Health detection method and system based on intelligent toilet
Data is collected through the image sensing element and temperature sensing element of the smart toilet, and the correlation matching is performed with user identity tags, which solves the problem of single function of traditional toilets, provides personalized health monitoring, and improves the accuracy of user experience and health monitoring. .
Patent Information
- Application Number
- CN202410105284.4
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-01-25
- Publication Date
- 2024-11-19
- Estimated Expiration
- 2044-01-25
AI Technical Summary
Traditional toilets have a single function and cannot adapt to the multifunctional needs of smart homes, resulting in poor user experience and lack of health monitoring functions.
The image sensing element and temperature sensing element in the smart toilet collect the color and temperature characteristics of the stool, combined with the user identity tag, perform multi-parameter correlation matching and sorting, and provide personalized health monitoring suggestions.
It realizes non-invasive and multi-parameter comprehensive analysis health monitoring, improves user experience, ensures the convenience and accuracy of health monitoring, and adapts to the multi-functional needs of smart homes.
Smart Images

Figure CN118037647B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of data processing, and in particular to a health detection method and system based on a smart toilet. Background Art
[0002] Smart home requires that all home appliances can be intelligent and provide multifunctional and user-friendly experience. As an indispensable device in the bathroom, smart toilets also need to be intelligent to meet multifunctional needs. Traditional toilets usually only provide basic defecation functions, which are simple and insufficient to meet the diverse needs of users in the era of smart home. Due to the lack of intelligent features, the user experience of traditional toilets is usually poor; in addition, traditional toilets cannot provide health monitoring functions, and users need more health data for self-monitoring and early disease prevention.
[0003] Therefore, a new solution is needed to improve the limitations of traditional toilets and enhance user experience through features such as intelligent health monitoring, while providing users with more functions and convenience, thereby better adapting to the multi-functional needs of smart homes and creating a higher level of convenience and comfort for users. Summary of the invention
[0004] This application provides a health detection method based on a smart toilet, aiming to solve the technical problem that traditional toilets have a single function and cannot adapt to the multi-functional needs of smart homes, resulting in a poor user experience.
[0005] In view of the above problems, the present application provides a health detection method and system based on a smart toilet.
[0006] The first aspect disclosed in the present application provides a health detection method based on a smart toilet, which is applied to a health detection system based on a smart toilet. The system is communicatively connected to the smart toilet, and the smart toilet includes an image sensor element and a temperature sensor element. The method includes: when a user activates the smart toilet and a preset state is met, the user terminal login information is exchanged to obtain a user identity tag; stool color characteristics and a dilution coefficient are collected through the image sensor element; stool temperature characteristics are collected through the temperature sensor element; user age and user gender are obtained based on the user identity tag; based on the user age and the user gender, the stool color characteristics, the dilution coefficient and the stool temperature characteristics are frequently associated with each other to obtain a high-frequency state set; the high-frequency state set is sorted according to the trigger frequency coefficient to obtain a high-frequency state sorting result, and sent to the user terminal.
[0007] Another aspect disclosed in the present application provides a health detection system based on a smart toilet, wherein the system is in communication connection with the smart toilet, wherein the smart toilet includes an image sensor element and a temperature sensor element, and the system is used in the above-mentioned health detection method based on the smart toilet, wherein the system includes: an identity tag acquisition module, wherein the identity tag acquisition module is used to obtain a user identity tag by exchanging user terminal login information when the user activates the smart toilet and meets a preset state; a stool color feature acquisition module, wherein the stool color feature acquisition module is used to acquire stool color features and a dilution coefficient through an image sensor element; and a stool temperature feature acquisition module, wherein the stool temperature feature acquisition module is used to acquire stool color features and a dilution coefficient through an image sensor element; and The stool temperature feature acquisition module is used to collect stool temperature features through a temperature sensing element; the user information acquisition module is used to obtain the user's age and gender based on the user identity tag; the high-frequency state acquisition module is used to perform frequent association state matching on the stool color feature, the thinness coefficient and the stool temperature feature based on the user's age and the user's gender to obtain a high-frequency state set; the sorting result acquisition module is used to sort the high-frequency state set according to the trigger frequency coefficient, obtain a high-frequency state sorting result, and send it to the user end.
[0008] One or more technical solutions provided in this application have at least the following technical effects or advantages:
[0009] Through the smart toilet, the user interacts with the login information when activating and meeting the preset state, thereby obtaining the user's identity tag and realizing accurate user identification; using the image sensor element, the color characteristics and dilution coefficient of the stool are successfully collected, which helps to understand the state and health of the stool more comprehensively; using the temperature sensing element, the temperature characteristics of the stool are successfully collected, providing additional physiological information to help health detection; according to the user identity tag, the user's age and gender information are successfully obtained, providing basic data for subsequent personalized analysis; by associating and matching the user's age and gender with multiple health parameters, a high-frequency state set is successfully obtained, which helps to discover health status patterns related to specific age and gender; the high-frequency state set is sorted according to the trigger frequency coefficient and sent to the user end, so that the user can get targeted suggestions according to his health status. In summary, the health detection method based on the smart toilet solves the technical problems that the traditional toilet has a single function and cannot adapt to the multi-functional needs of smart homes, resulting in poor user experience, provides non-invasive, multi-parameter comprehensive analysis and personalized health monitoring, realizes the health detection function of the toilet, and ensures the convenience and accuracy of health monitoring, which improves the user experience.
[0010] The above description is only an overview of the technical solution of the present application. In order to more clearly understand the technical means of the present application, it can be implemented in accordance with the contents of the specification. In order to make the above and other purposes, features and advantages of the present application more obvious and easy to understand, the specific implementation methods of the present application are listed below. BRIEF DESCRIPTION OF THE DRAWINGS
[0011] Figure 1 A flow chart of a health detection method based on a smart toilet is provided for the embodiment of the present application;
[0012] Figure 2 A schematic diagram of the structure of a health detection system based on a smart toilet is provided for the embodiment of the present application.
