Watching care support device and system

Abstract

To provide a care support system for grasping psychological stress or feelings of care related persons while performing watching monitoring on a care-requiring person using the care support system, the care support system being capable of being installed more simply by the care-requiring person without professional knowledge, laying construction work, or adjustment.SOLUTION: A care support system includes a loudspeaker terminal device disposed in the vicinity of a care-requiring person and watching platform means in a cloud connected to the terminal device via a communication network. The loudspeaker terminal device has an interactive loudspeaker having a voice recognition function and an AI (artificial intelligence) assistant function. The watching platform means comprises voice analysis means for grasping a psychological stress state or an abnormal condition of the care-requiring person or a care-giver by extracting a keyword and a voice feature amount of the care-requiring person or the care-giver on the basis of voice data on the care-requiring person or the care-giver acquired by the loudspeaker terminal device.SELECTED DRAWING: Figure 3

Description

[Technical Field]

[0001] The present invention relates to a monitoring care support device and system that monitors the daily lives, health conditions, and illnesses of elderly people, hospital patients, and people living in nursing facilities, and also detects psychological stress and abnormal situations in the care or nursing of caregivers and nurses. [Background technology]

[0002] Monitoring of care recipients and hospitalized patients in nursing homes and hospitals and identifying abnormalities are currently carried out by caregivers and nurses making rounds or through contact from the care recipient or patient, placing a heavy burden on caregivers and nursing staff. Furthermore, as the aging society continues to expand, the number of elderly people living alone, such as those living alone, has become increasingly important, making home visits rather than institutional care, and increasing the need for monitoring and supervision of individuals' daily lives and health. However, for those who require care or supervision, such as hospitalized patients, elderly people, and people with physical disabilities (hereinafter referred to as care recipients), even when they need supervision or care, access to care and supervision services is becoming increasingly difficult due to a shortage of nurses, caregivers (hereinafter referred to as caregivers), and facilities, as well as rising costs. Meanwhile, caregivers are prone to experiencing considerable psychological stress due to violence, verbal abuse, and unreasonable behavior from care recipients, which has recently led to conflicts and acts of violence between caregivers and care recipients.

[0003] Surveillance cameras are used as a monitoring system in such nursing homes and hospitals, but this is not desirable from the perspective of privacy, and it is difficult for a monitor to continue monitoring at all times.In addition, various methods have been proposed, such as using various sensors such as pressure sensors, temperature sensors, and ultrasonic sensors to monitor the breathing and heart rate of a person requiring care, or to detect when the person gets out of bed or lands in bed (Patent Document 1, etc.).

[0004] However, conventional monitoring systems and devices require specialized knowledge and skills to install and set up systems and devices equipped with sensors, and installation work and adjustments cannot be easily performed. Furthermore, once such systems are installed, new work and specialized knowledge are required to discontinue, change, expand, or add to the system. This creates a lack of convenience. To solve these problems, an invention using a smart speaker, as shown in Patent Document 2, has been proposed.

[0005] Furthermore, it is desirable to be able to grasp the psychological stress levels of caregivers (especially caregivers) involved in caregiving and nursing work on a daily basis, to prevent conflicts and violent acts between caregivers and those in need of care, and to reduce the turnover rate of caregivers. The present invention is an improved invention of the technical disclosure shown in Patent Document 2, and is a technical disclosure that meets the demand for monitoring those in need of care, as well as visualizing and understanding the psychological stress levels of those in need of care and caregivers in caregiving and nursing. [Prior art documents] [Patent documents]

[0006] [Patent Document 1] Japanese Patent Application Laid-Open No. 2013-78477 [Patent Document 2] Japanese Patent Publication No. 2022-89113 Summary of the Invention [Problem to be solved by the invention]

[0007] The problem to be solved is to provide the following monitoring, care and nursing support device and system (hereinafter referred to as care support system) in consideration of the above situation. (1) Providing a care support system that can be easily installed by care recipients without requiring specialized knowledge, installation work, or adjustments. (2) To provide a care support system that monitors the person requiring care while also grasping the psychological stress or emotions of those involved in care (especially the caregiver). [Means for solving the problem]

