Recommendation method for health services for the elderly based on bone conduction

Abstract

The present invention belongs to the field of health detection, and specifically relates to a method for recommending health services for the elderly based on bone conduction. The method disclosed in the present invention includes: collecting the sound of a human head and converting the sound into an audio signal; extracting identity feature information and health feature information in the audio signal; generating a request for health service recommendation based on the identity feature information and health feature information, and sending it to a server; and recommending health services to users based on the health service recommendation information returned by the server. The method of the present invention can realize the monitoring of human health based on bone conduction, and provides a user-friendly health monitoring method.

Description

Technical Field

[0001] The present invention belongs to the field of health detection, and in particular relates to a health service recommendation method for the elderly based on bone conduction. Background Art

[0002] Pillows are common daily necessities. People spend a lot of time sleeping in a day. The intelligence of pillows can improve people's life experience. Electronic devices are set in pillows to improve the functionality of pillows, that is, smart pillows.

[0003] For the elderly, smart pillows implant miniaturized and flexible smart controllers, various sensors and other electronic devices into the pillows, which can effectively monitor whether the user's various indicators are normal, and provide timely feedback to the person in charge so that appropriate measures can be taken for treatment. However, conventional ones only focus on data collection, and some use big data for processing. Because the information they need to provide contains private information and needs to rely on sufficient large model support, this poses a challenge to the health and privacy of users. Summary of the invention

[0004] An object of the present invention is to overcome the defects of the prior art and provide a method for solving the problems in the prior art to achieve monitoring of user health and form service recommendations.

[0005] The present invention first discloses a health service recommendation method for the elderly based on bone conduction, comprising:

[0006] collecting sounds from the occipital part and the contact part, and converting the sounds into audio signals;

[0007] Extracting identity feature information and health feature information from audio signals;

[0008] Generate a request for health service recommendation based on the identity feature information and the health feature information, and send it to the server;

[0009] Recommend health services to users based on the health service recommendation information returned by the server.

[0010] According to one embodiment of the present invention, the bone conduction monitoring device is a pillow, which includes a plurality of sound source sampling groups, each of which includes an audio transmitting unit and an audio receiving unit that are adjacently arranged.

[0011] According to an embodiment of the present invention, each sound source sampling group includes adjacently arranged audio transmitting units and audio receiving units separated by 6-8 cm; and the number of the audio transmitting units is 2, and the number of the audio receiving units is 1.

[0012] According to an embodiment of the present invention, the audio receiving unit is an ultrasonic transducer.

[0013] According to an embodiment of the present invention, a pressure sensor is further included, which is arranged below the audio transmitting unit and the audio receiving unit, and the sound source sampling group used for the test is determined according to the average value of the pressure sensor.

[0014] According to one embodiment of the present invention, identity feature information is extracted based on sound features in a frequency spectrum between 50 Hz and 2.5 kHz.

[0015] According to one embodiment of the present invention, health conditions are monitored based on a humidity sensor and a temperature sensor.

[0016] According to one embodiment of the present invention, audio of 100-200 Hz, 200-300 Hz, 300-500 Hz and 500-800 Hz in an audio signal is received, the loudness of the T time window is superimposed, the sum of the loudness of the audio of 100-200 Hz, 200-300 Hz, 300-500 Hz and 500-800 Hz is calculated, the sum of the loudness of the audio of 100-200 Hz, 200-300 Hz, 300-500 Hz and 500-800 Hz is standardized to obtain identity feature information, and whether the identity feature information is the same user is determined according to the difference between the identity feature information and the identity feature information corresponding to the standard loudness.

[0017] According to one embodiment of the present invention, an audio signal is received, the audio signal and a user identification are sent to a server, and a health service recommendation is performed based on a response of the server, wherein the health service recommendation is performed through an audio transmission unit.

[0018] According to one embodiment of the present invention, by monitoring the user's breathing frequency, temperature, humidity information, and the sound of the pillow and contact part, it is determined whether the user is in a sleeping state;

[0019] And execute health service recommendations when the user is in a non-sleeping state.

[0020] The present invention further discloses a bone conduction monitoring device, wherein the bone conduction is a pillow, and the pillow includes a plurality of sound source sampling groups, each of which includes an audio transmitting unit and an audio receiving unit that are adjacently arranged.

[0021] The present invention can achieve the following beneficial effects:

[0022] For user health monitoring, the pillow is used as a carrier to measure body temperature, pillow fitness, environmental sound and user voice.

