State slice generation method and apparatus

Abstract

The application relates to the technical field of data processing, and discloses a state slice generation method and device, which continuously collects multi-source data of a user, and extracts features from the multi-source data to obtain multi-source features; the current state of the user is identified based on the multi-source features to obtain a physiological state, an emotional state and a behavior state of the user; the state changes of the user are monitored based on the physiological state, the emotional state and the behavior state, wherein the state changes include at least one of a physiological state change, an emotional state change and a behavior state change; and a state slice corresponding to a key state change in the state changes is generated in response to the key state change. The application can convert continuous data into a structured state slice, and enables the user to identify self-state changes.

Description

Technical Field

[0001] This invention relates to the field of data processing technology, and in particular to a method and apparatus for generating state slices. Background Technology

[0002] With the rapid development of wearable devices, users can easily obtain health data such as heart rate, sleep, and activity levels. However, existing technologies still have many shortcomings in helping users identify changes in their own state, mainly in the following aspects: First, most existing health products only display raw data curves such as heart rate, HRV, and blood oxygen. Although users can see numerical changes, they find it difficult to understand the true physical and emotional state reflected by this data. For example, they cannot determine whether they are currently under high pressure, relaxed, focused, or fatigued. Second, even if users can see fluctuations in data in a single dimension, they cannot understand how their state changes over time. For example, they cannot identify when stress begins to rise, when emotions fluctuate, or when focus is broken—key moments of state change. Third, moments of state change that are of significant value in a user's life, such as the shift from relaxed to high pressure, or from focused to fatigue, cannot be automatically captured and recorded. In addition, users not only find it difficult to view changes in their state across different time dimensions, but also struggle to analyze the influence of different social interactions on their state. Summary of the Invention

[0003] This invention provides a method and apparatus for generating state slices to solve the technical problem that existing wearable devices cannot recognize changes in the user's own state.

[0004] Firstly, a method for generating state slices is provided, including: Continuously collect multi-source data from users, and extract features from the multi-source data to obtain multi-source features; Based on the multi-source features, the user's current state is identified to obtain the user's physiological state, emotional state, and behavioral state. Based on the physiological state, the emotional state, and the behavioral state, monitor the user's state changes, wherein the state changes include at least one of physiological state changes, emotional state changes, and behavioral state changes. In response to key state changes in the state changes, a state slice corresponding to the key state change is generated.

[0005] Secondly, a state slice generation apparatus is provided, comprising: The data acquisition and extraction unit is used to continuously acquire multi-source data from users and extract features from the multi-source data to obtain multi-source features. The state recognition unit is used to identify the user's current state based on the multi-source features, and to obtain the user's physiological state, emotional state and behavioral state. A monitoring unit is used to monitor changes in a user's state based on the physiological state, the emotional state, and the behavioral state, wherein the changes in state include at least one of changes in physiological state, changes in emotional state, and changes in behavioral state. The generation unit is used to generate a state slice corresponding to a key state change in response to the state change.

[0006] The aforementioned state slice generation method and apparatus continuously collect multi-source data from users and extract features from the multi-source data to obtain multi-source features. Based on the multi-source features, the user's current state is identified to obtain the user's physiological state, emotional state, and behavioral state. Based on the physiological state, emotional state, and behavioral state, changes in the user's state are monitored, wherein the state changes include at least one of physiological state changes, emotional state changes, and behavioral state changes. In response to key state changes in the state changes, a state slice corresponding to the key state change is generated. In this invention, by continuously collecting multi-source data and extracting features, the user's physiological state, emotional state, and behavioral state are identified based on the extracted multi-source features, thereby monitoring changes in the user's state. When a key state change is detected, a corresponding state slice is generated. Thus, continuous data is transformed into structured state slices, enabling users to recognize their own state changes. Attached Figure Description