[0013] Explanation of the reference numerals: identity tag acquisition module 10 , stool color feature acquisition module 20 , stool temperature feature acquisition module 30 , user information acquisition module 40 , high-frequency state acquisition module 50 , sorting result acquisition module 60 . DETAILED DESCRIPTION
[0014] The embodiment of the present application solves the technical problem that traditional toilets have a single function and cannot adapt to the multi-functional requirements of smart homes, resulting in a poor user experience by providing a health detection method based on a smart toilet.
[0015] After introducing the basic principles of the present application, various non-limiting implementation methods of the present application will be specifically described below in conjunction with the drawings in the specification.
[0016] Embodiment 1
[0017] like Figure 1 As shown, the embodiment of the present application provides a health detection method based on a smart toilet, which is applied to a health detection system based on a smart toilet. The system is connected to the smart toilet in communication. The smart toilet includes an image sensor element and a temperature sensor element. The method includes:
[0018] When the user activates the smart toilet and meets the preset status, the user terminal login information is exchanged to obtain the user identity tag;
[0019] The health detection method based on a smart toilet provided in the embodiment of the present application is applied to a health detection system based on a smart toilet, the system is communicatively connected with the smart toilet, the smart toilet includes an image sensor element and a temperature sensor element, wherein the smart toilet of the present application is a device that can only be used after obtaining user permission and signing a user privacy confidentiality agreement.
[0020] The smart toilet includes two main components, namely an image sensing component and a temperature sensing component. The image sensing component is used to capture and record images or video data of the user's feces. It includes a camera or an image sensor. The function of this component is to collect visual information of the feces, such as the color, shape and other characteristics of the feces; the temperature sensing component is used to measure the temperature of the user's feces. It includes a temperature sensor or a thermistor, which can sense the temperature changes of the feces. The temperature information of the feces reflects the user's gastrointestinal health status, such as the normal function of the digestive system or whether there are fever symptoms.
[0021] First, the user needs to activate the smart toilet, for example, by pressing a button or starting the toilet system through infrared sensing. After the user activates the toilet, check whether the user meets the preset state. The preset state means that the user is squatting on the toilet, which can be detected by the sensor on the toilet. When it is confirmed that the user meets the preset state, based on the user's indoor positioning data, it can be determined which user is defecating, obtain the user's login information, and obtain the user's identity tag based on the login information. This identity tag includes the user's unique identifier so that the user's data and health information can be associated with it.
[0022] The stool color characteristics and dilution coefficient are collected through the image sensor element;
[0023] An image sensor element is used to capture images or video data of the user's stool. These images or video data contain the color information of the stool and are represented by the numerical values of the RGB (red, green, and blue) color channels. The color of the stool can reflect certain health information, such as the state of the digestive system or the impact of food intake. By analyzing the numerical values of the RGB channels, the color characteristics of the stool can be determined.
[0024] The thinness coefficient indicates the dryness or thinness of the stool, that is, the concentration or consistency of the stool. The physical properties of the stool are measured using a dryness and thinness sensor element. By measuring the density, water content or other relevant parameters of the stool, it is determined whether the stool is dry or thin. These characteristics help to assess the user's gastrointestinal health status.
[0025] Furthermore, the stool color characteristics and dilution coefficient are collected through the image sensor element, including:
[0026] Collecting a defecation movement video through the image sensor element, wherein the defecation movement video includes a movement endpoint image;
[0027] Activate the dynamic feature analysis component to process the stool motion video and generate the rarefaction coefficient;
[0028] The static feature analysis component is activated to process the motion endpoint image to generate the stool color feature.
[0029] Through the image sensor element, frames of the stool movement video are continuously collected from the time the user sits on the toilet until the end of defecation, and the movement endpoint image is captured. The stool movement video refers to the video recording the user's defecation process, including all stages of defecation. Among them, the movement endpoint image refers to the image of the feces falling into the pit under the toilet after the user finishes defecation. This image contains information about the stool, including color, shape, etc., which is used for health analysis.
[0030] The dynamic feature parsing component is an algorithmic component designed to analyze and process the content of the stool motion video, including video processing technology, to extract key information about the defecation process. After the stool motion video is acquired, the dynamic feature parsing component is activated to start processing the stool motion video. This processing process includes video frame analysis, motion tracking and other related technologies to determine the key features of the defecation process. During the processing, the rarefaction coefficient is generated based on the appearance and density of the stool in the video. The rarefaction coefficient is used to indicate the dryness or rarefaction of the stool.
[0031] The static feature parsing component is also an algorithm component that is designed to analyze and process the content of static images. It is used to analyze the end-point image, that is, the image after defecation. After acquiring the end-point image, the static feature parsing component is activated to start processing the end-point image. This processing process includes image analysis, color recognition, feature extraction and other technologies to determine the features of the stool in the end-point image. During the processing, the static feature parsing component extracts information to generate the stool color feature. The stool color feature represents the color characteristics of the stool, including the numerical value of the RGB color channel or other color descriptions. This feature is used to analyze the color of the stool and information related to gastrointestinal health.
[0032] Furthermore, activating the dynamic feature analysis component to process the defecation motion video and generate the rarefaction coefficient includes:
[0033] The defecation motion video is processed through the video segmentation channel of the dynamic feature analysis component to generate a first cycle motion video, a second cycle motion video, and an Nth cycle motion video, wherein N is an integer, and N≥1;
[0034] Through the dry-thin feature analysis channel of the dynamic feature analysis component, the first period motion video, the second period motion video, and the Nth period motion video are traversed for processing to generate a first thinness coefficient, a second thinness coefficient, and the Nth thinness coefficient;
[0035] Performing cluster analysis on the first rarefaction coefficient, the second rarefaction coefficient, and up to the Nth rarefaction coefficient to obtain a first characteristic rarefaction coefficient, a second characteristic rarefaction coefficient, and up to the Mth characteristic rarefaction coefficient, where M is an integer and N≥M≥1;
[0036] The first characteristic dilution coefficient, the second characteristic dilution coefficient, and up to the Mth characteristic dilution coefficient are subjected to maximum value selection to obtain the dilution coefficient.
