtained.

[0068] The start and end times of the event records from different sources are extracted to obtain the event time range;

[0069] The start and end times of the event time range are cross-compared to obtain the time cross-comparison result;

[0070] By setting a preset overlap condition, event records from different sources that satisfy the overlap condition in the time crossover result are grouped and marked to obtain candidate overlapping records;

[0071] The start time, end time and source information in the candidate overlapping records are verified to obtain the verified overlapping records.

[0072] The overlapping records after verification are repeatedly integrated and arranged in chronological order to obtain overlapping event records.

[0073] This invention provides a method for recording the characteristics of infant and toddler care. It has the following beneficial effects:

[0074] 1. This invention adopts a technical solution that connects the relationships between events to form a parenting process chain, thereby achieving the technical effect of continuously recording the care process and realizing the correlation between care actions, changes in infant and toddler status and care results.

[0075] 2. This invention adopts a technical solution that combines credibility calculation and conflict resolution to achieve the technical effect of multi-source record correction and realize the consistent output of manual records, equipment identification and handover records. Attached Figure Description

[0076] Figure 1 This is a schematic diagram of the process of the present invention. Detailed Implementation

[0077] To enable those skilled in the art to understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are some, but not all, of the embodiments of the present invention. Other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort should fall within the scope of protection of the present invention.

[0078] The present invention will now be described in detail with reference to the accompanying drawings:

[0079] Please see the appendix Figure 1 This invention provides a method for recording the characteristics of infant and toddler care processes, applicable to handover scenarios in families, by postpartum caregivers, by grandparents, and in childcare institutions, for recording infant and toddler feeding, sleeping, diaper changing, crying, soothing, environmental changes, and care handover processes.

[0080] In one implementation, infant identification, caregiver identification, and device identification are first established. Infant identification is used to distinguish different infants, caregiver identification is used to distinguish family members, nannies, grandparents, or childcare workers, and device identification is used to distinguish cameras, sound acquisition devices, wearable devices, environmental sensors, and mobile terminals.

[0081] Multi-source basic data includes manually input data, equipment sensing data, and handover data.

[0082] Manually entered data comes from information recorded by caregivers on mobile devices or childcare terminals. This includes feeding start time, feeding end time, feeding method, feeding amount, diaper change time, diaper condition, bowel movement status, burping time, soothing methods, complementary food introduction records, care notes, and supplementary information. The manually entered data is generated simultaneously with the caregiver's identification, the terminal identification, the collection time, and the server's receiving time.

[0083] Device perception data originates from cameras, sound acquisition devices, wearable devices, and environmental sensors. Specifically, it includes cry clips, sound intensity, activity intensity, posture recognition results, sleep status, wakefulness status, ambient temperature, ambient humidity, light intensity, and ambient noise. During data generation, device identification, acquisition time, server reception time, and device operating status are recorded simultaneously.

[0084] The handover data originates from care handover operations and specifically includes the previous caregiver's identification, the next caregiver's identification, the handover time, the previous caregiver's remarks, the next caregiver's confirmation result, unconfirmed items, and supplementary explanations. The handover data is used to identify whether records between multiple caregivers are continuous, whether there are delayed entries, and whether there are gaps in records during the handover period.

[0085] The aforementioned manually input data, device-sensed data, and handover data are collected according to the infant's identification and arranged in a unified chronological order based on the collection time to obtain a set of original care records. Each original record includes at least the record number, infant's identification, data source, record type, collection time, receipt time, record content, and source identifier.

[0086] The original care records were classified by data type to obtain sound records, status recognition records, care action records, and environmental change records.

[0087] The audio recordings are segmented into silence intervals and adjacent segments are integrated. Specifically, continuous sound intervals are identified based on changes in sound intensity. Intervals with sound intensity above a preset sound threshold are classified as sound segments, and intervals with sound intensity below the preset threshold are classified as silence intervals. When the silence intervals between adjacent sound segments meet preset integration conditions, the adjacent sound segments are integrated into the same crying candidate segment; when the silence intervals do not meet the preset integration conditions, the adjacent sound segments are retained as different crying candidate segments. A crying candidate segment includes at least the start time, end time, duration, sound intensity change, and source device identifier.

[0088] The state recognition records are segmented based on state changes. Specifically, the segments are divided according to the change nodes between sleep, wakefulness, and activity states. When the same state continuously meets preset stability conditions, the continuous state interval is regarded as a sleep candidate segment or a wakefulness candidate segment. When the state changes repeatedly in a short period of time, the original state change is first retained, and then corrected according to the consistency reliability to avoid directly treating short-term misidentifications as stable events. A sleep candidate segment includes at least the start time, end time, state type, state duration, and source device identifier.

[0089] Action timing is extracted from caregiver action records. Specifically, the timing of actions such as feeding, diaper changing, burping, cuddling, soothing, bathing, and introducing complementary foods is extracted from manually entered and handover data. When caregivers input start and end times, the time from start to end is used as candidate action segments. When caregivers input a single time point, that time point is used as the action occurrence node, and candidate action segments are generated based on a preset action duration range. Each candidate action segment includes at least the action type, action occurrence time, caregiver identification, data entry source, and a supplementary data entry marker.