[0007] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the description of the embodiments of the present invention will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0008] Figure 1 This is a flowchart illustrating a state slice generation method in one embodiment of the present invention; Figure 2 yes Figure 1 A schematic diagram of a specific implementation of step S120; Figure 3 yes Figure 1 A schematic diagram of a specific implementation of step S140; Figure 4 This is a flowchart illustrating a state slice generation method in another embodiment of the present invention; Figure 5This is a schematic block diagram of a state slice generation device in one embodiment of the present invention; Figure 6 This is a schematic block diagram of a wearable device provided in an embodiment of the present invention. Detailed Implementation

[0009] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0010] The state slice generation method provided in this invention can be applied to wearable devices, such as smartwatches and smart rings. Currently, existing wearable devices cannot recognize changes in the user's own state. To address this problem, this invention proposes a state slice generation method. This method continuously collects multi-source data and extracts features. Based on the extracted multi-source features, it identifies the user's physiological state, emotional state, and behavioral state, thereby monitoring changes in the user's state. When a key state change is detected, a corresponding state slice is generated. Thus, continuous data is transformed into structured state slices, enabling the user to recognize changes in their own state. The invention will be described in detail below through specific embodiments.

[0011] Please see Figure 1 As shown, Figure 1 A schematic flowchart of a state slice generation method provided in an embodiment of the present invention includes the following steps: S110-S140.

[0012] S110. Continuously collect multi-source data from users and extract features from the multi-source data to obtain multi-source features.

[0013] Specifically, the multi-source data includes health data, behavioral data, voice data, and contextual information. The health data includes at least one of heart rate, HRV, blood oxygen, sleep duration, respiratory rate, body temperature, and inflammation prediction indicators. The behavioral data includes at least one of steps, sedentary time, exercise behavior, mobile phone usage time, App usage type, and travel trajectory. The voice data includes at least one of user-initiated recordings, periodically collected voice recordings, and scene-triggered recordings. The contextual information includes at least one of time, location, weather, schedule, public events, and social interaction objects. It should be noted that the wearable device continuously collects user health data such as heart rate, HRV, blood oxygen, sleep duration, respiratory rate, body temperature, and inflammation prediction indicators through built-in photoelectric sensors and temperature sensors.

[0014] The wearable device collects user behavioral data such as steps, movement behavior, sedentary time, and travel trajectory through its built-in accelerometer and gyroscope. It also collects user voice data through its built-in microphone, supporting various collection methods including user-initiated recording, system-periodic recording, and scene-triggered recording. The wearable device obtains system time and location information through its positioning module. After connecting to a mobile phone via Bluetooth, the wearable device synchronously acquires contextual information such as weather, schedules, public events, and social interactions. It should also be noted that the multi-source features include physiological features, behavioral features, voice features, and contextual features.

[0015] Step S110 specifically includes: extracting the physiological features from the health data using a physiological feature extraction method; extracting the behavioral features from the behavioral data using a behavioral feature extraction method; extracting the speech features from the speech data using a speech feature extraction method; and extracting the context features from the context information using a context feature extraction method.

[0016] It should be noted that the physiological feature extraction method includes at least one of time-domain analysis, frequency-domain analysis, and statistical analysis; the behavioral feature extraction method includes at least one of time-series statistics, behavioral pattern recognition, and activity frequency calculation. The speech feature extraction method includes at least one of speech signal processing, pre-trained speech model analysis, and acoustic feature extraction. The context feature extraction method includes at least one of information mapping, context association analysis, and entity recognition. It should also be noted that the physiological features include at least one of heart rate change trend, HRV fluctuation characteristics, blood oxygen fluctuation characteristics, sleep quality score, respiratory rate change characteristics, and body temperature change trend characteristics. The behavioral features include at least one of activity intensity level, step count change rate, sedentary frequency, device usage habit pattern, and travel activity level. The speech features include at least one of speech rate, pitch, volume, pause frequency, tone fluctuation, and emotional tendency. The context features include at least one of time period, location type, weather condition, schedule type, and social object type.

[0017] S120. Based on the multi-source features, identify the user's current state to obtain the user's physiological state, emotional state, and behavioral state.