[0037] The video segmentation channel is a functional module within the dynamic feature analysis component that is used to segment the defecation motion video into different periods. The channel uses video analysis techniques, such as motion detection or timestamp segmentation, to determine the boundaries of different defecation periods. The defecation motion video is a video that continuously records the user's defecation process, which includes multiple defecation periods, that is, the intermittent occurrence of defecation. The video segmentation channel divides the video into different defecation periods for processing.
[0038] During the processing, the video segmentation channel divides the defecation motion video into multiple defecation cycles, generating the first cycle motion video, the second cycle motion video, and so on to the Nth cycle motion video, each cycle corresponding to an intermittent occurrence of defecation. Where N is an integer representing the number of cycles into which the defecation motion video is segmented, and the value of N depends on the specific defecation situation to divide the video into different defecation cycles.
[0039] The dryness and thinness feature analysis channel is another functional module within the dynamic feature analysis component, which is specifically used to analyze the dryness and thinness features in the video. These features involve the appearance, density, or other dryness and thinness properties of the stool. Using the dryness and thinness feature analysis channel, the video of each defecation cycle is traversed. For each video, the dryness and thinness feature analysis channel extracts the dryness and thinness characteristics of the stool through image processing, density calculation, texture analysis and other technologies. Based on the results of the dryness and thinness feature analysis, the first thinness coefficient, the second thinness coefficient, and so on to the Nth thinness coefficient are generated. These thinness coefficients reflect the dryness and thinness of the stool in each defecation cycle.
[0040] Cluster analysis is used to divide data into similar groups, so that the data points in each group are similar to each other, while the data points between different groups are quite different. In this step, cluster analysis is used to group the rarefaction coefficients, and classify similar rarefaction coefficients into the same feature group. Similar rarefaction coefficients are clustered together through cluster analysis, and a representative rarefaction coefficient is assigned to each feature group, where the first feature rarefaction coefficient represents the typical dry-thin characteristics in the first feature group, the second feature rarefaction coefficient represents the typical dry-thin characteristics of the second feature group, and so on, until the Mth feature rarefaction coefficient.
[0041] Where M is an integer representing the final number of feature groups. The value of M can be selected according to the clustering results and analysis objectives. A smaller M value means dividing the rarefaction coefficient into fewer feature groups, while a larger M value means more feature groups.
[0042] For the first characteristic dilution coefficient, the second characteristic dilution coefficient, up to the Mth characteristic dilution coefficient, they are sorted from high to low according to the numerical size, and the maximum value is selected according to the sorting result. After the maximum value is selected, the characteristic dilution coefficient corresponding to the obtained value is used as the dilution coefficient. This dilution coefficient represents the defecation cycle with the most significant dry and dilute characteristics. Generally speaking, the thinner the stool, the worse the stool condition, so the thinnest one is selected as the dilution coefficient to ensure the greatest possible detection of abnormal user conditions.
[0043] Furthermore, activating the static feature analysis component to process the motion endpoint image to generate the stool color feature includes:
[0044] activating the static feature analysis component, processing the motion endpoint image, and obtaining a stool color feature set;
[0045] Performing a central tendency analysis on the stool color feature set to generate the stool color feature;
[0046] Among them, the static feature analysis component is obtained by supervised configuration of the convolutional neural network through the stool image record dataset and the stool RGB feature identification dataset.
[0047] The static feature analysis component is activated to first pre-process the motion endpoint image, including removing noise in the image to improve image clarity; adjusting the brightness and contrast of the image to ensure that color features are easier to distinguish; and also including operations such as scaling, cropping or rotating the image to meet processing needs.
[0048] The static feature parsing component uses a convolutional neural network (CNN) for feature extraction. CNN is a deep learning model that is widely used in image analysis and recognition tasks. In this step, CNN has been pre-configured for supervised learning and trained using a stool image record dataset and a stool RGB feature identification dataset. The supervised learning configuration of CNN enables it to learn and extract stool color features based on existing data, which means that CNN has learned to recognize RGB color features in stool images.
[0049] When an image is processed by CNN, the model extracts color features from the image. These color features are expressed in the form of RGB values, representing the brightness of the three color channels of red, green, and blue. After the color features of each motion endpoint image are extracted, these features are organized into a color feature set, which contains the color features in all motion endpoint images, and each feature is expressed in the form of RGB values.
[0050] Central tendency analysis is a statistical data analysis technique used to measure the trend or central position in a data set. In this step, central tendency analysis is used to analyze the stool color feature set to understand the central tendency or typical value of the stool color in the entire data set. Specifically, statistical methods such as mean, median, mode, etc. are used to analyze the central tendency of the stool color feature set. For example, by calculating the mean of the RGB color channel in the stool color feature set to obtain a value representing the central tendency of the color of the entire data set, the stool color feature is generated according to the result of the central tendency analysis, which represents the central tendency of the stool color.
[0051] Furthermore, the defecation motion video is processed through the video segmentation channel of the dynamic feature analysis component to generate a first cycle motion video, a second cycle motion video, and finally an Nth cycle motion video, including:
[0052] Decomposing the defecation motion video to obtain a defecation motion video frame sequence, wherein the defecation motion video frame sequence has a collection position label sequence and a collection time label sequence;
[0053] Based on the acquisition position tag sequence, extracting a first defecation motion video frame set belonging to the motion starting position from the defecation motion video frame sequence;
[0054] Traversing the first defecation motion video frame set, activating the intermittent feature identification node of the video segmentation channel for processing, and obtaining a second defecation motion video frame set, wherein the second defecation motion video frame set is a video frame with an intermittent feature identification;
[0055] Obtaining a motion cycle sequence according to the acquisition time tag sequence and the second defecation motion video frame set;
[0056] According to the motion cycle sequence, the period segmentation node of the video segmentation channel is activated to process the defecation motion video to generate the first period motion video, the second period motion video until the Nth period motion video.
[0057] Decompose the defecation motion video, that is, split the video into a series of separate image frames. This can be done through video processing software to divide the video into individual image frames. When collecting defecation motion videos, use a collection position tag to record the position information of each image frame, which can be the position of the toilet. The collection position tag will track the position of each image frame, and generate a corresponding collection position tag sequence according to the front and back order of the defecation motion video frames; similarly, during the collection process, record the time information of each image frame, and use the collection time tag to record the collection time of each image frame, usually in milliseconds. According to the front and back order of the defecation motion video frames, a corresponding collection time tag sequence is generated.