[0008] In order to solve the above-mentioned problems, the care support system of the present invention is a care support system having a speaker terminal device placed near a person requiring care and a monitoring platform means in the cloud connected to the terminal device via a communication network, wherein the speaker terminal device has an interactive speaker with a voice recognition function and an artificial intelligence assistant function, and the monitoring platform means is characterized by having a voice analysis means that grasps the psychological stress state and abnormal situations of the person requiring care or caregiver by extracting voice features and keywords of the person requiring care or caregiver based on the voice data of the person requiring care or caregiver acquired by the speaker terminal device.

[0009] In addition, the care support system of the present invention can also be configured so that the voice analysis means registers normal voice data acquired in advance from the person requiring care or the caregiver as reference data, and makes a relative comparison between the voice data acquired by the speaker terminal device and the registered reference data to estimate the psychological stress state of the person requiring care or the caregiver.

[0010] In addition, the care support system of the present invention has a monitoring means for monitoring a person in need of care who is connected to the cloud via a communication network, and the voice analysis means can be configured to issue an alert signal to the monitoring means when the voice feature acquired from the speaker terminal device or its change tendency exceeds a predetermined value, or when the appearance frequency of an acquired keyword or its change tendency exceeds a predetermined value.

[0011] The care support system according to the present invention includes a monitoring camera installed in a room where the care-requiring person lives and connected to the monitoring means by the communication network means, and the monitoring camera is driven by the issuance of the alert signal. It can also be configured as follows.

[0012] In addition, the care support system of the present invention can also be configured so that the voice analysis means includes a memory and learning means for storing and learning voice features, keywords, and the psychological stress state of the person requiring care or the caregiver based on the voice data of the person requiring care or the caregiver acquired by the speaker terminal device.

[0013] The care support system according to the present invention can also be configured so that the speaker terminal device is attached to a ceiling rosette in a room where the care-requiring person lives. [Effects of the Invention]

[0014] According to the care support system of the present invention, an interactive speaker terminal device installed near a person requiring care is connected to a monitoring platform means in the cloud via a communication network, and the person requiring care uses this speaker terminal device to monitor the person on a daily basis. The monitoring platform means also has a function for analyzing voice data and extracting voice features or keywords of the person requiring care and the caregiver. Because the voice analysis means for analyzing the speaker's voice from the voice data is stored on the cloud side, a user can understand the daily activity status (including life and death status) of the person requiring care simply by installing a speaker terminal device (such as a commercially available AI speaker) without the need for specialized knowledge or installation work, etc.

[0015] Furthermore, the care support system according to the present invention can be used to monitor the psychological stress and mental health of the care recipient and the caregiver through voice analysis, thereby preventing accidents.Furthermore, by configuring the system to issue an alert signal when the voice features or keywords of the care recipient or the caregiver exceed a predetermined value, efficient monitoring and care support can be performed without the need for continuous monitoring. [Brief explanation of the drawings]

[0016] [Figure 1] FIG. 1 is a diagram illustrating the configuration of a care support device and system according to the present invention. [Figure 2] FIG. 2 is a schematic explanatory diagram showing the flow of an output signal from the voice analysis means according to the present invention. [Figure 3] FIG. 3 is a block diagram illustrating a configuration for generating an alert signal from a voice input according to the present invention. [Figure 4] FIG. 4 is a flow chart illustrating the main processing steps of the present invention. DETAILED DESCRIPTION OF THE INVENTION

[0017] An embodiment of the care support system according to the present invention will be described in detail below with reference to the drawings. Note that any explanatory diagrams and drawings described in the following examples are drawn as outlines or schematic diagrams for the purpose of explaining the present invention, and the actual dimensions and shapes are not particularly limited. Furthermore, the circuit configuration, dimensions, materials, shapes, relative arrangements, and uses of the components are intended for use in homes and care / medical facilities, but unless otherwise specified, the technical scope of the invention is not intended to be limited to these. Furthermore, in this invention, the cloud-based monitoring platform means refers to software programs specialized for providing care support and monitoring to care recipients and caregivers, and the infrastructure environment that runs these programs.