[0018] Specifically, such as Figure 2As shown, step S120 includes steps S121-S122: S121, performing multimodal fusion analysis on the physiological features, behavioral features, voice features, and context features to obtain fusion analysis results; S122, inferring the fusion analysis results based on a preset state recognition model to obtain the user's physiological state, emotional state, and behavioral state. It should be noted that multimodal fusion analysis includes feature-level fusion and decision-level fusion. In feature-level fusion, feature transformation processing is performed on the physiological features, behavioral features, voice features, and context features respectively, and feature fusion weights are determined based on the availability status and quality assessment information of the physiological features, behavioral features, voice features, and context features. The transformed feature vectors are then weighted and aggregated according to the feature fusion weights to obtain a fused feature vector. Specifically, firstly, the physiological features... Behavioral characteristics Speech features and contextual features Through nonlinear mapping functions respectively Perform a feature transformation and project it onto a feature space of uniform dimension to obtain: in, The weight matrix is ​​a learnable matrix. For bias vectors, This is the transformed feature vector.

[0019] Secondly, availability status identifiers are defined for each modal feature. ,when This indicates that the feature is available. This indicates that the information is missing or unavailable; at the same time, quality assessment information is obtained. It includes quality indicators across multiple dimensions such as signal-to-noise ratio, completeness, timeliness, and confidence.

[0020] Furthermore, based on availability status and quality assessment information, the various quality indicators are first weighted and integrated to obtain a comprehensive quality value: in, For the first The allocation coefficients for each quality indicator are determined. Then, the fusion weights for each mode are calculated using a normalized exponential function. in, This is the temperature regulation coefficient, used to control the degree of influence of quality differences on weight allocation; satisfy When a certain mode is unavailable ( When the weight is set to zero, it achieves dynamic adaptive allocation.

[0021] Finally, the transformed feature vectors are weighted and aggregated according to the fusion weights to obtain the fused feature vector. : Understandably, feature-level fusion makes full use of the complementarity of multimodal features, while effectively suppressing the interference of low-quality or missing modalities on the fusion result through availability gating and multi-dimensional quality perception mechanisms, thereby improving the robustness and discriminative ability of feature representation.

[0022] In decision-level fusion, independent state inference is performed based on the physiological features, behavioral features, speech features, and contextual features, respectively, to obtain preliminary state judgment results and confidence levels for each of the physiological features, behavioral features, speech features, and contextual features. Then, the decision fusion weights are determined based on the availability state, quality assessment information, and confidence levels using the fusion feature vector. The preliminary state judgment results for each of the physiological features, behavioral features, speech features, and contextual features are then weighted and fused based on these decision fusion weights to obtain the fusion analysis result. Specifically, firstly, state recognition is performed on the physiological features, behavioral features, speech features, and contextual features (independent state recognition is performed on the four types of features using a pre-trained unimodal classification network), outputting two types of information: one is the preliminary state judgment result. First, it represents the user state category independently inferred from this modality; second, it represents the inference confidence level. This reflects the modality's level of confidence in the current judgment result. Each modality branch employs an independent classification network structure, avoiding mutual interference between modalities and ensuring the diversity and complementarity of inference results. Secondly, in the decision-level fusion stage, the availability status of each modality is comprehensively utilized. Quality assessment information Inference confidence level and the fused feature vector obtained by the aforementioned feature-level fusion. The global semantic information contained within is used to calculate the decision fusion weights for each modality. The fused feature vector carries multimodal collaborative representation information, which can further guide the decision layer in assessing the reliability of the judgment results for each branch. Decision Fusion Weights The calculation formula is as follows: in, The overall quality value is obtained by integrating quality assessment information. The modal correlation response function is extracted from the fused feature vector. , , This is a learnable adjustment coefficient used to balance the impact of confidence, quality, and global semantics on weight allocation.