[0090] Environmental change records are processed to extract change threshold intervals. Specifically, environmental temperature, humidity, light intensity, and noise are continuously monitored. When the change in any environmental state within a preset observation range meets a preset change threshold, that change interval is extracted as a candidate environmental change segment. When multiple environmental states change simultaneously, the environmental change intervals are integrated according to their temporal overlap. Each candidate environmental change segment includes at least the change type, start time, end time, direction, and magnitude.

[0091] Candidate segments of crying, sleeping, caregiving actions, and environmental changes are integrated, sorted by collection time, and the source of each segment is retained to obtain a set of event candidate segments.

[0092] The source type of the event candidate fragment set is identified to obtain source confidence. Source confidence expresses the reliability of the source of the event candidate fragment itself. Source types include real-time manual input, delayed manual entry, continuous device identification, single device identification, handover confirmation records, and unconfirmed handover records. Different initial confidence values ​​are set for different source types, and adjustments can be made based on historical confirmation results. For example, records confirmed by caregivers in real time when an event occurs may have higher source confidence, while delayed entry records may have lower source confidence; continuous device identification results may be higher than single device identification results.

[0093] Source credibility can be calculated using the following formula: ,

[0094] in, Indicates event candidate fragments The credibility of the source This indicates the data source supporting the candidate fragment of the event. Indicates the number of data sources. Indicates the source of the data Source weight, Indicates the source of the data Candidate fragments of events The level of support Indicates the source of the data The completeness of the records This represents a correction term to prevent the denominator from being zero.

[0095] The time reliability is determined by verifying the difference between the acquisition time and the reception time of the event candidate segment set. Specifically, the acquisition time and server reception time of each event candidate segment are extracted, and the time difference between them is calculated. When the time difference meets a preset time difference condition, the event candidate segment is marked as a normal time record; when the time difference does not meet the preset time difference condition, the event candidate segment is marked as a time record to be corrected or a delayed supplementary record. For time records to be corrected, they are not directly deleted, but their time reliability is reduced during process chain generation and conflict resolution.

[0096] Time reliability can be calculated using the following formula: ,

[0097] in, Indicates event candidate fragments Time reliability, Indicates the server receiving time. Indicates the collection time. This represents the time decay coefficient corresponding to the event type. This represents the penalty coefficient corresponding to the time correction marker. When there is no marker to be corrected for the event candidate segment... A lower value can be taken; when the event candidate segment is a delayed supplementary record, A higher value can be taken.

[0098] The consistency and reliability of the event candidate fragment set are obtained by comparing the supporting relationships of events within the same time period. Specifically, this involves: filtering event records within the same time period based on temporal overlap or temporal proximity; identifying the event type within the same time period; and comparing the correspondence between crying intensity, activity intensity, sleep state, and environmental state in the event records to be compared. If a crying candidate fragment corresponds to a higher sound intensity, higher activity intensity, or a state of wakefulness, a supporting relationship is formed; if a sleep candidate fragment corresponds to a lower sound intensity, lower activity intensity, and a stable environmental state, a supporting relationship is formed; if a sleep candidate fragment occurs simultaneously with continuous crying or high activity intensity, a mutually exclusive relationship is formed.

[0099] Consistency reliability can be calculated using the following formula: ,

[0100] in, Indicates event candidate fragments Consistency reliability Represents the normalization function. Indicates event candidate fragments With event candidate fragments The degree of time overlap between them This indicates a supportive or exclusive relationship. Supportive relationships are represented by positive values, while exclusive relationships are represented by negative values. Indicates event candidate fragments initial credibility, This represents a correction term to prevent the denominator from being zero.

[0101] The source credibility, time credibility, and consistency credibility are weighted and fused to obtain the comprehensive credibility. The comprehensive credibility can be calculated using the following formula:

[0102] ,

[0103] in, Indicates event candidate fragments Overall credibility Represents the normalization function. Indicates the basic bias parameter. , , This represents the weights of source credibility, time credibility, and consistency credibility. This indicates deductions incurred due to records awaiting correction, mutually exclusive records, or unconfirmed handover records. This indicates the weight of the deduction item.

[0104] The overall credibility score is correlated with the corresponding event candidate fragments to form the event credibility result. The event credibility result includes at least the event candidate fragment, source credibility, time credibility, consistency credibility, overall credibility score, and correction flag.

[0105] The event credibility results are filtered by event type to obtain central event records. Central event records can include crying, falling asleep, waking up, feeding, diaper changing, and soothing. Central event records are determined based on event type, overall credibility, and occurrence time. Events with an overall credibility lower than the preset credibility threshold are not included as central event records but can be retained as records pending confirmation.

[0106] The credibility results of events preceding the central event record are extracted under a prior time range constraint to obtain the prior event record. The prior event record is used to represent care actions, environmental changes, or state changes that may be related to changes in the infant's state before the occurrence of the central event. Care action records, environmental change records, and state records are extracted by searching backward from the start time of the central event record within a preset prior time range, while retaining the overall credibility of each record.

[0107] The event credibility results following the central event record are subjected to response time range constraints for extraction, resulting in subsequent event records. These subsequent event records represent changes in the infant's state and caregiving responses after the central event. Searching backwards according to the end time of the central event record or caregiving action record, changes in crying intensity, activity intensity, sleep state, and environmental state are extracted within a preset response time range.

[0108] The central event record, preceding event records, and subsequent event records are joined chronologically to obtain a single process record. During the joining process, events are sorted by occurrence time, and low-confidence, isolated records are filtered based on overall credibility. For cases where the same central event is associated with multiple preceding and subsequent events, records closely related to the central event are selected based on the association strength.

