Data processing system

Abstract

The system according to the present embodiment includes a motion detection unit, an analysis unit, a notification unit, a fall detection unit, an instruction unit, and a health monitoring unit. The motion detection unit is configured to detect a motion. The analysis unit is configured to analyze information detected by the motion detection unit. The notification unit is configured to notify based on information analyzed by the analysis unit. The fall detection unit is configured to detect a fall. The analysis unit is further configured to analyze information detected by the fall detection unit. The instruction unit is configured to instruct an emergency response based on information analyzed by the analysis unit. The health monitoring unit is configured to monitor a health. The analysis unit is further configured to analyze information monitored by the health monitoring unit. The notification unit is further configured to notify based on information analyzed by the analysis unit.

Description

Technical Field

[0001] The technology disclosed herein relates to a system. Background Technology

[0002] Patent Document 1 discloses a personalized chatbot control method executed by at least one processor, the method comprising: receiving user speech; adding the user speech to a prompt containing instructions related to a chatbot role; encoding the prompt; and inputting the encoded prompt into a language model to generate chatbot speech in response to the user speech.

[0003] Patent document 1: Japanese Patent Application Publication No. 2022-180282.

[0004] Existing technologies present challenges in effectively monitoring the safety and health of elderly people living alone and in responding quickly to emergencies. Summary of the Invention

[0005] The system described in this embodiment includes: a motion detection unit, an analysis unit, a notification unit, a fall detection unit, an instruction unit, and a health monitoring unit. The motion detection unit detects motions. The analysis unit analyzes the information detected by the motion detection unit. The notification unit issues notifications based on the information analyzed by the analysis unit. The fall detection unit detects falls. The analysis unit also analyzes the information detected by the fall detection unit. The instruction unit instructs emergency responses based on the information analyzed by the analysis unit. The health monitoring unit monitors health. The analysis unit also analyzes the information monitored by the health monitoring unit. The notification unit also issues notifications based on the information analyzed by the analysis unit. Attached Figure Description

[0006] Figure 1 This is a conceptual diagram illustrating an example of the configuration of a data processing system according to the first embodiment.

[0007] Figure 2 This is a conceptual diagram illustrating an example of the main functions of the data processing apparatus and smart device according to the first embodiment.

[0008] Figure 3 This is a conceptual diagram illustrating an example of the data processing system configuration in the second embodiment.

[0009] Figure 4 This is a conceptual diagram illustrating an example of the main functions of the data processing device and smart glasses according to the second embodiment.

[0010] Figure 5 This is a conceptual diagram illustrating an example of the data processing system configuration in the third embodiment.

[0011] Figure 6This is a conceptual diagram illustrating an example of the main functions of the data processing device and head-mounted terminal according to the third embodiment.

[0012] Figure 7 This is a conceptual diagram illustrating an example of the data processing system configuration in the fourth embodiment.

[0013] Figure 8 This is a conceptual diagram illustrating an example of the functions of the main parts of the data processing device and robot according to the fourth embodiment.

[0014] Figure 9 It represents an emotion graph that maps multiple emotions.

[0015] Figure 10 It represents an emotion graph that maps multiple emotions.

[0016] Explanation of reference numerals in the attached figures: Data processing systems 10, 210, 310, and 410 12 Data processing device 14 Smart devices 214 Smart Glasses 314 Head-mounted terminal 414 Robot. Detailed Implementation

[0017] Hereinafter, an example of an implementation of the system involved in this disclosure will be described with reference to the accompanying drawings.

[0018] First, let's explain the terms used in the following description.

[0019] In the following embodiments, the processor (hereinafter referred to as "processor") can be a single computing device or a combination of multiple computing devices. Furthermore, a processor can be a single computing device or a combination of multiple computing devices. Examples of computing devices include CPU (Central Processing Unit), GPU (Graphics Processing Unit), GPGPU (General-Purpose computing on Graphics Processing Units), APU (Accelerated Processing Unit), or TPU (Tensor Processing Unit), etc.