[0023] Understandably, the preliminary state judgment results of each modality are weighted and fused based on the decision fusion weights to obtain the final fusion analysis result. This decision-level fusion strategy retains the independent judgment ability of each modality, and modulates the weights by jointly modulating the multi-dimensional information, so that the modal with high confidence, high quality and consistent with the global semantics can gain greater discourse power. While ensuring the robustness of the system, it significantly improves the accuracy and interpretability of the final state judgment.

[0024] It should also be noted that the fusion analysis results are input into the state recognition model, which outputs the probability distribution of various states. The state type with the highest probability is selected as the final recognition result. This state recognition model is a multi-class neural network (such as a multilayer perceptron). It takes the fused feature vector as input, learns the complex nonlinear mapping between features and states through hidden layers, and uses the Softmax function to transform the output into the probability distribution of various states. This model is trained on a large amount of labeled data and can accurately assess the probability of each state occurring, ultimately outputting discrete state labels based on the maximum probability criterion. Furthermore, physiological states include high-pressure states, relaxed states, fatigued states, recovery states, and active states; emotional states include happy states, depressed states, angry states, anxious states, calm states, and focused states; and behavioral states include working states, exercising states, resting states, social states, and sleeping states.

[0025] S130. Based on the physiological state, the emotional state, and the behavioral state, monitor the user's state changes, wherein the state changes include at least one of physiological state changes, emotional state changes, and behavioral state changes.

[0026] Specifically, the physiological state, emotional state, and behavioral state are compared with a reference state to obtain a state comparison result. The reference state includes at least one of the state at a historical moment and the user's historical baseline state. The state comparison result includes at least one of the state difference value, state deviation degree, and state type change. If the state difference value exceeds a preset difference threshold, and / or the state deviation degree exceeds a preset deviation threshold, and / or the state type changes, then a state change is determined. It should be noted that the state at a historical moment refers to the state identified at the previous moment or multiple previous moments; the user's historical baseline state refers to a personal state benchmark established based on the user's long-term historical data; the state difference value represents the numerical difference between the current state and the state at a historical moment; the state deviation degree represents the magnitude of the deviation between the current state and the user's historical baseline state; and the state type change indicates whether the current state type has switched from the reference state type. It should also be noted that the preset difference threshold is dynamically set based on the standard deviation of historical state fluctuations; the preset deviation threshold is set based on the normal fluctuation range of the user's individual baseline; and state type switching includes situations such as a physiological state switching from relaxed to high-pressure, an emotional state switching from calm to anxious, and a behavioral state switching from work to rest. Understandably, by using multi-dimensional comparisons and multiple judgment conditions, the accuracy and sensitivity of state change detection are ensured, and false positives and false negatives are avoided.

[0027] S140. In response to the key state change in the state change, generate a state slice corresponding to the key state change.

[0028] Specifically, such as Figure 3As shown, step S140 includes steps S141-S144: S141, assessing the degree of change of the state change to obtain a change assessment result; S142, based on the change assessment result, filtering out the key state changes from the state changes; S143, in response to detecting the key state change, acquiring the time information, state information, and context information corresponding to the key state change; S144, based on the time information, the state information, and the context information, generating the state slice corresponding to the key state change. It should be noted that the change assessment result includes at least one of change magnitude, change rate, and state duration; the key state change satisfies at least one of the following conditions: the change magnitude exceeds a preset magnitude threshold, the change rate exceeds a preset rate threshold, the state duration exceeds a preset duration threshold, or the state type changes; the state slice includes a timestamp, duration, physiological state, emotional state, behavioral state, event information, associated person, state score, AI summary text, and recommended action. It should also be noted that the magnitude of change is calculated by the numerical difference between the current state and the reference state; the rate of change is calculated by the magnitude of state change per unit time; and the duration of the state is calculated by accumulating the duration the state is maintained. The preset magnitude threshold, preset rate threshold, and preset duration threshold are dynamically adjusted based on the user's historical state data to accommodate individual differences among users. Understandably, when generating a state slice, the time information includes the start timestamp and duration of the state change; the state information includes the specific type and intensity score of the physiological, emotional, and behavioral states; and the contextual information includes location, schedule, weather, and associated people. Furthermore, associated people are obtained through methods such as voice content recognition of names, calendar meeting participants, contact mapping, or identification of frequently interacting objects. The state score is calculated based on the significance and duration of the state change. The AI ​​summary text is generated by analyzing the time, state, and contextual information using a large language model to summarize the core content of the state slice. Recommended actions are generated based on the identified state type and the user's historical preferences to provide targeted state adjustment suggestions to the user.