[0109] The correlation strength can be calculated using the following formula: ,

[0110] in, Indicates an event and events The strength of the correlation between them and Indicates an event and events Overall credibility Indicates an event and events The time interval between This represents the time decay coefficient corresponding to the combination of event types. This represents the event type matching coefficient. Event type matching relationships include: reduced crying after feeding, decreased crying after soothing, decreased activity after diaper changing, and increased crying after environmental changes.

[0111] By aggregating multiple individual process records, a parenting process chain is obtained. Each parenting process chain includes at least the preceding event record, the central event record, the subsequent event record, the event credibility result, and the chronological order.

[0112] The system locates caregiving actions within the parenting process chain, resulting in caregiving action records. These records include feeding, cuddling, burping, diaper changing, soothing, and introducing complementary foods. The system uses the start and end times of each caregiving action record as a baseline to retrieve the state records preceding the action and the state records following the action.

[0113] The intensity of crying, activity, sleep status, and environmental status before the caregiving action were extracted to obtain the pre-action state record. The pre-action state record is used to represent the infant's and toddler's state and environmental status before the caregiving action occurred. The pre-action state baseline was calculated using continuous segments preceding the action.

[0114] The intensity of crying, activity, sleep status, and environmental status after the caregiving actions were recorded, resulting in a post-action status record. This record indicates whether the infant's condition improved, stabilized, or fluctuated again after the caregiving action was completed.

[0115] The changes in the pre-action and post-action state records are compared to obtain the state change results. These results include changes in crying intensity, activity intensity, sleep state, and environmental state. Changes in crying intensity can be determined by peak sound intensity, duration, and sound fragment density; changes in activity intensity can be determined by wearable device output or video posture changes; changes in sleep state can be determined by continuous sleep recognition results; and changes in environmental state can be determined by changes in ambient temperature, humidity, light intensity, and noise.

[0116] The result of the state change can be expressed by the following formula:

[0117] ,

[0118] in, Indicates caregiving actions The corresponding response change value, This indicates the degree to which the intensity of crying has decreased. Indicates the degree of decrease in activity intensity. Indicates the stability of sleep state. Indicates the degree of environmental stability. This indicates an unstable factor that may lead to further crying or waking up. , , , , Indicates the weight.

[0119] The response duration and stability of the state change results are extracted to obtain the care action response features. Specifically: the end node of the care action is extracted to obtain the response start point; the continuous decrease in crying intensity and activity intensity after the response start point is identified to obtain the state relief node; the time difference between the response start point and the state relief node is calculated to obtain the response duration; the continuous maintenance of sleep state and environmental state after the state relief node is determined to obtain the state stability result; the response duration and state stability result are used to determine the response effect to obtain the response effect result.

[0120] The response results can be categorized into effective responses, partially effective responses, and no significant responses. An effective response is defined as a decrease in crying intensity and activity level, coupled with a sustained sleep or quiet state. A partially effective response is defined as an improvement in some indicators. No significant response is defined as no improvement in crying intensity, activity level, or sleep state. Correlating response duration, stability of state, and response results yields the characteristics of caregiving action responses.

[0121] Event records generated by different caregivers and different devices within the same time period are categorized by source to obtain event records from different sources. These event records from different sources include manual records entered by different caregivers, device records generated by different devices, and handover confirmation records.

[0122] Start and end times are extracted from event records from different sources to obtain the event time range. For events with clear start and end times, the original start and end times are used directly; for events with a single time point, the event time range is generated by combining the event type; for delayed events, the entry time and the actual occurrence time are retained to avoid delayed entries directly overwriting real-time device records.

[0123] The start and end times of the event time ranges are cross-referenced to obtain the time cross-reference results. When the time ranges of two or more event records overlap or meet a preset proximity relationship, they are considered records that may have an overlapping relationship. The degree of time overlap can be calculated using the following formula:

[0124] ,

[0125] in, Indicates event logging With event log The degree of time overlap, , Indicates event logging The start and end times, , Represents event log The start and end times, This represents a correction term to prevent the denominator from being zero.

[0126] Preset overlap conditions are used to group event records from different sources that meet the overlap conditions in the time crossover results, resulting in candidate overlapping records. The start time, end time, and source information in the candidate overlapping records are verified to check the correspondence, and redundant records caused by duplicate uploads or duplicate records from the same device are eliminated. The verified records are then integrated and arranged in chronological order to obtain the overlapping event records.

[0127] Overlapping event records are identified through state mutual exclusion identification to obtain conflict event records. State mutual exclusion relationships include the simultaneous existence of a sleeping state and a continuous crying state within the same time period; the simultaneous existence of records showing completed feeding and unresolved crying before feeding within the same time period; and the simultaneous existence of records showing calmed and stable state and records showing continuously increasing activity intensity within the same time period. Instead of directly deleting any mutually exclusive record, the event credibility results corresponding to each mutually exclusive record are further read.

[0128] The credibility difference results are obtained by comparing the credibility results corresponding to the conflict event records. The credibility difference results can be calculated using the following formula:

[0129] ,

[0130] in, Record of conflict events Record of Conflicts The results show differences in credibility between the two. , Record of conflict events Record of Conflicts Overall credibility Record of conflict events Record of Conflicts The degree of time overlap, This indicates the strength of mutual exclusion between states.