[0058] When the decomposition and label collection are completed, this information is combined to form a sequence of video frames of stool movement. Each image frame is associated with its corresponding position label and time label. These label sequences are used to determine the position and time information of the stool.
[0059] The acquired position label sequence records the position information of each image frame. The movement starting position is the starting position of the stool movement. Determine which position labels correspond to the movement starting position. This can be achieved through predefined rules, such as determining the movement starting position based on the design of the toilet and the information of the position label.
[0060] When the starting position of the movement is determined, the defecation movement video frame sequence is traversed, and the image frames belonging to the starting position of the movement are extracted. These image frames constitute the first defecation movement video frame set. The first defecation movement video frame set contains image frames corresponding to the starting position of the movement, that is, the starting part of the defecation process. These image frames are used for subsequent analysis and processing.
[0061] Traverse each image frame in the first defecation motion video frame set, and on each image frame, activate the intermittent feature identification node of the video segmentation channel. This node is a model or algorithm whose goal is to identify intermittent features in the defecation process, such as the presence or absence of feces.
[0062] On each image frame, the image content is analyzed by the intermittent feature identification node to determine whether there is stool during defecation. This can be achieved through changes in image content, such as changes in color or pixel intensity. If the node detects that stool is not present, then the image frame is identified as having intermittent features, and the corresponding moments of such images are intermittent moments of defecation.
[0063] All image frames identified as having intermittent features are combined together to form a second defecation motion video frame set, which contains the intermittent moments in the defecation process, that is, the starting and ending positions of the defecation, so as to obtain the key time points in the defecation process.
[0064] The motion cycle includes the time period from the start of defecation to the end of defecation. The start and end of each motion cycle are determined by comparing the acquisition time tag sequence and the frames in the second defecation motion video frame set, wherein the start of the motion cycle can be matched with the first frame with intermittent characteristics, and the end of the motion cycle can be matched with the last frame with intermittent characteristics. According to the start and end of each motion cycle, a motion cycle sequence is constructed, which includes different motion cycles in the defecation process, and each motion cycle can be represented by the start time and end time of defecation.
[0065] Activate the period segmentation node of the video segmentation channel. The goal of this node is to segment the defecation motion video into different motion cycles. According to the motion cycle sequence, the defecation motion video is segmented into the first cycle motion video, the second cycle motion video, and so on until the Nth cycle motion video. The start and end time of each motion cycle is used to determine the time period of the segmented video. Each segmented video segment is processed independently to generate the corresponding period motion video. These period motion videos represent different motion cycles in the defecation process for subsequent analysis.
[0066] Further, through the dry-sparse feature parsing channel of the dynamic feature parsing component, the first period motion video, the second period motion video, and the Nth period motion video are traversed for processing to generate a first sparseness coefficient, a second sparseness coefficient, and the Nth sparseness coefficient, including:
[0067] The dry-sparse feature parsing channel includes a texture feature extraction node and a pixel feature extraction node;
[0068] According to the texture feature extraction node and the pixel feature extraction node, traverse the first periodic motion video frame sequence of the first periodic motion video to extract texture features, and generate a texture feature sequence and a pixel feature sequence;
[0069] Configure reference texture features and reference pixel features;
[0070] Traversing the texture feature sequence and the reference texture feature to perform distance evaluation and generate a first distance evaluation value sequence;
[0071] Traversing the pixel feature sequence and the reference pixel feature to perform distance evaluation and generate a second distance evaluation value sequence;
[0072] Performing mean calculation on the first distance evaluation value sequence to generate a first characteristic distance, and performing mean calculation on the second distance evaluation value sequence to generate a second characteristic distance;
[0073] The first characteristic distance and the second characteristic distance are respectively normalized and then added to generate the rarefaction coefficient.
[0074] The texture feature extraction node is a component used to capture the image texture in the stool motion video. Texture is the local structure in the image, which can be used to describe the details of the image and the changes in a specific area. The task of this node is to analyze each image frame to extract relevant information about the image texture. Among them, texture feature extraction can use various image processing techniques, such as gray-level co-occurrence matrix, local binary pattern (LBP), histogram of oriented gradients (HOG), etc. to capture texture features.
[0075] The pixel feature extraction node is a component used to analyze the pixel-level information of each image frame in the stool motion video, including the analysis of the image's color information, brightness information, pixel value distribution, etc. The task of this node is to extract information related to the pixel-level features of the image to help measure the dryness or looseness of the stool. These features are extracted to gain a more comprehensive understanding of the characteristics of the stool.
[0076] Input the first period motion video frame sequence of the first period motion video, which is a video clip containing continuous frames, traverse the first period motion video frame sequence, process each frame one by one, and for each frame, analyze the texture structure of the image through the texture feature extraction node, and capture the texture features of the image by using technologies such as calculating gray-level co-occurrence matrix, extracting local binary pattern (LBP), and calculating directional gradient histogram (HOG); similarly, for each frame, analyze the pixel-level features of the image through the pixel feature extraction node, including extracting color information, brightness information, pixel value distribution, etc., to generate pixel features. Continue this process until all frames of the first period motion video frame sequence are traversed, forming a texture feature sequence and a pixel feature sequence.
[0077] Baseline texture features are texture features defined by experts based on known stool samples or standards. These features include specific texture patterns, structures, color distribution, etc. These features represent the baseline texture characteristics of stool under known circumstances. Experts can select a set of representative texture features as the baseline texture features of stool based on their experience and knowledge.
[0078] Similarly, benchmark pixel features are pixel-level features defined by experts based on known stool samples or standards. These features include the range of color values, brightness, hue distribution, etc. These features represent the benchmark pixel characteristics of stool under known circumstances. Experts can select a set of representative pixel features as the benchmark pixel features of stool based on their experience and knowledge.
[0079] The above-mentioned benchmark texture features and benchmark pixel features are defined based on expert knowledge and standards. The purpose is to compare the texture features and pixel features of the stool image to be tested with known standards to evaluate the dryness or looseness of the stool. These benchmark features are the basic reference points for measuring stool characteristics.