[0029] Figure 4 This is a flowchart illustrating a state slice generation method according to another embodiment of the present invention, as shown below. Figure 4 As shown, in this embodiment, the method includes steps S110-S160. That is, in this embodiment, after step S140 in the above embodiment, the method further includes steps S150-S160.

[0030] S150. Organize the multiple state slices according to a time sequence to form a state slice set; S160. Display the set of state slices according to a preset time dimension, wherein the time dimension includes at least one of daily view, weekly view, monthly view, person view, and state view.

[0031] Specifically, the daily view displays representative status slices and an overall status analysis for the day; the weekly view displays status change trends and sentiment distribution over a week; the monthly view displays statistical results and patterns of long-term status data; the people view displays a set of status slices and interaction records associated with a specific person; and the status view displays the user's status distribution across different time periods. It should be noted that the status slice set is arranged in ascending order of timestamps and supports dynamic updates, automatically adding new status slices as they are generated. Switching between views is supported, allowing users to switch between different time dimensions by clicking or swiping. Furthermore, the representative status slices in the daily view are sorted by status rating, with a preset number of slices featuring the highest ratings displayed. Status change trends in the weekly view are displayed using line charts or area charts, and sentiment distribution is shown using pie charts or bar charts. Statistical results in the monthly view include the frequency of occurrence, duration distribution, and periodic pattern analysis of various statuses. Specific people in the people view are determined through user selection or recommendation, and interaction records include the number of status slices associated with that person, sentiment change trends, and interaction time distribution. The status distribution in the status view is displayed through heatmaps or radar charts, and can be filtered by time period, scene, or person. Each view allows clicking on a status slice to enter a details page to view complete information.

[0032] In one embodiment, such as this embodiment, after step S160, the method further includes: responding to the user's selection operation on the target time dimension, entering the corresponding review view; responding to the user's trigger operation on any target state slice, entering the details review page of the target state slice; responding to the user's trigger operation on the associated person in the target state slice, entering the person review page. It should be noted that the review view displays the state change trend, state distribution statistics, and a representative state slice list under the target time dimension. The details review page displays complete information about the target state slice, including timestamp, duration, physiological state, emotional state, behavioral state, event information, associated person, state score, AI summary text, and recommended actions. The person review page displays a collection of historical state slices related to the associated person, emotional change trends, and interaction records. Interaction records include the frequency and duration of interactions with the person, and the corresponding state distribution. It should also be noted that each page supports returning to the previous view and jumping between different pages, forming a complete review interaction path.

[0033] The state slice generation method in this invention has the following beneficial effects: by converting continuous multi-source data into structured state slices, users can clearly identify the time and type of their own state changes, solving the technical problem that existing wearable devices cannot identify changes in the user's own state; specifically, by multimodal fusion analysis of health data, behavioral data, voice data, and contextual information, a comprehensive identification of the user's physiological state, emotional state, and behavioral state is achieved, improving the accuracy and comprehensiveness of state recognition; through state change monitoring and key state change filtering, moments of state change of significant value in the user's life are automatically captured without the need for manual recording by the user, improving the automation and convenience of state recording; It supports multi-dimensional display and review, including daily, weekly, monthly, person, and status views, enabling users to understand the patterns of their own status changes from different time dimensions and related object perspectives. The generated structured status slices contain rich information such as time, status, events, people, AI summaries, and recommended actions, which can provide AI Agents with structured input of user status, supporting personalized suggestion generation and proactive intervention services.

[0034] It should be understood that the sequence number of each step in the above embodiments does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

[0035] The software tools or components not belonging to our company that appear in the embodiments of this application are merely examples and do not represent actual use.