[0131] Pre-defined discrepancy conditions are used to determine the retention of conflict event records where the credibility difference results meet these conditions. Specifically: when the credibility difference between conflict events is sufficient to distinguish reliability, the event record with the higher overall credibility is retained, while the event record with the lower overall credibility is treated as supplementary explanation or low-credibility record; when the credibility difference results do not meet the discrepancy conditions, no conflict event record is deleted, but instead marked as pending confirmation, resulting in pending confirmation event records. Pending confirmation event records can be presented to caregivers for confirmation during the next handover.

[0132] The retained event records and pending event records are integrated to obtain the conflict resolution results. The conflict resolution results include at least the conflict event records, overlapping time ranges, confidence difference results, retained event records, pending event records, and processing tags.

[0133] Integrating the parenting process chain, caregiving response characteristics, and conflict resolution results, the system outputs a record of infant and toddler parenting process characteristics. This record may include an event timeline, a process chain summary, credibility indicators, caregiving response results, and handover prompts.

[0134] The event timeline displays feeding, sleep, diaper changing, crying, soothing, and environmental changes in chronological order. The process chain summary shows the relationships between preceding, central, and subsequent event records. The credibility indicator displays the source credibility, time credibility, consistency credibility, and overall credibility of each record. The care action response result shows changes in crying intensity, activity level, sleep status, and environmental status after a specific care action. The handover prompts display pending confirmation records, delayed entries, conflicting records, and process segments requiring the next caregiver's attention.

[0135] In one output method, the characteristics of infant and toddler care process can be recorded using the following data structure:

[0136] Record number; infant / toddler identification; record date; preceding event record; central event record; subsequent event record; care action response characteristics; event credibility result; conflict resolution result; handover prompts; update time.

[0137] Through the above processing, a process record with source, credibility, context, and response results is output, enabling multiple caregivers to understand the daily parenting rhythm and care effects of infants and young children based on the same set of records.

[0138] In a preferred embodiment, the historical parenting process chain can be statistically analyzed to form an individualized parenting rhythm baseline. Specifically, according to the historical parenting process chain of the same infant, the occurrence time intervals, durations, and adjacent intervals of feeding, sleeping, diaper changing, crying, and soothing are statistically analyzed; when the number of valid records meets the preset sample conditions, the parenting rhythm baseline corresponding to the infant is generated; when the number of valid records does not meet the preset sample conditions, the original process records and insufficient sample markers are output.

[0139] Individualized parenting rhythm baselines are used to help determine lead time ranges, response time ranges, time difference conditions, overlap conditions, and difference conditions. This approach avoids directly using uniform standards to evaluate the differences in parenting processes among different infants and young children, ensuring that the process chain generation and care action response characteristics are aligned with the actual care rhythms of specific infants and young children.

[0140] By adopting the above specific implementation methods, the present invention can transform manually input data, device-sensed data, and handover data into a set of event candidate fragments, and calculate the event credibility result through source credibility, time credibility, and consistency credibility, thereby reducing the impact of omissions, misrecordings, delayed supplementary recordings, and device misidentification on the reliability of records.

[0141] Furthermore, this invention connects the event credibility results into a parenting process chain by using the pre-event time range and the response time range, transforming the recorded content from isolated events into a continuous expression between pre-event, central event, and post-event, which can reflect the impact of care actions on changes in the infant's state.

[0142] Furthermore, the present invention generates care action response features by means of response duration, state stability results and response effect results, which can record the actual state changes after the care action, rather than simply showing that the care action has occurred.

[0143] Furthermore, this invention generates conflict resolution results through time overlap identification, state mutual exclusion identification, and credibility difference comparison, enabling conflicts between records from multiple caregivers and multiple devices to be retained, corrected, or prompted for confirmation, thereby improving the credibility and usability of handover records.

[0144] Example 1:

[0145] This embodiment applies to the transition between home and childcare. Infants and toddlers are cared for by childcare providers in the morning, with data collected simultaneously from manual input, device sensing, and environmental changes. Existing recording methods typically only display "when crying," "when feeding," and "when soothing," allowing caregivers to review individual events and making it difficult to determine whether environmental changes occurred before crying or whether the infant's condition truly recovered after soothing.

[0146] In this invention, multi-source basic data are collected during the infant and toddler care process to obtain a set of original care records; the original care records are segmented to obtain a set of candidate event segments; then the event credibility results are calculated, and the event credibility results are linked together to obtain a care process chain.

[0147] Table 1: Selected Records from the Collection of Original Parenting Records

[0148] 08:05 Environmental sensors Environmental change record Environmental noise is caused by Rise to 08:08 wearable devices Status recognition record Activity intensity from Rise to 08:12 Sound acquisition equipment Sound recording The child cried continuously, and the intensity of the crying was high. 08:14 Childcare providers Caregiver Action Record Perform hugging and comforting 08:18 Sound acquisition equipment Sound recording The intensity of crying decreased to 08:21 wearable devices Status recognition record Activity intensity decreased to 08:25 Camera Status recognition record Stable sleep

[0149] When processing the above records, the sound records are first segmented into silent intervals and adjacent segments are integrated to obtain crying candidate segments; the state recognition records are segmented into state change segments to obtain sleep candidate segments; the caregiver action records are extracted into action time segments to obtain caregiver action candidate segments; and the environmental change records are extracted into change threshold interval segments to obtain environmental change candidate segments.