[0018] In recent years, interactive speakers with voice recognition and artificial intelligence assistant functions, known as smart speakers or AI speakers (hereinafter referred to as interactive speakers or speaker terminal devices), have become popular. These interactive speakers are devices that enable various operations to be performed by the AI ​​assist function when the user speaks to the speaker via a communication network such as the Internet. By speaking to or giving commands through these interactive speakers, the AI ​​(artificial intelligence) assist function can obtain various information such as weather forecasts, traffic information, and news by voice, control the operation of home appliances, curtains, doors, etc., and can also be used for entertainment such as music and readings.

[0019] Many of these interactive speakers have AI (artificial intelligence) assistance functions, voice recognition functions, and control / command functions stored in the cloud of the service provider (platform side) connected via a communication network, while the speaker terminal only has a voice input device (microphone), a speaker, and a communication device for connecting to the network.This means that the speaker terminal can be an inexpensive, small device, and general-purpose products are easily available on the market.

[0020] The present invention provides a care support device and system that uses an easily available interactive speaker terminal device to be used as an AI speaker for everyday purposes such as obtaining various information, controlling home appliances, and playing music, while also monitoring and watching over those in need of care without the need for any special equipment or processing. [Example]

[0021] Fig. 1 is an explanatory diagram of the configuration of a care support system according to the present invention. In the residence 1 of a care-requiring person 10, an interactive speaker terminal device 11 is placed near the care-requiring person 10. This speaker terminal device 11 is small and lightweight, making it easily portable, and the care-requiring person 10 can wear it at all times, for example, by hanging it around their neck. Furthermore, if the care-requiring person 10 is recuperating in bed, it can also be installed on the ceiling above the bed or near the bed. When the speaker terminal device 11 is installed on the ceiling, it is desirable to give the microphone in the terminal device directional characteristics.

[0022] Furthermore, the speaker terminal devices 11 can be installed in multiple rooms so that the care-requiring person can respond from any room even if they move. In this case, multiple speaker terminal devices 11 can be installed by directly attaching them to ceiling rosettes on the ceiling.

[0023] The speaker terminal device 11 is connected by wireless communication to a communication connection device 13 for connecting to an external communication network 12 within the residence 1. Many of these connections can be configured using WiFi connections, but are not limited to this and can also be configured using telephone lines, power line circuits (PLC), etc. The communication network 12 is connected to the Internet and to various services provided by various Internet providers and platform providers, and receives voices and conversations uttered by the care-requiring person 10 from the speaker terminal device 11 as commands, obtains information such as traffic information and weather forecasts based on the commands, and causes the speaker terminal device 11 to output voices or to control home appliances, curtains, windows, etc. connected to the communication network 12.

[0024] This communication network 12 is connected to a monitoring platform means 15 in the cloud 14. This monitoring platform means 15 stores a voice analysis means 20 that analyzes the voice of the care recipient 10 sent from the speaker terminal device 11. This voice analysis means 20 acquires and analyzes the voice of the care recipient 10 and the voice of the caregiver 16 or the conversation situation between the two, and outputs the psychological stress levels of the care recipient 10 and the caregiver 16 or abnormal situations in care from the voice situation output by the voice analysis means 15.

[0025] The communication network 12 is also connected to a monitoring means 17 installed in a monitoring center or the like that watches over and monitors the care-requiring person 10, and functions to watch over the care-requiring person 10 or give care instructions in response to commands output from the voice analysis means 20 of the watching platform means 15. The monitoring means 17 can also be configured to start up a monitoring camera 18 installed in the residence 1 of the care-requiring person 10 in the event of an emergency or other need, and monitor the condition of the care-requiring person 10.