[0020] In the following implementation, the labeled RAM (Random Access Memory) is a memory that temporarily stores information and is used by the processor as working memory.

[0021] In the following embodiments, the labeled memory is one or more non-volatile storage devices used to store various programs and various parameters, etc. Examples of non-volatile storage devices include flash memory (SSD (Solid State Drive)), disk (e.g., hard disk) or magnetic tape, etc.

[0022] In the following implementation, the labeled Communication I / F (Interface) is an interface that includes a communication processor and an antenna, etc. The Communication I / F is responsible for communication between multiple computers. Examples of communication standards applicable to the Communication I / F include wireless communication standards such as 5G (5th Generation Mobile Communication System), Wi-Fi (registered trademark), or Bluetooth (registered trademark).

[0023] In the following implementation, "A and / or B" is synonymous with "at least one of A and B". That is, "A and / or B" means that it can be only A, only B, or a combination of A and B. Furthermore, in this specification, when "and / or" connects more than three items, the same approach as "A and / or B" applies.

[0024] [First Implementation] Figure 1 An example of the configuration of the data processing system 10 according to the first embodiment is shown.

[0025] like Figure 1 As shown, the data processing system 10 includes a data processing device 12 and an intelligent device 14. An example of the data processing device 12 is a server.

[0026] The data processing device 12 includes a computer 22, a database 24, and a communication I / F 26. The computer 22 includes a processor 28, RAM 30, and memory 32. The processor 28, RAM 30, and memory 32 are connected to a bus 34. Furthermore, the database 24 and the communication I / F 26 are also connected to the bus 34. The communication I / F 26 is connected to a network 54. An example of the network 54 includes a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[0027] The smart device 14 includes a computer 36, a receiver 38, an output device 40, a camera 42, and a communication I / F 44. The computer 36 includes a processor 46, RAM 48, and memory 50. The processor 46, RAM 48, and memory 50 are connected to a bus 52. In addition, the receiver 38, output device 40, and camera 42 are also connected to the bus 52.

[0028] The receiving device 38 includes a touchscreen 38A and a microphone 38B, etc., for receiving user input. The touchscreen 38A receives user input generated by contact with an indicator (e.g., a pen or finger). The microphone 38B receives user input generated by sound by detecting the user's voice. The control unit 46A sends data representing user input received via the touchscreen 38A and microphone 38B to the data processing unit 12. In the data processing unit 12, a specific processing unit 290 (see...) Figure 2 Get the data that represents user input.

[0029] The output device 40 includes a display 40A and a speaker 40B, etc., and presents data to the user by outputting data in a user-perceptible form (e.g., sound and / or text). The display 40A displays visual information such as text and images according to instructions from the processor 46. The speaker 40B outputs sound according to instructions from the processor 46. The camera 42 is a small digital camera equipped with an optical system such as a lens, aperture, and shutter, and an imaging element such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor.

[0030] Communication I / F 44 is connected to network 54. Communication I / F 44 and 26 are responsible for sending and receiving various information between processor 46 and processor 28 via network 54.

[0031] Figure 2 An example of the main functions of the data processing device 12 and the smart device 14 is shown.

[0032] like Figure 2 As shown, in the data processing apparatus 12, specific processing is performed by the processor 28. A specific processing program 56 is stored in the memory 32. The specific processing program 56 is an example of a "program" as understood in this disclosure. The processor 28 reads the specific processing program 56 from the memory 32 and executes the read specific processing program 56 on the RAM 30. Specific processing is implemented by the processor 28 operating as a specific processing unit 290 based on the specific processing program 56 executed on the RAM 30.

[0033] The memory 32 stores a data generation model 58 and an emotion-specific model 59. The data generation model 58 and the emotion-specific model 59 are used by the specific processing unit 290. The specific processing unit 290 can use the emotion-specific model 59 to infer the user's emotions and perform specific processing based on the user's emotions. The emotion inference function using the emotion-specific model 59 (emotion-specific function) includes various inferences and predictions related to the user's emotions, but is not limited to this example. Furthermore, emotion inference and prediction also include, for example, emotion analysis (analysis).