[0023] The characteristic information of the human body is obtained through a device set close to the head, and an identity identification is formed based on this. A service recommendation request can be formed based on the identity characteristic information and health characteristic information, and a service recommendation is formed based on the service recommendation request. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] Figure 1 The flowchart schematically shows the method for recommending health services for the elderly based on bone conduction according to the present invention.

[0025] Figure 2 A schematic diagram schematically shows a flow chart of monitoring pillow status of a health service recommendation method for the elderly based on bone conduction according to the present invention;

[0026] Figure 3 The flowchart of monitoring pillow status of the bone conduction-based health service recommendation method for the elderly according to the present invention is schematically shown. DETAILED DESCRIPTION

[0027] In order to enable those skilled in the art to better understand the solution of the present invention, the technical solution in the embodiment of the present invention will be clearly described below in conjunction with the drawings in the embodiment of the present invention. Obviously, the described embodiment is a part of the embodiment of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present invention.

[0028] The term "comprising" and any variations thereof in the specification and claims of the present invention and the above drawings are intended to cover non-exclusive inclusions. For example, a process, method or system, product or device comprising a series of steps or units is not limited to the listed steps or units, but may optionally include steps or units not listed, or may optionally include other steps or units inherent to these processes, methods, products or devices.

[0029] The purpose of the present invention is to provide a method for recommending health services for users, especially elderly users, and the health services are implemented based on scenarios that are frequently used in daily life.

[0030] Figure 1 A schematic diagram schematically shows a flow chart of a method for recommending health services for the elderly based on bone conduction according to the present invention, as shown in FIG. Figure 1 As shown, in one embodiment of the present invention, a method for recommending health services for the elderly based on bone conduction is disclosed, comprising:

[0031] collecting sounds from the occipital part and the contact part, and converting the sounds into audio signals;

[0032] Extracting identity feature information and health feature information from audio signals;

[0033] Generate a request for health service recommendation based on the identity feature information and the health feature information, and send it to the server;

[0034] Recommend health services to users based on the health service recommendation information returned by the server.

[0035] The above method can achieve:

[0036] For user health monitoring, the pillow is used as a carrier to measure body temperature, pillow fitness, environmental sound and user voice.

[0037] The characteristic information of the human body is obtained through a device set close to the head, and an identity identification is formed based on this. A service recommendation request can be formed based on the identity characteristic information and health characteristic information, and a service recommendation is formed based on the service recommendation request.

[0038] The bone conduction involved in the present invention is a sound conduction method, that is, by converting sound into mechanical vibrations of different frequencies, sound waves are transmitted through the human skull, bony labyrinth, inner ear lymph, spiral organ, auditory nerve, and auditory center. Compared with the classic sound conduction in which the diaphragm generates sound waves, bone conduction saves many steps of sound wave transmission, can achieve clear sound restoration in a noisy environment, and the sound waves will not affect others because they spread in the air.

[0039] The present invention receives sound through a sound receiving structure arranged on the pillow to achieve user identification and obtain health-related information, thereby avoiding interference of wearing specific instruments on the human body.

[0040] The above health recommendation method can be performed based on the following health service recommendation device, which is a pillow, and the pillow includes a plurality of sound source sampling groups, each of which includes an audio transmitting unit and an audio receiving unit that are adjacently arranged;

[0041] Each sound source sampling group includes an audio transmitting unit and an audio receiving unit that are adjacently arranged at a distance of 6-8 cm; the number of the audio transmitting units is 2, and the number of the audio receiving unit is 1;

[0042] The bone conduction unit is a bone conduction speaker, which is matched with one or more microphone units for use;

[0043] The audio receiving unit is an ultrasonic transducer and a microphone;

[0044] The ultrasonic transducer includes an ultrasonic Doppler probe. A plurality of ultrasonic Doppler probes can be provided. The ultrasonic Doppler probes can be provided on the pillow so as to contact the human head. When the sound of the human head is received, the sound is converted into an audio signal for processing.

[0045] The ultrasonic Doppler probe includes a 2MHz or 3MHz transducer, which can be selectively converted into an audio signal for processing;

[0046] The ultrasound Doppler probe may be electrically connected to one or more amplifying circuits, sampling circuits, memories, buffers, and processors.

[0047] The pillow includes several sound source sampling groups for obtaining different sounds when the pillow is used. The sounds come from the environment and the sounds inside the user's body. After the environmental sounds are processed, it can be obtained whether there is noise around, and the sounds inside the user's body can be used to obtain information such as snoring, heartbeat and other health information.