[0150] Next, using the crying candidate segment formed at 8:12 as the central event record, the credibility results of the events before the central event record are subjected to a prior time range constraint extraction to obtain the prior event records corresponding to the increase in environmental noise and the increase in activity intensity; the credibility results of the events after the central event record are subjected to a response time range constraint extraction to obtain the subsequent event records corresponding to the comforting and soothing, the decrease in crying intensity, the decrease in activity intensity, and the stabilization of sleep state.

[0151] Table 2: Results of the parenting process chain generated by this invention

[0152] Pre-event log Increased environmental noise and increased activity intensity As evidence of changes in state before crying occurs Central Event Log Infants and toddlers cry continuously, and the intensity of the crying is... As the central event of this process chain Post-event logging After being comforted and soothed, the intensity of crying decreased to Record of state changes after caregiving actions Post-event logging Activity intensity reduced to Stable sleep As a record of care outcomes Output Environmental changes and increased activity – crying – comforting and soothing – improved condition – stable sleep Obtain the continuous parenting process chain

[0153] As can be seen from the above processing, if the existing isolated event recording method is used, it can only show "08:05 Noise increases", "08:12 Crying", "08:14 Soothing", and "08:25 Falling asleep". Caregivers need to judge for themselves whether there is a relationship between the events, and different caregivers may have different understandings.

[0154] This invention integrates environmental changes, infant states, caregiving actions, and care outcomes into a continuous process chain by linking the credibility of events before and after they occur. This clearly expresses "what happened before the crying, what actions the caregiver took, and how the infant's state changed after the actions." Therefore, this embodiment verifies that the technical solution of this invention, which uses the linking of events before and after they occur to form a parenting process chain, can achieve the technical effect of continuously recording the care process and realize the correlation between caregiving actions, changes in the infant's state, and care outcomes.

[0155] Example 2:

[0156] This embodiment applies to a scenario where multiple caregivers alternate care. Infants and toddlers are cared for sequentially by family caregivers and childcare caregivers in the afternoon, with simultaneous recordings from manual input, audio capture devices, cameras, and handover confirmation. Due to manual re-recording, device upload delays, and device recognition errors, conflicting event records may appear within the same time period.

[0157] In this invention, the source type of the event candidate fragment set is identified to obtain the source credibility; the difference between the acquisition time and the reception time of the event candidate fragment set is verified to obtain the time credibility; the support relationship of events within the same time period is compared to obtain the consistency credibility; and then the source credibility, time credibility, and consistency credibility are weighted and fused to obtain the event credibility result.

[0158] Table 1: Multi-source event records and reliability results within the same time period

[0159] 14:30-14:38 Camera Infants and young children are asleep 0.82 0.91 0.38 0.70 14:31-14:37 Sound acquisition equipment Continuing to cry 0.86 0.94 0.88 0.89 14:33-14:36 wearable devices Increased activity intensity 0.81 0.90 0.84 0.85 14:40 Supplementary recording Family caregivers I fell asleep at 2:30 PM. 0.62 0.55 0.35 0.50 14:42 Childcare providers The infant was still crying during the handover. 0.76 0.88 0.86 0.83

[0160] In the aforementioned data, both camera recordings and manually recorded data show that the infant was asleep. However, sound capture devices, wearable devices, and caregiver handover records all indicate that the infant was still crying or becoming more active. If current technology were to display these records simultaneously, the next caregiver would see both conflicting records of "asleep" and "still crying," making it difficult to determine which record is reliable.

[0161] This invention first identifies overlapping events by analyzing event records generated by different caregivers and devices within the same time period. Then, it identifies mutually exclusive states within these overlapping event records to obtain conflicting event records. For cases where "sleep state" and "continuous crying" coexist, these are identified as mutually exclusive states, and the corresponding event credibility results are compared for discrepancies.

[0162] Table 2: Results of Conflict Identification and Resolution

[0163] Camera sleep recording and sound crying recording Handing over crying records and vocal crying records Sleep duration 0.7, crying duration 0.89 Records of crying are retained, while sleep records are marked as low-confidence references. Manually record sleep status and device crying status. The supplementary recording was delayed and inconsistent with the audio and activity intensity records. Sleeping record: 0.5; Crying record: 0.89 Real-time device records will not be overwritten; supplementary records will be treated as pending confirmation records. Manually record sleep status and device crying status. Sleep state and increased activity level are inconsistent Sleep record 0.7, activity record 0.85 Records of increased activity levels are retained, and a prompt is made to review sleep records. Handing over crying records and vocal crying records The events are of the same type and occur within a similar time period. Handover record 0.83, audio record 0.89 The two elements form a supportive relationship, increasing the credibility of the crying incident.

[0164] The above process does not simply delete conflicting records, nor does it synchronize all records. Instead, it establishes a comprehensive credibility score based on source credibility, time credibility, and consistency credibility, and combines time overlap identification and state mutual exclusion identification to generate conflict resolution results. Event records with significantly high credibility are retained and added to the infant care process feature record; event records with insufficient credibility differences or requiring manual judgment are added as pending confirmation event records to the handover prompt item.

[0165] In this embodiment, the output handover content can be expressed as follows: From 14:31 to 14:37, both the sound acquisition device and the wearable device support the infant being in a state of crying and increased activity, and the handover record of the caregiver is consistent with the records of the aforementioned devices; the manually recorded sleep record is inconsistent with the multi-source record, and there is a lag in the recording, therefore it is marked as a record to be confirmed. The next caregiver can directly obtain the reliable status record for this time period, as well as the contradictory records that need further confirmation.