[0026] 2 is a schematic explanatory diagram showing the flow of signals from the voice analysis means 15 to the monitoring means 17. The voice analysis means 15 is configured to include a voice feature extraction unit 21 that calculates features from input voice, and a keyword extraction unit 22 that extracts and processes keywords. The two output signals of the voice feature extraction unit 21 and the keyword extraction unit 22 are signals that analyze the input voices and conversations of the care-requiring person 10 and the caregiver 16 and correlate with the level of psychological stress and abnormal situations.

[0027] As for speech features, changes in parameters such as speech pitch (frequency), pitch period, and speech power of conversations and speech are known to indicate a person's psychological stress level, while as for keywords, there are keywords indicating emergency / abnormal situations, abusive language, abuse, etc. The output signal of the speech analysis means 20 is sent to psychological stress level estimation means 23, which estimates the psychological stress level and the possibility of an abnormal situation based on the speech features of the input speech, the number and frequency of occurrence of keywords, and their changing trends. This stress level estimation means 23 estimates the stress level (high, medium, low, etc.) and the possibility of an abnormal situation occurring according to the input stress level and urgency level.

[0028] When the stress level estimation means 23 estimates that the stress level is high or that there is a high possibility of an abnormal situation, an alert signal is issued from the alert signal issuing means 24. This alert signal is sent to the monitoring response means 25 of the monitoring center 17 via the communication network 12. The monitoring response means 25 can be configured to use an alert lamp or a warning sound as the alert (warning) signal issued by the alert signal issuing means 24. When an alert signal is issued, a caregiver will rush to the scene in response to the monitoring means 17 or will check the situation via the speaker terminal device 11. Furthermore, when audio indicating abusive language or abuse between the care-requiring person 10 and / or the caregiver 16 is detected as a possible abnormal situation, the monitoring camera 18 for the care-requiring person 10, which is normally not in operation, can be activated to check the condition of the care-requiring person 10 through camera footage.

[0029] Here, the monitoring means 17 is installed in a facility that monitors the care-requiring person 10 or in a location accessible to the caregiver, and is expected to be installed in a central control room in a hospital, a care facility, a monitoring center at a security company, etc., but is not limited to these, and is also expected to be installed in places where the care-requiring person is cared for and monitored, such as visiting caregivers and relatives.

[0030] Next, details of the configuration for estimating psychological stress levels and abnormal situations through voice analysis will be described. FIG. 3 is a block diagram illustrating a configuration for estimating psychological stress levels and abnormal situations based on voice input from the care recipient 10 and the caregiver 16 and issuing an alert signal. Voices recorded by a microphone in a speaker terminal device 11 are input to a monitoring platform means 15 stored in a cloud 14 via a communication network 12. A general-purpose smart speaker or the like can be used as this speaker terminal device 11. Preferably, the system is configured to detect only when the care recipient 10 and the caregiver 16 are speaking, and not to activate in other situations such as silence or background noise.

[0031] The input voice signal is divided into fixed intervals (frames), and multiple voice frames are stored sequentially in the frame recording unit 31. The number of frames is preferably as large as possible depending on the clarity of the voice input and the recording capacity, but is not particularly limited. Voice frames stored for a predetermined period are overwritten and updated when the storage capacity reaches the limited area. Non-voice parts, such as silent parts, noise, and background sounds in the stored frames, are deleted and discarded, and only the voice parts are extracted and sent to the feature extraction unit 21 and keyword extraction unit 22 of the voice analysis means 15.

[0032] The feature extraction unit 21 extracts features for each input frame. These features include parameters including the pitch (frequency), pitch period, and voice power of the voice input. These parameters are known to correlate with changes in emotions such as joy, anger, sadness, and pleasure, and although there are individual differences, the voice pitch, pitch period, and voice power change with emotional fluctuations. Therefore, by comparing the features with those of a previously acquired sample of the subject's normal speech, the subject's emotional fluctuations can be determined.