[0036] In one embodiment, a state slice generation device 200 is provided, which corresponds one-to-one with the state slice generation method in the above embodiments. As shown in Figure 5, the state slice generation device 200 includes an acquisition and extraction unit 201, a state identification unit 202, a monitoring unit 203, and a generation unit 204. Detailed descriptions of each functional module are as follows: The acquisition and extraction unit 201 is used to continuously acquire multi-source data from users and extract features from the multi-source data to obtain multi-source features; The state recognition unit 202 is used to identify the user's current state based on the multi-source features, and obtain the user's physiological state, emotional state and behavioral state; The monitoring unit 203 is used to monitor changes in the user's state based on the physiological state, the emotional state, and the behavioral state, wherein the changes in state include at least one of changes in physiological state, changes in emotional state, and changes in behavioral state. The generation unit 204 is used to generate a state slice corresponding to the key state change in response to the key state change in the state change.

[0037] In one embodiment, the acquisition and extraction unit 201 is specifically used for: The physiological features are extracted from the health data using a physiological feature extraction method; The behavioral features are extracted from the behavioral data using a behavioral feature extraction method. The speech features are extracted from the speech data using a speech feature extraction method; The context features are extracted from the context information using a context feature extraction method.

[0038] In one embodiment, the state recognition unit 202 is specifically used for: Multimodal fusion analysis is performed on the physiological features, behavioral features, speech features, and contextual features to obtain the fusion analysis results; Based on a preset state recognition model, the fusion analysis results are inferred to obtain the user's physiological state, emotional state, and behavioral state.

[0039] In one embodiment, the monitoring unit 203 is specifically used for: The physiological state, emotional state, and behavioral state are compared with a reference state to obtain a state comparison result. The reference state includes at least one of the state at a historical moment and the user's historical baseline state. The state comparison result includes at least one of the state difference value, the degree of state deviation, and the change in state type. If the state difference value exceeds a preset difference threshold, and / or the state deviation exceeds a preset deviation threshold, and / or the state type changes, then the state change is determined.

[0040] In one embodiment, the generation unit 204 is specifically used for: The degree of change of the state is evaluated to obtain the change evaluation result; Based on the change assessment results, the key state changes are selected from the state changes; In response to detecting the key state change, the time information, state information and context information corresponding to the key state change are obtained; Based on the time information, the state information, and the context information, a state slice corresponding to the key state change is generated.

[0041] In one embodiment, the state slice generation apparatus 200 further includes: An organization forming unit is used to organize multiple state slices according to a time sequence to form a state slice set; The display unit is used to display the set of state slices according to a preset time dimension, wherein the time dimension includes at least one of daily view, weekly view, monthly view, person view and state view; The first response entry unit is used to respond to the user's selection of the target time dimension and enter the corresponding playback view; The second response entry unit is used to respond to the user's trigger operation on any target state slice and enter the details review page of the target state slice; The third response entry unit is used to respond to the user's trigger operation on the associated person in the target state slice and enter the person review page.

[0042] In one embodiment, the display unit is specifically used for: The daily view displays representative state slices and an overall state analysis for the day. The weekly view displays the trend of state changes and the distribution of emotions over a period of one week; The monthly view displays the statistical results and patterns of change of long-term state data; The character view displays the set of state slices and interaction records associated with a specific character; The status view displays the distribution of the user's status across different time periods.

[0043] The aforementioned state slice generation device can be implemented as a computer program, which can, for example, Figure 6 It runs on the wearable device shown.

[0044] Please see Figure 6 , Figure 6 This is a schematic block diagram of a wearable device provided in an embodiment of the present invention. The wearable device 300 is a device capable of generating state slices.

[0045] See Figure 6 The wearable device 300 includes a processor 302, a memory, and a network interface 305 connected via a system bus 301. The memory may include a non-volatile storage medium 303 and internal memory 304.