[0034] In the smart device 14, specific processing is performed by the processor 46. A specific processing program 60 is stored in the memory 50. The specific processing program 60 is used in conjunction with the data processing system 10. The processor 46 reads the specific processing program 60 from the memory 50 and executes the read specific processing program 60 on the RAM 48. Specific processing is implemented by the processor 46 operating as a control unit 46A based on the specific processing program 60 executed on the RAM 48. Furthermore, the smart device 14 may also have the same data generation model and emotion-specific model as the data generation model 58 and the emotion-specific model 59, and use these models to perform the same processing as the specific processing unit 290.

[0035] Furthermore, other devices besides the data processing device 12 may also have the data generation model 58. For example, a server device (e.g., a generation server) may have the data generation model 58. In this case, the data processing device 12 obtains the processing results (prediction results, etc.) using the data generation model 58 by communicating with the server device that has the data generation model 58. In addition, the data processing device 12 may be a server device or a user-owned terminal device (e.g., a mobile phone, robot, home appliance, etc.). Next, an example of the processing performed by the data processing system 10 of the first embodiment will be described.

[0036] (Example) The system described in this invention is a solution that utilizes AI to support the safety and health of elderly people living alone. This system combines motion sensors, fall detection devices, and a medical monitoring system to ensure the safety of the elderly at home and support emergency response through AI assistance. Furthermore, it provides a safe and comfortable living environment by enhancing communication between users, family members, and caregivers. For example, motion sensors are used to detect the elderly person's movements. These sensors are installed throughout the room and can monitor the elderly person's movements in real time. For example, they can detect the elderly person's movements when they move around the room to identify any abnormalities. When an abnormality is detected, the AI ​​analyzes the information and notifies family members or caregivers if necessary. Next, fall detection devices are used to detect falls. These devices are worn by the elderly person and can detect falls immediately. For example, when an elderly person falls, the device sends information to the AI, which instructs on emergency response. This enables rapid response and ensures the safety of the elderly person. In addition, a medical monitoring system is used to monitor the elderly person's health status. The medical monitoring system can measure the elderly person's vital signs such as body temperature, heart rate, and blood pressure in real time and send the information to the AI. The AI ​​analyzes this data and notifies family members or caregivers when an abnormality is detected. For example, when an elderly person's body temperature rises sharply, the AI ​​analyzes this information and notifies family members or caregivers for a rapid response. Furthermore, it enhances communication between users and their families and caregivers. For instance, the AI ​​can periodically report the elderly person's condition to family members or caregivers, or enable direct communication between family members, caregivers, and the elderly person in emergencies. This allows for real-time monitoring of the elderly person's condition, ensuring their peace of mind. Thus, this invention, by combining motion sensors, fall detection devices, and medical monitoring systems, utilizes AI to support the safety and emergency response of the elderly at home. Moreover, by enhancing communication between users and their families and caregivers, it provides a safe and comfortable living environment. Therefore, this system, supporting the safety and