[0033] It is desirable to obtain speech samples in advance from subjects engaged in care, nursing, and monitoring, such as the care-requiring person 10 and the caregiver 16, and store the features in the feature database 32. However, it is also possible to obtain sample speech from subjects while the system is running to some extent, store it in the recording and learning means 33, extract features, and store them in the feature database 32 for use.

[0034] The speech features of the subject extracted by the feature extraction unit 21 are sent to the occurrence amount calculation unit 35, which calculates the tendency (vector) of the speech features. The features (speech pitch, pitch period, speech power, etc.) extracted for each predetermined frame are compared with the feature database 32, and at the same time, the difference between each frame is obtained to determine how they are changing over time. In other words, the amount of instantaneous (sudden) large changes in speech pitch, voice volume (power), or pitch period, or amounts that differ from normal vector changes, is calculated.

[0035] Meanwhile, the keyword extraction unit 22 extracts keywords correlated with joy, anger, sadness, and happiness from the speech of the input frames. These keywords are mainly extracted from abusive language, name-calling, attacks on physical or psychological defects, prejudice, harassment, and scathing remarks, particularly in the context of caregiving and nursing, but are not limited to these and also include representative keywords for joy, anger, sadness, and happiness. A language dictionary database 34 can be used for this keyword extraction.

[0036] The language dictionary database 34 contains pre-prepared speech patterns for specific languages, such as slanderous language, and languages ​​in specific fields, such as emotions. While these speech patterns can be extracted using general-purpose standard speech patterns used in text-to-speech conversion, it is preferable to acquire a speech sample from the subject in advance and register it in the language dictionary for more reliable extraction. Similarly to the acquisition of samples for extracting the subject's features, sample speech from the subject can be acquired while the system is running for a certain period of time, stored in the recording and learning means 33, and related keywords can be extracted and stored in the language dictionary database 34 for use. The keyword extraction unit 22 extracts keywords from specific fields, such as slanderous language and emotions, from the care recipient 10 and / or caregiver 16. The keywords extracted by the keyword extraction unit 22 are sent to the frequency calculation unit 36, which calculates the frequency and percentage of each keyword for each field.

[0037] The output data of the feature occurrence amount calculation unit 35 and the output data calculated by the keyword frequency amount calculation unit 36 ​​are sent to the likelihood calculation unit 37. The likelihood calculation unit 37 calculates the feature change and trend, which are the output data of the occurrence amount calculation unit 35, and the trend and frequency of the keyword frequency, which are the output data of the frequency amount calculation unit 36, over a relatively long period. The feature extraction and keyword extraction described above are extracted for each frame and are instantaneous data. However, because the subject's emotions and stress state may rise and fall instantaneously (rapidly), it is necessary to grasp this frame-by-frame data over a predetermined period (e.g., every few minutes) longer than the frame interval. The likelihood calculation unit 37 extracts the data for each frame obtained by the speech feature extraction and keyword extraction as data accumulated or averaged over a predetermined period. The speech feature data and keyword data thus obtained by the likelihood calculation unit 37 are sent to the stress level estimation unit 38.

[0038] The stress level estimation unit 38 uses the changes, levels, and trends in the voice feature data and keyword frequency data to level the emotional fluctuations of the subject person requiring care and the caregiver, emotional differences from normal care and nursing conditions, and abnormal situations, and estimates the degree of accumulated psychological stress or the possibility of an abnormal situation occurring, etc. Such levels of psychological stress conditions are reference data that can be used for comparison and estimation to level and estimate whether or not there is a deviation from psychological stress in normal care and nursing, and are not intended to perform a medical stress diagnosis.