[0048] By placing a bone conduction monitoring device on the pillow, the health status of the human body can be determined based on daily scenarios. Since sleep time occupies a large part of the day, analyzing it can obtain more stable information and avoid potential health risks when no one is watching. For elderly users, when the sleep time is short, by monitoring the user's status, sleep analysis can be provided, and the emission of specific sound waves can be used to assist in improving the user's quality of life.

[0049] According to one embodiment of the present invention, each sound source sampling group includes an audio transmitting unit and an audio receiving unit that are adjacently arranged at an interval of 6-8 cm; and the number of the audio transmitting units is 2, and the number of the audio receiving unit is 1. For example, a pillow can be provided with 3-4 groups of sound source sampling groups. The sound source sampling group can be in direct contact with the human head in the form of a soft cushion to provide a better sound reception effect. At present, pillows containing bone conduction headphones already exist on the market, and the description of their structure is omitted here.

[0050] The bone conduction unit used in the present invention is a bone conduction speaker, which can be further matched with one or more microphone units for use. Some attempts have been made to set an ultrasonic transducer as a microphone unit, but a separate microphone for collecting sounds inside the human body and ambient sounds can achieve better results.

[0051] That is, setting up an ultrasonic transducer can obtain the sound inside the human body, while the sound obtained by the microphone includes not only the sound inside the human body, but also the sound of the environment. The microphone alone cannot quickly obtain the characteristic information inside the human body, but after sampling the sound source obtained by the microphone and comparing it with the ultrasonic transducer, the sound that is different from the environment can be obtained. For the superposition and removal of sound waves in a frequency band, reference can be made to existing signal processing.

[0052] For example, a high-pass filter is used to pre-emphasize and frame-window the sound source obtained by at least a microphone to obtain multiple frame signals corresponding to each sound source signal; a fast Fourier transform is performed on the multiple frame signals to obtain spectral features; the frequency features are used as environmental sound features, and the base of the obtained human body sound is removed based on this. During the removal, adaptive filtering or spectral subtraction method is used to remove irrelevant environmental sounds to obtain more sounds containing human body sound features.

[0053] Furthermore, the location of the sound source sampling group should be considered. Taking into account the sleeping posture of the person and the actual length of the pillow, the sound source sampling groups should be set at intervals to balance the sampling, sound source effect and overall cost.

[0054] According to an embodiment of the present invention, the audio receiving unit is an ultrasonic transducer.

[0055] The ultrasonic transducer comprises an ultrasonic Doppler probe, and a plurality of the ultrasonic Doppler probes can be provided. By being provided on a pillow, the ultrasonic transducer can contact the human head, and when receiving the sound of the human head, the sound is converted into an audio signal for processing.

[0056] The available ultrasound Doppler probes include 2MHz or 3MHz, and the transducer can be optionally converted into an audio signal for processing.

[0057] The ultrasonic Doppler probe can be electrically connected to one or more amplifying circuits, sampling circuits, memories, buffers and processors. The electrical connection here refers to the connection for transmitting communication signals such as current signals, voltage signals, pulse signals, etc. through cables, data lines, etc. In order to reduce the corresponding volume during design, chips and circuits with low power consumption should be selected. The chip can be a central processing unit, or other general-purpose processors, digital signal processors, application-specific integrated circuits, off-the-shelf programmable gate arrays or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components, etc.

[0058] The ultrasonic Doppler probe is used to collect sound signals from the human head. The ultrasonic Doppler probe can acquire corresponding signals from the human head, such as the ear side, head and breathing sounds.

[0059] The ultrasonic Doppler probe here includes a piezoelectric ceramic sensor, which can be covered with a film to ensure protection against water and to ensure that the signal does not drift, through which a voltage signal can be obtained.

[0060] The optional amplifier circuit can be used to amplify the sound signal, and a bipolar transistor or a field effect transistor can be selected.

[0061] The sampling circuit here can be implemented by any sampling device with frequency sampling function. Considering the metabolism and heart rate range of the elderly, sampling can be performed within the frequency range of 10-500Hz, and a 16, 24 or 32-bit analog-to-digital conversion device can be selected.

[0062] Since the heartbeat of a healthy individual and the internal sounds of the human body are stable over a period of time, the signal obtained after sampling the sound is also predictable within a time span.

[0063] However, simply relying on sound of one frequency cannot form user identity recognition. If sounds in different frequency ranges are obtained and weighted calculations are performed to obtain feature values, it will have recognition significance.