[0080] Each feature in the texture feature sequence is traversed one by one. For each texture feature, the distance between it and the benchmark texture feature is calculated. This can be done using different distance measurement methods, such as Euclidean distance, Manhattan distance, cosine similarity, etc. The distance value represents the similarity between the texture feature to be tested and the benchmark texture feature. A smaller distance value represents a higher similarity, and a larger distance value represents a lower similarity. For each texture feature, a distance evaluation value is generated. These evaluation values form a first distance evaluation value sequence, which includes the distance evaluation results of the texture feature to be tested and the benchmark texture feature.
[0081] The pixel feature sequence is traversed in the same manner to perform distance evaluation to generate a second distance evaluation value sequence, which will not be described in detail here for the sake of brevity.
[0082] Traverse all evaluation values in the first distance evaluation value sequence, sum these evaluation values, and then divide them by the total number of evaluation values to calculate the average value. The average value represents the average difference between the texture features of the stool image to be tested and the benchmark texture features. The mean calculation result is used as the first feature distance. The first feature distance is a metric used to describe the average difference between the texture features of the stool image to be tested and the benchmark texture features.
[0083] The second distance evaluation value sequence is averaged in the same manner to generate a second characteristic distance, which is used to describe the average difference between the pixel features of the stool image to be tested and the reference pixel features.
[0084] The first feature distance and the second feature distance are normalized respectively. Normalization maps the distance value to a specific range, usually between 0 and 1. Normalization can use various methods, such as Min-Max normalization, to ensure that the two distance values have a similar range.
[0085] The normalized first feature distance and the second feature distance are added together to generate a density coefficient, which is used to indicate the dryness or dryness of the stool image to be tested. A lower density coefficient indicates thicker stool, and a higher density coefficient indicates thinner stool. This coefficient is used for subsequent health testing and status matching.
[0086] Collect stool temperature characteristics through temperature sensing elements;
[0087] The temperature sensing element is located inside the smart toilet or in contact with the surface of the toilet to accurately measure the temperature of the stool. When the user's stool enters the smart toilet and contacts the temperature sensing element, the temperature sensing element begins to measure the temperature of the stool and generates a stool temperature feature. This temperature feature represents the actual temperature of the stool and can reflect the user's gastrointestinal health status, such as whether there are fever symptoms or other temperature-related problems. Collecting temperature information helps monitor any abnormal changes in the stool and thus detect potential health problems early.
[0088] Furthermore, the temperature sensing element is used to collect the temperature characteristics of the stool, including:
[0089] Collecting a stool temperature feature set through the temperature sensing element;
[0090] A central tendency analysis is performed on the stool temperature feature set to generate the stool temperature feature.
[0091] When the user defecates into the smart toilet, a temperature sensing element is used to measure and record the user's stool temperature at preset time intervals. These recorded measurement values constitute a stool temperature feature set. The temperature feature represents the actual temperature of the stool. The preset time interval can be set according to actual conditions and specific needs.
[0092] Central tendency analysis is a statistical data analysis technique used to measure the trend or central position in a data set. In this step, central tendency analysis is used to analyze the stool temperature feature set to understand the central tendency or typical value of stool temperature in the entire data set. Specifically, statistical methods such as mean, median, mode, etc. are used to analyze the central tendency of the stool temperature feature set. For example, the mean of the stool temperature in the stool temperature feature set is calculated to obtain a numerical value representing the central tendency of the stool temperature in the entire data set. Based on the results of the central tendency analysis, the stool temperature feature is generated, which represents the central tendency of the stool temperature.
[0093] According to the user identity tag, obtain the user age and user gender;
[0094] Based on the user identity tag, the tag is used to look up the user's age and gender information in a database that stores user information, where the user's age and gender are information provided when the user registers or sets up the smart toilet, or can be updated in subsequent user configurations. This information is used to personalize the user's health analysis and evaluation, which helps provide more personalized and accurate health recommendations.
[0095] Based on the user age and the user gender, frequently associated state matching is performed on the stool color feature, the thinness coefficient and the stool temperature feature to obtain a high-frequency state set;
[0096] Frequently associated state matching is a data mining technology used to identify feature combinations that frequently appear together in a data set. Specifically, the user's age and gender information is used as a condition to screen out stool color characteristics, thinness coefficient, and stool temperature characteristics that are related to a specific age and gender combination. By matching age and gender information with other features, high-frequency state sets related to different populations can be identified. High-frequency state sets are generated based on matching conditions and association rules. They contain stool color characteristics, thinness coefficient, and stool temperature characteristics that frequently appear under a specific age and gender combination. These high-frequency state sets represent gastrointestinal health conditions related to the user's age and gender.
[0097] Furthermore, based on the user age and the user gender, the stool color feature, the thinness coefficient and the stool temperature feature are matched with frequent associated states to obtain a high-frequency state set, including:
[0098] Taking the user age and the user gender as constraints, collecting samples of the stool color feature, the thinness coefficient and the stool temperature feature in a cloud database to obtain a first gastrointestinal state type set, wherein the first gastrointestinal state type set has a retrieval frequency tag set;
[0099] Based on the retrieval frequency tag set, a second gastrointestinal state type set that meets the retrieval frequency threshold is filtered from the first gastrointestinal state type set and is set as the high-frequency state set.
[0100] The user's age and gender are used as constraints for sample collection, which means that gastrointestinal status data is only obtained from samples that meet these constraints. A cloud database is accessed, which contains multiple recorded gastrointestinal status sample data, including stool color characteristics, thinness coefficient, and stool temperature characteristics.
[0101] According to the constraint conditions, sample data that meet the conditions are screened out, and the screened sample data are subjected to category analysis to determine different gastrointestinal state types, wherein each state type has a retrieval frequency label set for describing the frequency of occurrence of the state in a specific age and gender group. According to age, gender, stool color characteristics, thinness coefficient, and stool temperature characteristics, a first gastrointestinal state type set is generated, and this set includes different gastrointestinal state types, each type of which is accompanied by a retrieval frequency label set.