[0039] The quantification of psychological stress according to the present invention indicates a deviation from normal conditions, but by understanding the subject's stress and emotional ups and downs in conjunction with the frequency of keyword occurrence, it is possible to reduce the possibility of an abnormal situation occurring. For example, if abusive language appears frequently as a keyword in a situation where psychological stress is elevated or emotional ups and downs change significantly during caregiving or nursing, it is assumed that some abnormal situation has occurred between the care recipient 10 and the caregiver 16, and an alert signal can be issued, and an inquiry can be made from the monitoring center 17 via the interactive speaker 11, a caregiver can be dispatched, or the monitoring camera 18 can be activated to monitor the condition of the care recipient 10.

[0040] Furthermore, if the keyword extraction function of this monitoring system extracts an emergency keyword (for example, "Help!", "Someone come!", or "Scream") that indicates an emergency, it is possible that an emergency has occurred in the subject, even if the emotional fluctuations in the voice features do not indicate an abnormal situation, and it is possible that an alert signal will be issued. In this way, by providing an emergency keyword detection function, daily monitoring can be carried out via the speaker terminal device 11 even for people living alone or without a caregiver.

[0041] In the monitoring system of the present invention, voice data with different characteristics is acquired by extracting features and extracting keywords as the voice analysis means 15. The data obtained by extracting the features of the input voice (voice pitch, voice power, pitch period) is used to primarily grasp emotional fluctuations, and furthermore, keyword extraction, combined with the frequency of abusive language, allows for early estimation of psychological stress levels and the possibility of abnormal situations, making it possible to support the prevention and early detection of abnormal situations and accidents during caregiving and nursing.

[0042] In the present invention, large-capacity databases such as the feature database 32 in the voice analysis means 15, the memory / learning means 33, and the language dictionary database 34, as well as voice analysis processing programs that require fast processing speeds, are all stored on the cloud 14, and the speaker terminal device 11 placed near the care-requiring person 10 can be a commercially available interactive speaker 11 known as a smart speaker or AI speaker, and can be used as a monitoring support device for the care-requiring person. Such an interactive speaker terminal device 11 does not require any special specifications or programs, and can monitor the care-requiring person 10 simply by connecting to the monitoring platform in the cloud 14, and can also grasp the emotional ups and downs and stress levels of the caregiver and the care-requiring person during care and nursing.

[0043] Next, the voice analysis process that constitutes the present invention will be explained with reference to Figure 4. Figure 4 is a flow diagram showing the main processing steps required to implement the present invention. Voice analysis is initiated by an activation utterance from the care-requiring person 10 or the caregiver 16, such as a pre-registered utterance such as "Hello, Mr. / Ms. X!" or "Come on! Let's examine him / her!" However, if an emergency keyword is uttered, the system automatically enters an activated state and starts the voice analysis process.

[0044] When speech analysis begins, only voiced segments are detected (step 01: shown as S01, and so on). Unvoiced segments are not recorded or analyzed, and so recording capacity is omitted. The input speech in the detected voiced segments is divided into frame units of a predetermined period (S02), and recorded as multiple frame-unit speech (S03).

[0045] The presence or absence of an emergency keyword is first detected from the multiple frame-unit voices (S04). As described above, this emergency keyword is a voice that indicates an emergency situation or a situation where the care recipient 10 or the caregiver 16 is calling for help. However, even if an emergency keyword is detected, it does not necessarily mean that an emergency has occurred, as it may be a joke or a metaphorical expression in the conversation. Therefore, even if an emergency keyword is detected, the conversations and related phrases before and after the emergency keyword, as well as repetition of the emergency keyword, are detected (S05). This emergency keyword and the conversations and related phrases before and after it are referenced (S06). If it is determined that the emergency keyword has been uttered, a notification is sent to the monitoring means 17 in the monitoring center, and an emergency alert is output (S15). Here, it is also possible to not provide the urgency estimation step (S06), and if an emergency keyword is detected, it is considered to be an emergency and an alert signal is issued.

[0046] If no urgent keyword is detected in the urgent keyword detection process (S04), the output speech is subjected to both the speech feature detection process (S07) and the keyword detection process (S10). In the feature detection process (S07), data on one or a combination of speech pitch, volume power, and speech period is acquired, and the occurrence rate of speech features is calculated. Next, the calculated features are subjected to a process to calculate their frequency (S08).