[0166] Therefore, this embodiment verifies that the technical solution of combining credibility calculation and conflict resolution adopted by the present invention can achieve the technical effect of multi-source record correction and realize the consistent output of manual records, equipment identification and handover records.

[0167] Comparative Experiment 1:

[0168] Experimental Scenario: This experiment is set up in a home-childcare handover scenario.

[0169] Infants and toddlers were cared for by caregivers in the morning, during which environmental sensors, sound recording devices, wearable devices, cameras, and manual input records from caregivers were simultaneously collected. The experiment focused on verifying whether the present invention could connect scattered records of environmental changes, crying, caregiving actions, and recovery into a continuous chain of parenting processes.

[0170] The experimental subjects were the same infants and toddlers whose care records were generated during the same morning care period. The experimental data included records of changes in environmental noise, changes in activity intensity, crying sounds, soothing and comforting experiences, and sleep patterns.

[0171] Two processing methods are set up:

[0172] control group Existing isolated event logging methods Display the occurrence time and content of each event according to the timeline. experimental group Method of the present invention The credibility of candidate event segments is calculated, and the relationships between them are connected to generate a parenting process chain.

[0173] Experimental steps:

[0174] Collect basic data from multiple sources within the same care period to form a set of original parenting records;

[0175] Classify sound recordings, status recognition records, care action records, and environmental change records;

[0176] The sound recordings are segmented into silence intervals, the state recognition records are segmented into state changes, the action time of the caregiver action records is extracted, and the change threshold range of the environmental change records is extracted.

[0177] The control group outputs the event timeline in chronological order;

[0178] The experimental group calculated the source credibility, time credibility, and consistency credibility of the candidate event fragments to obtain the event credibility results;

[0179] The experimental group used the crying event as the central event, extracting records of environmental changes and increased activity forward, and records of soothing actions, crying relief, decreased activity, and sleep stability backward.

[0180] Compare the differences between the two sets of output results in terms of event association expression, care action response expression, and handover comprehensibility.

[0181] Experimental data table:

[0182] 08:05 Environmental sensors Environmental change record Environmental noise is caused by Rise to 08:08 wearable devices Status recognition record Activity intensity from Rise to 08:12 Sound acquisition equipment Sound recording The child cried continuously, and the intensity of the crying was high. 08:14 Childcare providers Caregiver Action Record Perform hugging and comforting 08:18 Sound acquisition equipment Sound recording The intensity of crying decreased to 08:21 wearable devices Status recognition record Activity intensity decreased to 08:25 Camera Status recognition record Stable sleep

[0183] Experimental Results Table:

[0184] Event presentation method Displays the time of each event. A process chain is formed based on preceding events, central events, and subsequent events. The experimental group was able to express the relationship between events before and after. Help in determining the cause of crying The message indicated that crying occurred at 08:12. The crying was preceded by increased ambient noise and heightened activity. The experimental group was able to retain evidence of changes in state before crying. Caregiver Response Expression The display shows that the reassurance was carried out at 08:14. The child showed a decrease in crying intensity, a decrease in activity level, and stable sleep after being soothed. The experimental group was able to express changes in state after caregiving actions. Handover understandability The next caregiver needs to determine the relevance of the event themselves. The output is: "Environmental change and increased activity - crying - comforting and soothing - improved condition - stable sleep". The experimental group reduced the cost of handover and understanding. Output completeness Isolated Event Timeline Continuous parenting process chain and care action response characteristics The experimental group recorded complete information.

[0185] The experimental results showed that the control group could only output isolated events such as "increased environmental noise," "crying," "soothing," and "falling asleep," and could not stably express the sequential relationships between events. By connecting the sequential relationships of the event credibility results, the experimental group was able to form a continuous parenting process chain of "environmental change and increased activity - crying - cuddling and soothing - relief - stable sleep."

[0186] Therefore, this comparative experiment verifies that the event-before-after relationship connection method adopted in Example 1 of the present invention can correlate care actions, changes in infant state and care results, and achieve the technical effect of continuously recording the care process.

[0187] Comparative Experiment 2:

[0188] Experimental scenario: This experiment was set up in a scenario of multiple caregivers taking turns providing care.

[0189] Infants and toddlers are cared for sequentially by family caregivers and childcare providers in the afternoon, with simultaneous manual record-keeping, device identification, sound capture, wearable device detection, and handover confirmation records. The experiment focuses on verifying whether the present invention can identify conflicts and output reliable resolution results when contradictory records exist within the same time period.

[0190] The experiment involved recordings of multiple events from the same infant between 14:30 and 14:42. During this time period, there were records of "sleep state", "continuous crying", "increased activity intensity" and "still crying at handover", which showed characteristics of mutually exclusive states and conflicting sources.

[0191] Two processing methods are set up:

[0192] control group Existing synchronous display method Display all records, both manual and device-generated, without performing credibility calculations or conflict resolution. experimental group Method of the present invention Calculate source credibility, time credibility, and consistency credibility, and perform time overlap identification, state mutual exclusion identification, and conflict resolution.

[0193] Experimental steps:

[0194] Collect multi-source records generated by cameras, audio acquisition devices, wearable devices, family caregivers, and childcare providers within the same time period;

[0195] Distinguish the source of each event record and extract the event start time, end time, collection time, and reception time;

[0196] The source credibility, time credibility, and consistency credibility of the event candidate segments are calculated and weighted and fused to obtain the comprehensive credibility.