[0102] Set a retrieval frequency threshold, which is a numerical threshold used to filter out state types whose frequencies in the retrieval frequency label set are higher than or equal to the threshold. It can be set according to specific needs and actual conditions. Traverse each state type in the first gastrointestinal state type set, compare its retrieval frequency label with the retrieval frequency threshold. If the frequency of a state type is higher than or equal to the threshold, filter it into the second gastrointestinal state type set. Based on all the filtered state types in the second gastrointestinal state type set, a high-frequency state set is formed. This set includes gastrointestinal state types that appear more frequently in specific age and gender groups.
[0103] The high frequency state set is sorted according to the trigger frequency coefficient, and a high frequency state sorting result is obtained and sent to the user end.
[0104] The trigger frequency coefficient is an indicator used to measure the frequency of occurrence of each feature combination in the high-frequency state set. It reflects the significance of a specific feature combination. A high trigger frequency coefficient means that the feature combination often appears in the high-frequency state set, and a low trigger frequency coefficient means that the feature combination appears less frequently. The data will be sorted according to the trigger frequency coefficient, where the feature combination with a higher trigger frequency coefficient will be ranked first, indicating that it is more significant. This sorting can help users more easily identify gastrointestinal health trends related to their age and gender.
[0105] The sorted high-frequency status set will be sent to the user for review or analysis by medical professionals. Users can view the sorted results to understand the gastrointestinal health status related to their age and gender, as well as measures or suggestions that may need to be taken. This sorting helps users better understand their health condition so that they can take appropriate actions.
[0106] In summary, the health detection method based on the smart toilet provided in the embodiment of the present application has the following technical effects:
[0107] 1. Through the smart toilet, the user interacts with the login information when activating and meeting the preset status, thereby obtaining the user's identity tag and realizing accurate user identity recognition;
[0108] 2. Using image sensing elements, the color characteristics and dilution coefficient of stool are successfully collected, which helps to more comprehensively understand the state and health of stool; using temperature sensing elements, the temperature characteristics of stool are successfully collected, providing additional physiological information to assist in health detection;
[0109] 3. Based on the user identity tag, the user's age and gender information is successfully obtained, providing basic data for subsequent personalized analysis;
[0110] 4. By associating and matching the user's age and gender with multiple health parameters, a high-frequency state set is successfully obtained, which helps to discover health status patterns related to specific age and gender. The high-frequency state set is sorted according to the trigger frequency coefficient and sent to the user end, so that users can get targeted suggestions based on their health status.
[0111] In summary, this health detection method based on smart toilets solves the technical problem that traditional toilets have a single function and cannot adapt to the multi-functional needs of smart homes, resulting in a poor user experience. It provides non-invasive, multi-parameter comprehensive analysis and personalized health monitoring, realizes the health detection function of the toilet, ensures the convenience and accuracy of health monitoring, and improves the user experience.
[0112] Embodiment 2
[0113] Based on the same inventive concept as the health detection method based on the smart toilet in the aforementioned embodiment, Figure 2 As shown, the present application provides a health detection system based on a smart toilet, which is applied to a health detection system based on a smart toilet. The system is communicatively connected to the smart toilet, and the smart toilet includes an image sensor element and a temperature sensor element. The system includes:
[0114] An identity tag acquisition module 10, which is used to exchange user terminal login information and obtain a user identity tag when the user activates the smart toilet and meets a preset state;
[0115] A stool color feature acquisition module 20, the stool color feature acquisition module 20 is used to acquire stool color features and a rarefaction coefficient through an image sensor element;
[0116] The stool temperature characteristic acquisition module 30 is used to acquire stool temperature characteristics through a temperature sensing element;
[0117] A user information acquisition module 40, the user information acquisition module 40 is used to obtain the user age and user gender according to the user identity tag;
[0118] A high-frequency state acquisition module 50, the high-frequency state acquisition module 50 is used to perform frequent association state matching on the stool color feature, the thinness coefficient and the stool temperature feature based on the user age and the user gender to obtain a high-frequency state set;
[0119] The sorting result acquisition module 60 is used to sort the high-frequency state set according to the trigger frequency coefficient, obtain the high-frequency state sorting result, and send it to the user end.
[0120] Furthermore, the system comprises:
[0121] A motion video acquisition module, the motion video acquisition module is used to acquire a defecation motion video through the image sensor element, wherein the defecation motion video includes a motion endpoint image;
[0122] A rarefaction coefficient generating module, the rarefaction coefficient generating module is used to activate the dynamic feature analysis component, process the defecation motion video, and generate the rarefaction coefficient;
[0123] The stool color feature generation module is used to activate the static feature analysis component, process the movement endpoint image, and generate the stool color feature.
[0124] Furthermore, the system further includes a rarefaction coefficient acquisition module to perform the following operation steps:
[0125] The defecation motion video is processed through the video segmentation channel of the dynamic feature analysis component to generate a first cycle motion video, a second cycle motion video, and an Nth cycle motion video, wherein N is an integer, and N≥1;
[0126] Through the dry-thin feature analysis channel of the dynamic feature analysis component, the first period motion video, the second period motion video, and the Nth period motion video are traversed for processing to generate a first thinness coefficient, a second thinness coefficient, and the Nth thinness coefficient;
[0127] Performing cluster analysis on the first rarefaction coefficient, the second rarefaction coefficient, and up to the Nth rarefaction coefficient to obtain a first characteristic rarefaction coefficient, a second characteristic rarefaction coefficient, and up to the Mth characteristic rarefaction coefficient, where M is an integer and N≥M≥1;
[0128] The first characteristic dilution coefficient, the second characteristic dilution coefficient, and up to the Mth characteristic dilution coefficient are subjected to maximum value selection to obtain the dilution coefficient.
[0129] Furthermore, the system also includes a stool color feature generation module to perform the following operation steps:
[0130] activating the static feature analysis component, processing the motion endpoint image, and obtaining a stool color feature set;
[0131] Performing a central tendency analysis on the stool color feature set to generate the stool color feature;
[0132] Among them, the static feature analysis component is obtained by supervised configuration of the convolutional neural network through the stool image record dataset and the stool RGB feature identification dataset.