[0064] For this, information data of 20 people were collected through cluster analysis, corresponding to 30Hz, 50Hz, 100Hz, 200Hz and 500Hz. Multiple different data were obtained by acquiring audio signals of specified frequencies. By using three or more of 30Hz, 50Hz, 100Hz, 200Hz and 500Hz as factors for cluster analysis, it was found that they had good correlation and the accuracy was higher than 90%. Therefore, three of 30Hz, 50Hz, 100Hz, 200Hz and 500Hz can be selected for signal conversion based on this, and the loudness of the corresponding converted sound signals can be obtained, and the identity feature information can be obtained according to the proportion of each loudness.

[0065] Since this process is to include more information, no filter may be set to filter out noise.

[0066] Furthermore, since this process does not seek to achieve a long-term monitoring effect, a judgment can be made at the time of first contact without the need for continuous sound collection while the user is sleeping.

[0067] According to an embodiment of the present invention, a pressure sensor is further included, which is arranged below the audio transmitting unit and the audio receiving unit, and the sound source sampling group used for the test is determined according to the average value of the pressure sensor.

[0068] When performing service recommendations, different recommendation operations can be performed according to the scenarios, wherein the scenario judgment can be determined based on the contact information.

[0069] If there is no initialized identity feature information when the user touches the pillow for the first time, the readings of all pressure sensors are obtained, and the pressure sensor with the maximum pressure is obtained, which is used as the main sensor, and its corresponding audio transmitting unit and audio receiving unit are further used, and the sound source sampling group is used to collect the sound source.

[0070] If the user has touched the pillow and has initialized identity feature information, the pressure sensor can be periodically activated to determine the actual sampled sound source sampling group. During this process, only information related to the user's health should be collected, without re-collecting sound sources related to the identity information.

[0071] If all the pressure sensors cannot obtain pressure values ​​within the threshold, a timer is started. If the value of the pressure sensor does not increase to exceed the threshold within a period of time, it can be determined that the user has left the pillow.

[0072] Further, such as Figure 2 If all pressure sensors cannot obtain pressure values ​​within the threshold, the timer is started. If the value of the pressure sensor does not increase by more than the threshold within a time threshold T1, it can be determined that the user has left the pillow. At this time, the second timer is started. If the value of the pressure sensor does not increase within another time threshold T2, it is determined that there is no need to monitor the pillow state.

[0073] That is, by setting the pressure sensor, the sound reception process can be optimized in a targeted manner.

[0074] Furthermore, the pressure sensor and time can be used to determine whether the user is suffering from insomnia, and targeted audio can be emitted through bone conduction headphones to help the user sleep.

[0075] According to one embodiment of the present invention, identity feature information is extracted based on sound features in a frequency spectrum between 50 Hz and 2.5 kHz.

[0076] As before, information data of 20 people were collected through cluster analysis. SPSS cluster analysis determined that three of 30Hz, 50Hz, 100Hz, 200Hz and 500Hz could be selected for signal conversion, and the corresponding loudness of the converted sound signal could be obtained, and identity feature information could be obtained based on the proportion of each loudness.

[0077] Specifically, when determining according to the loudness value of a single frequency, for example, 30 Hz, 100 Hz and 200 Hz are selected, their loudness is standardized, and the weights are set to 0.33 respectively. Then, when the loudness after standardization is 0.6, 0.7, and 1 respectively, the overall eigenvalue can be obtained by calculating the square of the loudness as 0.33*0.6*0.6+0.33*0.7*0.7+0.33*1*1=0.1188+0.200+0.33=0.694. If the eigenvalue measured by the user is within this range (such as a threshold of 5% up and down), the users are considered to be the same user.

[0078] The present invention performs identity authentication based on bone conduction based on the following characteristic information:

[0079] When using a pillow with sound collection equipment, the surrounding environment, the vibrations generated by the human vocal cords, and the vibrations generated by the human head can be collected by the microphone, where the base frequencies of the environment and human voice are between 80Hz and 300Hz, and the volume of the environment is similar within a period of time, and the base can be determined by collecting audio signals within a period of time and superimposing them, while the frequency of the bone conduction bite sound is between 100Hz and 2.5kHz. This also reveals that if it is possible to apply short-term high-frequency sound to the human head and receive the sound in the bone conduction frequency band, it is possible to identify the object using the pillow based on this.

[0080] To achieve this process, it should be based on the following approach:

[0081] Based on the previous theory and verification, if a wider frequency is monitored, not only more data needs to be collected, but the analysis is more difficult and unnecessary. If the differences in human bodies and the usage scenarios of pillows are taken into consideration, the problem to be solved for the extraction of identity information is the identification of one or a limited number of objects, and during use, an object will continuously generate multiple data, which are only useful for identifying the object, not for all objects.