health of elderly people living alone, assists in their safety and emergency response at home. Specifically, this system uses a dedicated data acquisition module to collect high-dimensional data (such as triaxial acceleration vectors, time-series tensors of temperature, humidity, and illuminance, and continuous value arrays of vital signs) from multiple sensor devices and aggregates it in real time to a central processing unit. The central processing unit first performs noise removal (such as Butterworth filtering and moving average), outlier removal, and standardization (such as Z-score normalization) in the data preprocessing department, and then inputs the data into the AI ​​inference module. The AI ​​inference module can use convolutional neural networks (CNN) or recurrent neural networks (RNN) for action detection, LSTM or GRU for fall detection, and multilayer perceptron (MLP) or decision tree-based ensemble learning (such as random forest) for health status analysis.Examples of AI inputs include 3D vector sequences from accelerometers sampled at 1-second intervals (e.g., t=0s: [0.12, -0.03, 0.98], t=1s: [0.15, -0.01, 0.97]), continuous arrays of body temperature, heart rate, and blood pressure values ​​every 5 minutes (e.g., body temperature 36.5℃, heart rate 72bpm, blood pressure 120 / 80mmHg), and binary sequences of ON / OFF statuses of motion sensors in each room (e.g., living room: 1, bedroom: 0, corridor: 1). The AI ​​outputs structured data such as anomaly detection scores (continuous values ​​from 0.0 to 1.0), anomaly type labels (e.g., fall, wandering, abnormal vital signs), and urgency levels (e.g., urgent, requires attention, normal). For example, output examples might include "Anomaly score 0.92, type: fall, urgency: urgent" or "Anomaly score 0.45, type: abnormal vital signs, urgency: requires attention." These outputs are compared with preset benchmark values ​​(e.g., anomaly scores above 0.8) in the threshold determination department. If the data exceeds the threshold, it is forwarded to the notification or instruction department. The notification department automatically selects the optimal notification method (push notification, SMS, email, voice call, etc.) to send the information, taking into account the family member's or caregiver's device information (smartphone, tablet, PC, etc.) and geographical location information (GPS coordinates, address). The instruction department automatically instructs actions such as calling an ambulance, immediately contacting nearby family members, and notifying nursing service institutions based on the urgency. Vital sign data from the medical monitoring system is analyzed using AI's time-series anomaly detection algorithms (e.g., autoregressive models, LSTM-based prediction models) to accurately detect sudden changes or deviations from the norm. In addition, the system databases past anomaly detection history and the response history of family members and caregivers for continuous online learning of the AI ​​model and personalized optimization of notification and instruction content. This not only automates manual operations but also achieves a comprehensive improvement in computer technologies, including high-dimensional analysis of sensor data, multi-module collaboration, real-time anomaly detection, automatic notification, and emergency instruction. The technological benefits include: improved anomaly detection accuracy (significantly reducing false positives and false negatives) compared to traditional manual care or simple threshold-based methods; faster response times (real-time notifications and instructions); improved data management efficiency (automatic recording and historical analysis); and optimized communication load (notifications only when necessary, distributed processing). Specific application areas include a variety of use cases such as care for elderly people living alone, home-based medical support, safety management of nursing facilities, remote health monitoring, support for people with disabilities, and child and pet care.