[0197] The control group displays all records synchronously in chronological order without making any conflict judgments.

[0198] The experimental group performed time overlap identification on event records within the same time period to obtain overlapping event records;

[0199] The experimental group performed state mutual exclusion identification on the overlapping event records, and identified records such as "sleep state" and "continuous crying" and "increased activity intensity" as conflicting events;

[0200] The experimental group reads the overall credibility of the conflict events and compares the credibility differences.

[0201] Based on the results of the confidence level differences, events with higher confidence levels are retained as events, while events with lower confidence levels or those requiring manual confirmation are marked as low-confidence references or records to be confirmed.

[0202] Compare the differences between the two sets of output results in terms of conflict identification, record correction, and handover prompts.

[0203] Experimental data and reliability results table:

[0204] 14:30-14:38 Camera Infants and young children are asleep 0.82 0.91 0.38 0.70 14:31-14:37 Sound acquisition equipment Continuing to cry 0.86 0.94 0.88 0.89 14:33-14:36 wearable devices Increased activity intensity 0.81 0.90 0.84 0.85 14:40 Supplementary recording Family caregivers I fell asleep at 2:30 PM. 0.62 0.55 0.35 0.50 14:42 Childcare providers The infant was still crying during the handover. 0.76 0.88 0.86 0.83

[0205] Conflict Identification and Resolution Results Table:

[0206] Camera sleep recording and sound crying recording Handing over crying records and vocal crying records Sleep duration 0.7, crying duration 0.89 Records of crying are retained, while sleep records are marked as low-confidence references. Manually record sleep status and device crying status. The supplementary recording was delayed and inconsistent with the audio and activity intensity records. Sleeping record: 0.5; Crying record: 0.89 Real-time device records will not be overwritten; supplementary records will be treated as pending confirmation records. Manually record sleep status and device crying status. Sleep state and increased activity level are inconsistent Sleep record 0.7, activity record 0.85 Records of increased activity levels are retained, and a prompt is made to review sleep records. Handing over crying records and vocal crying records The events are of the same type and occur within a similar time period. Handover record 0.83, audio record 0.89 The two elements form a supportive relationship, increasing the credibility of the crying incident.

[0207] Comparison of experimental results:

[0208] Conflict identification It displays both "asleep" and "still crying" simultaneously. Identifying sleep states and persistent crying states are mutually exclusive. The experimental group was able to find contradictory records. Credibility judgment No distinction is made between real-time device recordings and delayed supplementary recordings. The overall credibility is obtained based on source credibility, time credibility, and consistency credibility. Reliability of experimental group energy recording Conflict handling All records are displayed synchronously, requiring manual judgment. High-confidence records are retained, while low-confidence records are used as references or pending verification. The experimental group can output the digestion results. Handover Notice The next caregiver needs to assess the conflict themselves. The message explicitly states that "the crying records from 14:31 to 14:37 are highly reliable; if the child has fallen asleep, confirmation is required before recording." The experimental group improved handover availability. Multi-source consistent output Contradictory records from multiple sources The audio recordings, activity records, and handover records form a supporting relationship. The experimental group achieved multi-source recording correction.

[0209] The experimental results showed that the control group displayed manual records, equipment records and handover records simultaneously, which caused the next caregiver to see conflicting information such as "asleep" and "still crying" at the same time, making it impossible to directly determine which record was reliable.

[0210] The experimental group calculated the overall credibility by considering source credibility, time credibility, and consistency credibility. By combining time overlap identification, state mutual exclusion identification, and credibility difference comparison, it was able to retain crying records and activity enhancement records with higher credibility, while marking sleep records and delayed supplementary records with lower credibility as low-credibility references or records to be confirmed.

[0211] Therefore, this comparative experiment verifies that the combination of credibility calculation and conflict resolution adopted in Example 2 of the present invention can achieve the technical effect of multi-source record correction and realize the consistent output of manual records, equipment identification and handover records.

[0212] Embodiments of the present invention have been presented and described. It will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to the embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A method for recording the characteristics of infant and toddler care, characterized in that, include: Collect multi-source basic data during the infant and toddler care process to obtain a set of original care records; The original parenting record set is segmented to obtain a set of event candidate segments; The source credibility, time credibility, and consistency credibility are calculated for the set of candidate event fragments to obtain the event credibility result; By connecting the credibility results of the events sequentially, a parenting process chain can be obtained; By comparing the intensity of crying, activity, sleep state, and environmental state before and after caregiving actions, the response characteristics of caregiving actions were obtained. Conflict identification is performed on event records generated by different caregivers and different devices within the same time period to obtain conflict resolution results; Integrate the parenting process chain, care action response characteristics, and conflict resolution results to output a record of infant and toddler parenting process characteristics.

2. The feature recording method according to claim 1, characterized in that, The original parenting record set is segmented to obtain a set of event candidate segments, including: The original care record set is classified by data type to obtain sound records, status recognition records, care action records, and environmental change records; The sound recording is segmented into silent intervals and adjacent segments are integrated to obtain crying candidate segments; The state recognition record is segmented into state changes to obtain sleep candidate segments; The action time is extracted from the care action record to obtain candidate care action segments; The environmental change records are subjected to change threshold range extraction to obtain candidate environmental change segments; The candidate segments of crying, sleeping, caregiving actions, and environmental changes are integrated to obtain a set of event candidate segments.