[0046] The non-volatile storage medium 303 can store an operating system 3031 and a computer program 3032. When the computer program 3032 is executed, it causes the processor 302 to execute a state slice generation method.

[0047] The processor 302 provides computing and control capabilities to support the operation of the entire wearable device 300.

[0048] The internal memory 304 provides an environment for the execution of the computer program 3032 in the non-volatile storage medium 303. When the computer program 3032 is executed by the processor 302, the processor 302 can execute a state slice generation method.

[0049] This network interface 305 is used for network communication with other devices. Those skilled in the art will understand that... Figure 6 The structure shown is merely a block diagram of a portion of the structure related to the present invention and does not constitute a limitation on the wearable device 300 to which the present invention is applied. The specific wearable device 300 may include more or fewer components than shown in the figure, or combine certain components, or have different component arrangements.

[0050] The processor 302 is used to run a computer program 3032 stored in a memory to implement any embodiment of the state slice generation method described above.

[0051] It should be understood that, in this embodiment of the invention, the processor 302 may be a Central Processing Unit (CPU), or it may be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or any conventional processor.

[0052] It will be understood by those skilled in the art that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a storage medium, which is a computer-readable storage medium. The computer program is executed by a processor in the computer system to implement the process steps of the embodiments of the above methods.

[0053] Therefore, the present invention also provides a storage medium. This storage medium can be a computer-readable storage medium. The storage medium stores a computer program. When executed by a processor, the computer program causes the processor to perform any embodiment of the state slice generation method described above.

[0054] The storage medium can be any computer-readable storage medium capable of storing program code, such as a USB flash drive, portable hard drive, read-only memory (ROM), magnetic disk, or optical disk.

[0055] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of both. To clearly illustrate the interchangeability of hardware and software, the components and steps of the various examples have been generally described in terms of functionality in the foregoing description. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this invention.

[0056] In the several embodiments provided by this invention, it should be understood that the disclosed apparatus and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative. For example, the division of each unit is merely a logical functional division, and there may be other division methods in actual implementation. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed.

[0057] The steps in the method of this invention can be adjusted, merged, or reduced in order according to actual needs. The units in the device of this invention can be merged, divided, or reduced according to actual needs. Furthermore, the functional units in the various embodiments of this invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0058] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a wearable device to execute all or part of the steps of the methods described in the various embodiments of the present invention.

[0059] In the above embodiments, the descriptions of each embodiment have different focuses. For parts that are not described in detail in a certain embodiment, please refer to the relevant descriptions in other embodiments.

[0060] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Since these modifications and variations fall within the scope of the claims and their equivalents, this invention also intends to include these modifications and variations.

[0061] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any person skilled in the art can easily conceive of various equivalent modifications or substitutions within the technical scope disclosed in the present invention, and these modifications or substitutions should all be covered within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A method for generating state slices, characterized in that, include: Continuously collect multi-source data from users, and extract features from the multi-source data to obtain multi-source features; Based on the multi-source features, the user's current state is identified to obtain the user's physiological state, emotional state, and behavioral state. Based on the physiological state, the emotional state, and the behavioral state, monitor the user's state changes, wherein the state changes include at least one of physiological state changes, emotional state changes, and behavioral state changes. In response to key state changes in the state changes, a state slice corresponding to the key state change is generated.

2. The state slice generation method as described in claim 1, characterized in that, The multi-source features include physiological features, behavioral features, speech features, and contextual features; The step of identifying the user's current state based on the multi-source features to obtain the user's physiological state, emotional state, and behavioral state includes: Multimodal fusion analysis is performed on the physiological features, behavioral features, speech features, and contextual features to obtain the fusion analysis results; Based on a preset state recognition model, the fusion analysis results are inferred to obtain the user's physiological state, emotional state, and behavioral state.