[0082] Therefore, the actual improvement of the present invention is based on the fact that when an object uses a pillow, it can produce an approximate response to the transmitted audio of a fixed frequency. The above-mentioned technology is commonly used in the prior art. The common one is to transmit ultrasonic waves to obtain the reflection of the human body and thus obtain an image. Here, in order to obtain data with distinction, the identity feature information here is the ratio of audio signals of different wavelength frequencies. Obviously, when the posture contact angles of the human body are inconsistent, a variety of methods should be set to perform pattern recognition and matching. If combined with the data of the pressure sensor, the difficulty of the problem can be further reduced.

[0083] According to one embodiment of the present invention, health conditions are monitored based on a humidity sensor and a temperature sensor.

[0084] The temperature sensor is used to obtain the temperature of the human head to determine whether the body temperature is normal; the humidity sensor is used to measure the humidity of the pillow. The humidity of the pillow can be caused by sweat on the human head or due to excessive indoor humidity. Therefore, it is necessary to combine the data of multiple sensors to determine whether the human body is healthy.

[0085] According to one embodiment of the present invention, audio of 100-200 Hz, 200-300 Hz, 300-500 Hz and 500-800 Hz in an audio signal is received, the loudness of the T time window is superimposed, the sum of the loudness of the audio of 100-200 Hz, 200-300 Hz, 300-500 Hz and 500-800 Hz is calculated, the sum of the loudness of the audio of 100-200 Hz, 200-300 Hz, 300-500 Hz and 500-800 Hz is standardized to obtain identity feature information, and whether the identity feature information is the same user is determined according to the difference between the identity feature information and the identity feature information corresponding to the standard loudness.

[0086] When the loudness of a single frequency is calculated, the loudness of a single frequency is determined; when a microphone is selected as the sound source, the loudness in a region is integrated to obtain a characteristic value, such as selecting the audio of 100-200Hz, 200-300Hz, 300-500Hz and 500-800Hz in the received audio signal, superimposing the loudness of the T time window, calculating the sum of the loudness of the audio of 100-200Hz, 200-300Hz, 300-500Hz and 500-800Hz, and normalizing the sum of the loudness of the audio of 100-200Hz, 200-300Hz, 300-500Hz and 500-800Hz to obtain identity feature information, and determining whether it is the same user based on the difference between the identity feature information and the identity feature information corresponding to the standard loudness. Its calculation scheme is similar to the previous one.

[0087] According to one embodiment of the present invention, an audio signal is received, the audio signal and a user identification are sent to a server, and a health service recommendation is performed based on a response of the server, wherein the health service recommendation is performed through an audio transmission unit.

[0088] When the user's identification is selected, the audio signal features are extracted. Since these audio signal features come from bone conduction and ultrasonic transducers, most of the ambient sounds can be filtered out, thereby better protecting the user's privacy and being able to meet the recommendation of services based on the user's characteristic voice.

[0089] When making service recommendations, since sound features are obtained through microphones and ultrasonic transducers, the features can be recorded or extracted locally, and combined with body temperature, pressure sensor and humidity data, the data is sent to the health recommendation server to obtain recommended services.

[0090] To this end, the recommended device can be configured with an IoT SIM card for communication, or connected to the network through a wireless connection module.

[0091] According to an embodiment of the present invention, please refer to Figure 3 , by monitoring the user's breathing frequency, temperature, humidity information and the sound of the pillow and contact parts, it can determine whether the user is in a sleeping state;

[0092] And execute health service recommendations when the user is in a non-sleeping state.

[0093] The sleep state can be further judged in combination with time, such as setting 21:00 to 7:00 the next day as sleep time, and not executing health service recommendations in this time zone, otherwise after receiving the server's recommendation message, executing the server's recommended operation.

[0094] Outside of the above-mentioned time, such as when the user touches the pillow during the day, the bone conduction device can emit a sound based on the recommendation information from the server, thereby realizing the recommendation of health services.

[0095] When performing health service recommendations, a recommended processing network can be built based on typical health data. For example, a neural network can be built based on machine learning. By building a human health monitoring model, a series of data and health values ​​are obtained as inspection data, and the model is iteratively trained through parameter adjustment and optimization. After obtaining the most accurate model, it is applied, and by inputting existing health data, human health monitoring results are output and feedback is given. In this process, new data can be received, and duplicate data can be removed to obtain an updated network.