[0047] The speech features identified in this process are then passed to a likelihood detection process (S09) to acquire their trends (vectors). In this process, sudden or gradual increases, decreases, or sudden changes in the features are identified.

[0048] Meanwhile, in the keyword detection step (S10), keywords related to psychological stress states related to caregiving and nursing are extracted by comparing them with a pre-configured dictionary database. The extracted keywords have their frequency calculated (S11), and their trends are obtained in the likelihood detection process (S12). Here, the feature likelihood detection step (S09) and keyword likelihood detection (S12) have the same function of detecting appearance trends (vectors), but the detection targets are different.

[0049] After each likelihood detection step is completed, the psychological stress level is determined based on whether the input voice features and / or keywords change or tendencies significantly different from normal, or whether a sudden rise or fall or discontinuity is detected in the change trend (vector) (S13). The determination result is displayed as a numerical scale ranging from high to low stress, or is classified and output as a level such as emergency, caution required, monitoring, or stable (S14).

[0050] The psychological stress level is determined, and the output stress estimate is continuously recorded and learned (S16), and is stored as psychological stress data in a feature database, keyword language database, etc., and is also fed back to the feature detection step processing. This completes the voice analysis process.

[0051] Next, an example in which the present invention described above is used as a monitoring system will be described. Normally, the speaker terminal device 11 is always connected to the communication network 12 by being constantly powered on, and its usage status can be confirmed at the monitoring center 17. The user (person requiring care 11) uses the speaker terminal device 11 on a daily basis to collect information such as news, weather forecasts, and traffic information, to play music, and to operate home appliances and devices, using the interactive response function of the speaker terminal device 11, so that the monitoring center 17 can understand that the person requiring care is living their daily life. In particular, even if the person requiring care 10 is not operating the terminal device 11, the presence of the person requiring care 10 can often be confirmed by the sounds of their movements and noise.

[0052] However, the monitoring means 17 in the monitoring center is configured to ask questions (e.g., "How are you?" and "How are you feeling today?") to check the user's safety at predetermined time intervals (e.g., every 4 to 6 hours) and wait for a response from the care-requiring person 10. These questions can also be set to be asked automatically at set time intervals. If there is no response from the care-requiring person 10 to these questions, an alert signal is sent to the monitoring means 17, and the monitoring means 17 as described above takes direct action on the care-requiring person 10.

[0053] When the care-requiring person 10 is away from home or is unable to communicate with or does not use the speaker terminal device 11 for a certain period of time due to sleeping or personal routine, the user (care-requiring person 11) stores this fact in the terminal device 11 or notifies the monitoring means 17 in advance and sets a question exclusion period. This exclusion period can also be set automatically by learning the user's daily routine.

[0054] Next, an example of monitoring psychological stress and emotional fluctuations of the caregiver 16 and the care recipient 10 during care and nursing work will be described. When the caregiver 16 (visiting caregiver or nurse) visits the care recipient 10 to provide care and nursing, or when the care recipient 10 requests support from the monitoring center 17, the caregiver 16 issues the activation voice as described above, thereby entering a state of watching over the caregiver 16 and monitoring stress and emotions.

[0055] While the caregiver 16 is providing care and nursing, the voices and conversations of both parties are analyzed and continuously recorded and learned by a voice analysis means 15 in the cloud 14 to analyze the emotions and stress states of the caregiver 16 and the care recipient 10. By recording and learning these emotions and stress states, the monitoring means 17 estimates changes or trends in the psychological stress and mental state of both parties. The monitoring means 17 periodically provides appropriate guidance based on the stress and emotional changes of the caregiver 16 and the care recipient 10, and issues necessary treatment or warnings.