3. The state slice generation method as described in claim 2, characterized in that, The multimodal fusion analysis includes feature-level fusion and decision-level fusion. The step of performing multimodal fusion analysis on the physiological features, behavioral features, speech features, and contextual features to obtain the fusion analysis result includes: The physiological features, behavioral features, speech features, and context features are respectively subjected to feature transformation processing. Based on the availability status and quality assessment information of the physiological features, behavioral features, speech features, and context features, feature fusion weights are determined. The transformed feature vectors are weighted and aggregated according to the feature fusion weights to perform feature-level fusion and obtain a fused feature vector. Independent state inferences are performed based on the physiological features, behavioral features, speech features, and contextual features respectively to obtain preliminary state judgment results and confidence levels for each of the physiological features, behavioral features, speech features, and contextual features. Then, a decision fusion weight is determined based on the availability state, quality assessment information, and confidence levels using the fusion feature vector. The preliminary state judgment results for each of the physiological features, behavioral features, speech features, and contextual features are then weighted and fused based on the decision fusion weight to perform the decision-level fusion and obtain the fusion analysis result.

4. The state slice generation method as described in claim 1, characterized in that, The step of monitoring changes in a user's state based on the physiological state, emotional state, and behavioral state includes: The physiological state, emotional state, and behavioral state are compared with a reference state to obtain a state comparison result. The reference state includes at least one of the state at a historical moment and the user's historical baseline state. The state comparison result includes at least one of the state difference value, the degree of state deviation, and the change in state type. If the state difference value exceeds a preset difference threshold, and / or the state deviation exceeds a preset deviation threshold, and / or the state type changes, then the state change is determined.

5. The state slice generation method as described in claim 1, characterized in that, The step of generating a state slice corresponding to a key state change in response to the state change includes: The degree of change of the state is evaluated to obtain the change evaluation result; Based on the change assessment results, the key state changes are selected from the state changes; In response to detecting the key state change, the time information, state information and context information corresponding to the key state change are obtained; Based on the time information, the state information, and the context information, a state slice corresponding to the key state change is generated.

6. The state slice generation method as described in claim 5, characterized in that, The change assessment result includes at least one of the change magnitude, change rate, and state duration; the key state change satisfies at least one of the following conditions: the change magnitude exceeds a preset magnitude threshold, the change rate exceeds a preset rate threshold, the state duration exceeds a preset duration threshold, or the state type is switched; the state slice includes a timestamp, duration, physiological state, emotional state, behavioral state, event information, associated person, state score, AI summary text, and recommended action.

7. The state slice generation method according to any one of claims 1-5, characterized in that, Following the step of generating state slices corresponding to the key state changes, the method further includes: The multiple state slices are organized according to a time series to form a state slice set; The set of state slices is displayed according to a preset time dimension, wherein the time dimension includes at least one of daily view, weekly view, monthly view, person view, and state view.

8. The state slice generation method as described in claim 7, characterized in that, The step of displaying the set of state slices according to a preset time dimension includes: The daily view displays representative state slices and an overall state analysis for the day. The weekly view displays the trend of state changes and the distribution of emotions over a period of one week; The monthly view displays the statistical results and patterns of change of long-term state data; The character view displays the set of state slices and interaction records associated with a specific character; The status view displays the distribution of the user's status across different time periods.

9. The state slice generation method as described in claim 7, characterized in that, After the step of displaying the set of state slices according to a preset time dimension, the method further includes: In response to the user's selection of the target time dimension, the corresponding playback view is displayed. In response to a user's triggering action on any target state slice, the user is taken to the details review page of the target state slice. In response to a user's triggering action on a person associated with the target state slice, the user is taken to the person replay page.

10. A state slice generation apparatus, characterized in that, include: The data acquisition and extraction unit is used to continuously acquire multi-source data from users and extract features from the multi-source data to obtain multi-source features. The state recognition unit is used to identify the user's current state based on the multi-source features, and to obtain the user's physiological state, emotional state and behavioral state. A monitoring unit is used to monitor changes in a user's state based on the physiological state, the emotional state, and the behavioral state, wherein the changes in state include at least one of changes in physiological state, changes in emotional state, and changes in behavioral state. The generation unit is used to generate a state slice corresponding to a key state change in response to the state change.

Images