3. The feature recording method according to claim 1, characterized in that, The source credibility, time credibility, and consistency credibility are calculated for the set of candidate event fragments to obtain the event credibility result, including: The source type of the candidate event fragment set is identified to obtain the source credibility. The time reliability is obtained by verifying the difference between the acquisition time and the reception time of the event candidate fragment set. The consistency confidence level is obtained by comparing the event support relationships within the same time period of the set of event candidate segments. The source credibility, time credibility, and consistency credibility are weighted and fused to obtain the comprehensive credibility. The overall credibility score is correlated with the corresponding event candidate fragments to obtain the event credibility score result.

4. The feature recording method according to claim 1, characterized in that, By connecting the event credibility results sequentially, a parenting process chain is obtained, including: The event credibility results are filtered by event type to obtain the central event record; Extract the preceding time range constraint from the event credibility results prior to the central event record to obtain the preceding event record; The response time range constraint is extracted from the event credibility results after the central event record to obtain the subsequent event record; The central event record, preceding event record, and subsequent event record are sequentially connected in time to obtain a single process record; The individual process records are summarized to obtain the parenting process chain.

5. The feature recording method according to claim 1, characterized in that, By comparing the intensity of crying, activity, sleep state, and environmental state before and after caregiving actions, the response characteristics of caregiving actions are obtained, including: Care actions are located within the parenting process chain to obtain care action records; The intensity of crying, activity, sleep state, and environmental state before the care action was recorded were extracted to obtain the pre-action state record; The intensity of crying, activity, sleep state, and environmental state after the care actions are recorded are extracted to obtain the post-action state record; The changes in the state records before and after the action are compared to obtain the state change results. The response duration and state stability are extracted from the state change results to obtain the care action response characteristics.

6. The feature recording method according to claim 1, characterized in that, Conflict identification is performed on event records generated by different caregivers and different devices within the same time period to obtain conflict resolution results, including: Time overlap identification is performed on event records generated by different caregivers and different devices within the same time period to obtain overlapping event records; The overlapping event records are subjected to state mutual exclusion identification to obtain conflict event records; The credibility results corresponding to the conflict event records are compared to obtain the credibility difference results. Preset difference conditions, retain the conflict event records that meet the difference conditions for the credibility difference results, and obtain the retained event records; mark the conflict event records that do not meet the difference conditions for the credibility difference results as pending confirmation, and obtain the pending confirmation event records. The retained event records and unconfirmed event records are integrated to obtain the conflict resolution result.

7. The feature recording method according to claim 3, characterized in that, The acquisition time and reception time difference are verified for the set of candidate event segments to obtain time reliability, including: The collection time is extracted from the set of candidate event segments to obtain the collection time record; The reception time is extracted from the set of candidate event segments to obtain the reception time record; The time difference is calculated by analyzing the acquisition time record and the reception time record to obtain the time difference result; A preset time difference condition is used to mark the time reliability of candidate event segments whose time difference results meet the time difference condition, thereby obtaining normal time records. Time correction marks are applied to candidate event segments whose time difference results do not meet the time difference conditions to obtain time records to be corrected. The credibility values ​​of the normal time records and the time records to be corrected are assigned to obtain the time credibility.

8. The feature recording method according to claim 3, characterized in that, The event candidate fragment set is compared with the event support relationship within the same time period to obtain the consistency confidence level, including: Time segment matching is performed on the set of candidate event segments to obtain event records within the same segment; The event type is identified from the same segment of event records to obtain the event records to be compared; The corresponding relationships between the crying intensity, activity intensity, sleep state, and environmental state in the event records to be compared are compared to obtain the state correspondence results; Support relationship tags are applied to event records that match the same event type in the state correspondence results to obtain support event records; The event records that conflict with the same event type in the state correspondence results are marked with mutual exclusion relationships to obtain mutually exclusive event records; The consistency reliability is obtained by assigning consistent values ​​to the supporting event records and the mutual exclusion event records.

9. The feature recording method according to claim 5, characterized in that, The response duration and state stability of the state change results are extracted to obtain care action response features, including: Extract the end node of the care action from the state change result to obtain the response start point; The continuous decrease in crying intensity and activity intensity after the response starting point is identified to obtain the state relief node; The response duration is obtained by calculating the time difference between the response start point and the state relief node; The sleep state and environmental state after the state relief node are continuously maintained to obtain a state stability result; The response duration and state stability results are evaluated to determine the response effect, and the response effect results are obtained. The response duration, state stability result, and response effect result are correlated to obtain the care action response characteristics.

10. The feature recording method according to claim 6, characterized in that, Time overlap identification is performed on event records generated by different caregivers and different devices within the same time period to obtain overlapping event records, including: By distinguishing the sources of event records generated by different caregivers and different devices within the same time period, event records from different sources can be obtained. The start and end times of the event records from different sources are extracted to obtain the event time range; The start and end times of the event time range are cross-compared to obtain the time cross-comparison result; By setting a preset overlap condition, event records from different sources that satisfy the overlap condition in the time crossover result are grouped and marked to obtain candidate overlapping records; The start time, end time and source information in the candidate overlapping records are verified to obtain the verified overlapping records. The overlapping records after verification are repeatedly integrated and arranged in chronological order to obtain overlapping event records.

Images