[0133] Furthermore, the system further includes a motion video generation module to perform the following operation steps:
[0134] Decomposing the defecation motion video to obtain a defecation motion video frame sequence, wherein the defecation motion video frame sequence has a collection position label sequence and a collection time label sequence;
[0135] Based on the acquisition position tag sequence, extracting a first defecation motion video frame set belonging to the motion starting position from the defecation motion video frame sequence;
[0136] Traversing the first defecation motion video frame set, activating the intermittent feature identification node of the video segmentation channel for processing, and obtaining a second defecation motion video frame set, wherein the second defecation motion video frame set is a video frame with an intermittent feature identification;
[0137] Obtaining a motion cycle sequence according to the acquisition time tag sequence and the second defecation motion video frame set;
[0138] According to the motion cycle sequence, the period segmentation node of the video segmentation channel is activated to process the defecation motion video to generate the first period motion video, the second period motion video until the Nth period motion video.
[0139] Furthermore, the system further includes a rarefaction coefficient generation module to perform the following operation steps:
[0140] The dry-sparse feature parsing channel includes a texture feature extraction node and a pixel feature extraction node;
[0141] According to the texture feature extraction node and the pixel feature extraction node, traverse the first periodic motion video frame sequence of the first periodic motion video to extract texture features, and generate a texture feature sequence and a pixel feature sequence;
[0142] Configure reference texture features and reference pixel features;
[0143] Traversing the texture feature sequence and the reference texture feature to perform distance evaluation and generate a first distance evaluation value sequence;
[0144] Traversing the pixel feature sequence and the reference pixel feature to perform distance evaluation and generate a second distance evaluation value sequence;
[0145] Performing mean calculation on the first distance evaluation value sequence to generate a first characteristic distance, and performing mean calculation on the second distance evaluation value sequence to generate a second characteristic distance;
[0146] The first characteristic distance and the second characteristic distance are respectively normalized and then added to generate the rarefaction coefficient.
[0147] Furthermore, the system also includes a stool temperature feature generation module to perform the following operation steps:
[0148] Collecting a stool temperature feature set through the temperature sensing element;
[0149] A central tendency analysis is performed on the stool temperature feature set to generate the stool temperature feature.
[0150] Furthermore, the system further includes a high-frequency state set acquisition module to perform the following operation steps:
[0151] Taking the user age and the user gender as constraints, collecting samples of the stool color feature, the thinness coefficient and the stool temperature feature in a cloud database to obtain a first gastrointestinal state type set, wherein the first gastrointestinal state type set has a retrieval frequency tag set;
[0152] Based on the retrieval frequency tag set, a second gastrointestinal state type set that meets the retrieval frequency threshold is filtered from the first gastrointestinal state type set and is set as the high-frequency state set.
[0153] Through the above detailed description of the health detection method based on the smart toilet in this specification, those skilled in the art can clearly understand the health detection system based on the smart toilet in this embodiment. For the device disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant parts can be referred to the method part description.
[0154] The above description of the disclosed embodiments enables those skilled in the art to implement or use the present application. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application will not be limited to the embodiments shown herein, but will conform to the widest scope consistent with the principles and novel features disclosed herein.
Claims
1. A detection method based on an intelligent toilet, characterized in that: Applied to a detection system based on a smart toilet, the system is in communication connection with the smart toilet, the smart toilet includes an image sensing element and a temperature sensing element, and the method includes: When the user activates the smart toilet and meets the preset status, the user terminal login information is exchanged to obtain the user identity tag; The stool color characteristics and dilution coefficient are collected through the image sensor element; Collect stool temperature characteristics through temperature sensing elements; According to the user identity tag, obtain the user age and user gender; Based on the user age and the user gender, frequently associated state matching is performed on the stool color feature, the thinness coefficient and the stool temperature feature to obtain a high-frequency state set; Sorting the high-frequency state set according to the trigger frequency coefficient, obtaining a high-frequency state sorting result, and sending it to the user end; Through the image sensor element, the stool color characteristics and dilution coefficient are collected, including: Collecting a defecation movement video through the image sensor element, wherein the defecation movement video includes a movement endpoint image; Activate the dynamic feature analysis component to process the stool motion video and generate the rarefaction coefficient; Activate the static feature analysis component to process the motion endpoint image to generate the stool color feature; Activating the dynamic feature analysis component, processing the defecation motion video, and generating the rarefaction coefficient, including: The defecation motion video is processed through the video segmentation channel of the dynamic feature analysis component to generate a first cycle motion video, a second cycle motion video, and an Nth cycle motion video, wherein N is an integer, and N≥1; Through the dry-thin feature analysis channel of the dynamic feature analysis component, the first period motion video, the second period motion video, and the Nth period motion video are traversed for processing to generate a first thinness coefficient, a second thinness coefficient, and the Nth thinness coefficient; Performing cluster analysis on the first rarefaction coefficient, the second rarefaction coefficient, and up to the Nth rarefaction coefficient to obtain a first characteristic rarefaction coefficient, a second characteristic rarefaction coefficient, and up to the Mth characteristic rarefaction coefficient, where M is an integer and N≥M≥1; Selecting the maximum value of the first characteristic dilution coefficient, the second characteristic dilution coefficient, and the Mth characteristic dilution coefficient to obtain the dilution coefficient; Activating the static feature analysis component, processing the motion endpoint image, and generating the stool color feature, including: activating the static feature analysis component, processing the motion endpoint image, and obtaining a stool color feature set; Performing a central tendency analysis on the stool color feature set to generate the stool color feature; The static feature analysis component is obtained by supervised configuration of the convolutional neural network through the stool image record dataset and the stool RGB feature identification dataset; The defecation motion video is processed through the video segmentation channel of the dynamic feature analysis component to generate a first cycle motion video, a second cycle motion video, and finally an Nth cycle motion video, including: Decomposing the defecation motion video to obtain a defecation motion video frame sequence, wherein the defecation motion video frame sequence has a collection position label sequence and a collection time label sequence; Based on the acquisition position tag sequence, extracting a first defecation motion video frame set belonging to the motion starting position from the defecation motion video frame sequence; Traversing the first defecation motion video frame set, activating the intermittent feature identification node of the video segmentation channel for processing, and obtaining a second defecation motion video frame set, wherein the second defecation motion video frame set is a video frame with an intermittent feature identification; Obtaining a motion cycle sequence according to the acquisition time tag sequence and the second defecation motion video frame set; According to the motion cycle sequence, the period segmentation node of the video segmentation channel is activated to process the defecation motion video to generate the first period motion video, the second period motion video until the Nth period motion video.