[0056] If an emergency keyword or abusive language is detected, it is determined that this may be due to abusive language or violent behavior by the care-requiring person 10 or the caregiver 16, and an alert signal is sent to the monitoring means 17. When a keyword indicating such an emergency is uttered, the system automatically detects the emergency keyword uttered by any person, not limited to the care-requiring person 10 or the caregiver 17, and goes into monitoring mode, issuing an alert signal to the monitoring means 17 and enabling image monitoring by the monitoring camera 18.

[0057] In such a care support system, the speaker terminal device 11 is inexpensive and readily available, so no installation work or specialized knowledge is required. A monitoring system can be configured simply by installing an interactive speaker and connecting it to a monitoring platform for not only sick and physically disabled people in nursing homes, hospitals, and home care facilities, but also elderly people living alone and healthy people. Furthermore, as long as there is a communication connection environment such as Wi-Fi, the system can be easily installed in multiple rooms or anywhere. Furthermore, since the system of the present invention allows for continuous monitoring of one's own psychological stress and emotional ups and downs, it can be easily applied to issues with caregivers and those requiring care, accident prevention, and mental and physical health management. [Industrial Applicability]

[0058] With this invention, simply by installing a simple two-way speaker terminal device, it is possible to monitor a person in need of care and continuously grasp the psychological stress and emotional fluctuations of the caregiver and the person in need of care, without requiring any specialized knowledge or construction work. This means that systems can be easily constructed not only in the nursing and medical fields, but also in companies, factories, individuals, etc., expanding the scope of industrial applicability in a wide range of fields. [Explanation of symbols]

[0059] 10. People requiring care 11 Interactive speaker terminal device 12. Communication Networks 13 Communication connection devices (WiFi) 14. Cloud 15 Monitoring platform means 16. Caregiver 17 Monitoring measures 18. Surveillance Cameras 20 Voice analysis tools 21 Feature extraction unit 22 Keyword Extraction Section 23 Stress level estimation method 24 Alert signal issuing means 25 Surveillance Response Measures 31 Frame Recording Unit 32 Feature Database 33 Recording and learning tools 34 Language Dictionary Database 35 Appearance Calculation Unit 36 Frequency calculation part 37 Likelihood calculation section 38 Stress level estimation unit

Claims

1. A care support system having a speaker terminal device arranged near a person requiring care and a cloud-based monitoring platform means connected to the terminal device via a communication network, The speaker terminal device has an interactive speaker with a voice recognition function and an artificial intelligence assistant function, The care support system is characterized in that the monitoring platform means is equipped with a voice analysis means that grasps the psychological stress state and abnormal situations of the person requiring care or the caregiver by extracting voice features and keywords of the person requiring care or the caregiver based on the voice data of the person requiring care or the caregiver acquired by the speaker terminal device.

2. The care support system of claim 1, characterized in that the voice analysis means includes means for registering normal voice data acquired in advance from the person requiring care or the caregiver as reference data, comparing the voice data acquired by the speaker terminal device with the registered reference data, and estimating the psychological stress state of the person requiring care or the caregiver.

3. the care support system includes a monitoring means for monitoring a care-requiring person connected to the cloud via a communication network, The care support system of claim 2, characterized in that the voice analysis means issues an alert signal to the monitoring means when the voice feature acquired from the speaker terminal device or its change tendency exceeds a predetermined value, or when the appearance frequency of the acquired keyword or its change tendency exceeds a predetermined value.

4. the care support system includes a monitoring camera installed in a room where the care-requiring person resides and connected to the monitoring means via the communication network means; 4. The care support system according to claim 3, wherein the monitoring camera is activated in response to the issuance of the alert signal.

5. The care support system of claim 2, characterized in that the voice analysis means includes a memory / learning means for storing and learning voice features, keywords, and psychological stress states of the person requiring care or the caregiver based on the voice data of the person requiring care or the caregiver acquired by the speaker terminal device.

6. 3. The care support system according to claim 2, wherein the speaker terminal device is attached to a ceiling rosette in a room where the person requiring care lives.

Images