[0096] That is, the neural network can have the following inputs {M, E, T, P, S}, where M is the sound inside the human body acquired by the transducer on the pillow, E is the environmental sound, T is the temperature of the human body, P is the data acquired by the activated pressure sensor, and S is the status data, which further includes heart rate, breathing rate and health identification data.

[0097] During training, the Mel-frequency cepstral coefficients of internal human sounds and environmental sounds, human body temperature and pressure data are used as input, and the heart rate, respiratory rate and health information obtained through other means are used as output for training.

[0098] When training, the neural network used can be a multi-layer type, consisting of an input layer, multiple hidden layers and an output layer. The input layer, hidden layer and output layer have units for simulating the synaptic function of brain neural circuits, and the neural network is optimized by machine learning to obtain acceptable prediction data.

[0099] In the present invention, human physiological parameters may include body temperature, heart rate, respiratory rate, environmental sound and snoring information data, wherein body temperature data may be obtained based on a sensor set on a pillow, heart rate data may be obtained based on received sound feature data, respiratory rate may be obtained using a sound processing network, environmental sound may be obtained based on environmental sound, and snoring information data may be obtained based on environmental sound. When adjusting the input, the corresponding network may be adaptively adjusted.

[0100] Furthermore, health services can be configured as output data of the configuration, that is, corresponding service recommendations are given when feature data is received.

[0101] The embodiments described above are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features may be replaced by equivalents. Such modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention, and should all be included in the protection scope of the present invention.

Claims

1. A method for recommending health services for the elderly based on bone conduction, characterized in that: include: collecting sounds from the occipital part and the contact part, and converting the sounds into audio signals; Extracting identity feature information and health feature information from the audio signal, wherein the identity feature information is a ratio of audio signals of different wavelength frequencies; Generate a request for health service recommendation based on the identity feature information and the health feature information, and send it to the server; Recommend health services to users based on the health service recommendation information returned by the server.

2. The bone conduction-based health service recommendation method for the elderly as claimed in claim 1, characterized in that The bone conduction monitoring device is a pillow, which includes a plurality of sound source sampling groups, each of which includes an audio transmitting unit and an audio receiving unit that are adjacently arranged.

3. The bone conduction-based health service recommendation method for the elderly as claimed in claim 2, characterized in that Each sound source sampling group includes an audio transmitting unit and an audio receiving unit that are adjacently arranged and spaced 6-8 cm apart; and the number of the audio transmitting units is 2, and the number of the audio receiving unit is 1.

4. The bone conduction-based health service recommendation method for the elderly as claimed in claim 2, characterized in that The audio receiving unit is an ultrasonic transducer.

5. The bone conduction-based health service recommendation method for the elderly as claimed in claim 4, characterized in that It also includes a pressure sensor, which is arranged below the audio transmitting unit and the audio receiving unit, and the sound source sampling group used for the test is determined according to the average value of the pressure sensor.

6. The bone conduction-based health service recommendation method for the elderly as claimed in claim 1, characterized in that The identity feature information is extracted based on the sound features in the frequency spectrum from 50Hz to 2.5kHz.

7. The bone conduction-based health service recommendation method for the elderly as claimed in claim 1, characterized in that Health condition monitoring based on humidity sensors and temperature sensors.

8. The bone conduction-based health service recommendation method for the elderly as claimed in claim 1, characterized in that Audio of 100-200 Hz, 200-300 Hz, 300-500 Hz and 500-800 Hz in an audio signal is received, the loudness of the T time window is superimposed, the sum of the loudness of the audio of 100-200 Hz, 200-300 Hz, 300-500 Hz and 500-800 Hz is calculated, the sum of the loudness of the audio of 100-200 Hz, 200-300 Hz, 300-500 Hz and 500-800 Hz is standardized to obtain identity feature information, and whether the identity feature information is the same user is determined according to the difference between the identity feature information and the identity feature information corresponding to the standard loudness.

9. The bone conduction-based health service recommendation method for the elderly as claimed in claim 1, characterized in that An audio signal is received, the audio signal and a user identification are sent to a server, and a health service recommendation is performed based on a response of the server, wherein the health service recommendation is performed through an audio transmission unit.

10. The bone conduction-based health service recommendation method for the elderly as claimed in claim 9, characterized in that By monitoring the user's breathing rate, temperature, humidity information, and the sound of the pillow and contact parts, it is determined whether the user is in a sleeping state; and when the user is in a non-sleeping state, health service recommendations are implemented.

Images