2. The method according to claim 1, characterized in that Through the dry-thin feature parsing channel of the dynamic feature parsing component, the first period motion video, the second period motion video, and the Nth period motion video are traversed for processing to generate a first thinness coefficient, a second thinness coefficient, and the Nth thinness coefficient, including: The dry-sparse feature parsing channel includes a texture feature extraction node and a pixel feature extraction node; According to the texture feature extraction node and the pixel feature extraction node, traverse the first periodic motion video frame sequence of the first periodic motion video to extract texture features, and generate a texture feature sequence and a pixel feature sequence; Configure reference texture features and reference pixel features; Traversing the texture feature sequence and the reference texture feature to perform distance evaluation and generate a first distance evaluation value sequence; Traversing the pixel feature sequence and the reference pixel feature to perform distance evaluation and generate a second distance evaluation value sequence; Performing mean calculation on the first distance evaluation value sequence to generate a first characteristic distance, and performing mean calculation on the second distance evaluation value sequence to generate a second characteristic distance; The first characteristic distance and the second characteristic distance are respectively normalized and then added to generate the rarefaction coefficient.
3. The method according to claim 1, characterized in that Through the temperature sensing element, the temperature characteristics of the stool are collected, including: Collecting a stool temperature feature set through the temperature sensing element; A central tendency analysis is performed on the stool temperature feature set to generate the stool temperature feature.
4. The method according to claim 1, characterized in that Based on the user age and the user gender, the stool color feature, the thinness coefficient and the stool temperature feature are matched with frequent associated states to obtain a high-frequency state set, including: Taking the user age and the user gender as constraints, collecting samples of the stool color feature, the thinness coefficient and the stool temperature feature in a cloud database to obtain a first gastrointestinal state type set, wherein the first gastrointestinal state type set has a retrieval frequency tag set; Based on the retrieval frequency tag set, a second gastrointestinal state type set that meets the retrieval frequency threshold is filtered from the first gastrointestinal state type set and is set as the high-frequency state set.
5. The detection system based on the intelligent toilet is characterized by: The system is in communication connection with a smart toilet, and the smart toilet includes an image sensor element and a temperature sensor element, and is used to implement the detection method based on the smart toilet according to any one of claims 1 to 4. The system includes: An identity tag acquisition module, which is used to exchange user terminal login information and obtain a user identity tag when the user activates the smart toilet and meets a preset state; A stool color feature acquisition module, the stool color feature acquisition module is used to acquire stool color features and a dilution coefficient through an image sensor element; A stool temperature characteristic acquisition module, which is used to acquire stool temperature characteristics through a temperature sensing element; A user information acquisition module, the user information acquisition module is used to obtain the user age and user gender according to the user identity tag; A high-frequency state acquisition module, the high-frequency state acquisition module is used to perform frequent association state matching on the stool color feature, the thinness coefficient and the stool temperature feature based on the user age and the user gender to obtain a high-frequency state set; A sorting result acquisition module, the sorting result acquisition module is used to sort the high-frequency state set according to the trigger frequency coefficient, obtain the high-frequency state sorting result, and send it to the user end; A motion video acquisition module, the motion video acquisition module is used to acquire a defecation motion video through the image sensor element, wherein the defecation motion video includes a motion endpoint image; A rarefaction coefficient generating module, the rarefaction coefficient generating module is used to activate the dynamic feature analysis component, process the defecation motion video, and generate the rarefaction coefficient; A stool color feature generation module, the stool color feature generation module is used to activate the static feature analysis component, process the movement endpoint image, and generate the stool color feature; The system also includes a rarefaction coefficient acquisition module to perform the following operation steps: The defecation motion video is processed through the video segmentation channel of the dynamic feature analysis component to generate a first cycle motion video, a second cycle motion video, and an Nth cycle motion video, wherein N is an integer, and N≥1; Through the dry-thin feature analysis channel of the dynamic feature analysis component, the first period motion video, the second period motion video, and the Nth period motion video are traversed for processing to generate a first thinness coefficient, a second thinness coefficient, and the Nth thinness coefficient; Performing cluster analysis on the first rarefaction coefficient, the second rarefaction coefficient, and up to the Nth rarefaction coefficient to obtain a first characteristic rarefaction coefficient, a second characteristic rarefaction coefficient, and up to the Mth characteristic rarefaction coefficient, where M is an integer and N≥M≥1; Selecting the maximum value of the first characteristic dilution coefficient, the second characteristic dilution coefficient, and the Mth characteristic dilution coefficient to obtain the dilution coefficient; The system also includes a stool color feature generation module to perform the following operation steps: activating the static feature analysis component, processing the motion endpoint image, and obtaining a stool color feature set; Performing a central tendency analysis on the stool color feature set to generate the stool color feature; The static feature analysis component is obtained by supervised configuration of the convolutional neural network through the stool image record dataset and the stool RGB feature identification dataset; The system also includes a motion video generation module to perform the following operation steps: Decomposing the defecation motion video to obtain a defecation motion video frame sequence, wherein the defecation motion video frame sequence has a collection position label sequence and a collection time label sequence; Based on the acquisition position tag sequence, extracting a first defecation motion video frame set belonging to the motion starting position from the defecation motion video frame sequence; Traversing the first defecation motion video frame set, activating the intermittent feature identification node of the video segmentation channel for processing, and obtaining a second defecation motion video frame set, wherein the second defecation motion video frame set is a video frame with an intermittent feature identification; Obtaining a motion cycle sequence according to the acquisition time tag sequence and the second defecation motion video frame set; According to the motion cycle sequence, the period segmentation node of the video segmentation channel is activated to process the defecation motion video to generate the first period motion video, the second period motion video until the Nth period motion video.