[0037] The safety support system described in this embodiment includes a motion detection unit, an analysis unit, a notification unit, a fall detection unit, an instruction unit, and a health monitoring unit. The motion detection unit detects the movements of the elderly person. For example, the motion detection unit is installed throughout the room to monitor the elderly person's movements in real time. The motion detection unit can detect movements such as walking, standing up, and sitting down. The motion detection unit can detect the elderly person's movements using motion sensors. Motion sensors are installed throughout the room to monitor the elderly person's movements in real time. For example, it can detect the elderly person's movements when they move around the room to confirm whether there are any abnormalities. The analysis unit analyzes the information detected by the motion detection unit. The analysis unit can use AI to analyze motion data and detect abnormalities. The analysis unit can also use machine learning algorithms to analyze motion data and detect abnormalities. The notification unit makes notifications based on the information analyzed by the analysis unit. For example, the notification unit notifies family members or caregivers based on information analyzed by AI. The notification unit notifies family members or caregivers when an abnormality is detected. The fall detection unit detects falls in the elderly person. The fall detection unit is worn on the elderly person's body and can detect falls immediately. The fall detection unit can detect falls using an accelerometer. An accelerometer is worn on the elderly person's body to immediately detect falls. An instruction unit instructs emergency response based on information analyzed by the analysis unit. For example, the instruction unit instructs emergency response based on information analyzed by AI. Upon detecting a fall, the instruction unit instructs the call for an ambulance or contact with family members. A health monitoring unit monitors the elderly person's health. The health monitoring unit can measure the elderly person's vital signs such as body temperature, heart rate, and blood pressure in real time. The health monitoring unit can monitor the elderly person's health status through a medical monitoring system. The medical monitoring system can measure the elderly person's vital signs such as body temperature, heart rate, and blood pressure in real time and send the information to AI. Thus, the safety support system involved in this embodiment supports the safety and health of the elderly person by detecting their movements, falls, and health status, analyzing, notifying, and responding to emergencies. Specifically, this safety support system, as a movement detection unit, installs multiple movement sensors (such as infrared sensors, ultrasonic sensors, image sensors, etc.) on the ceiling or walls of the room to acquire the elderly person's position coordinates and movement status (such as stillness, walking, pre-fall signs, etc.) as three-dimensional vector data at 1-second intervals. The motion detection unit sends this sensor data as a time-series tensor (e.g., a 30×3 matrix for a 30-second motion history) to the central processing unit. The parsing unit first performs preprocessing on the received motion data, such as noise removal (e.g., moving average filtering, outlier removal) and standardization (Z-score normalization), and then uses convolutional neural networks (CNN) or recurrent neural networks (RNN) for feature extraction and anomaly detection.The AI ​​input examples are continuous vector sequences such as t=0s: [0.1, 0.2, 0.0], t=1s: [0.15, 0.18, 0.02], etc., and the output is structured data such as anomaly score (0.0~1.0), anomaly type (e.g., fall, wandering, normal), and urgency level (urgent, need attention, normal). The fall detection unit wears an accelerometer or gyroscope sensor on the elderly person's waist or wrist to sample three-axis acceleration data at 100Hz (e.g., X:0.98, Y:-0.12, Z:0.05), and analyzes fall-specific patterns (e.g., rapid acceleration changes, posture changes) through time-series AI models such as LSTM or GRU. The fall detection unit's AI output examples are "fall score 0.95, urgency level: urgent" or "fall score 0.45, urgency level: need attention". Every 5 minutes, the Health Monitoring Department retrieves vital signs such as body temperature, heart rate, and blood pressure from the medical monitoring system and inputs them as a continuous value array (e.g., body temperature 36.7℃, heart rate 74 bpm, blood pressure 122 / 78 mmHg) into the AI. Health status analysis employs a Multilayer Perceptron (MLP) or decision tree ensemble (e.g., Random Forest), outputting anomaly scores and anomaly types (e.g., fever, tachycardia, hypotension). The Notification Department receives anomaly judgment results from the Analysis Department, Fall Detection Department, and Health Monitoring Department, compares them with preset benchmark values ​​in the Threshold Judgment Department (e.g., anomaly score above 0.8). If the threshold is exceeded, it automatically selects the optimal notification method (push notification, SMS, email, voice call, etc.) based on the family member's or caregiver's device information (smartphone, tablet, PC, etc.) and geographical location information (GPS coordinates, address). The Instruction Department automatically instructs actions such as calling an ambulance, immediately contacting nearby family members, and notifying nursing service providers based on the urgency level. This series of processes differs from traditional manual care or simple threshold judgment methods, achieving comprehensive improvements in computer technologies such as high-dimensional sensor data AI analysis, multi-module collaboration, real-time anomaly detection, notification, and indication. Technical benefits include improved anomaly detection accuracy (significantly reducing false positives and false negatives), faster response speed (real-time notifications and indications), improved data management efficiency (automatic recording and historical analysis), and optimized communication load (notifications only when necessary, distributed processing). Specific application areas include a variety of use cases such as care for elderly people living alone, home-based medical support, safety management of nursing facilities, remote health monitoring, support for people with disabilities, and child and pet care.

[0038] The motion detection unit is installed throughout the room to monitor the elderly person's movements in real time. "Real time" here refers to updates occurring at the second or minute level. The motion detection unit detects the elderly person's movements using motion sensors. For example, it detects the elderly person's movements as they move around the room to identify any abnormalities. Thus, by monitoring the elderly person's movements in real time, abnormalities can be detected early. Specifically, this motion detection unit installs various motion sensors (such as infrared sensors, ultrasonic sensors, image sensors, etc.) on the ceiling or walls to acquire the elderly person's position coordinates and movement status (such as stillness, walking, pre-fall signs, wandering, etc.) as three-dimensional vector data at intervals of 1 second or higher. The motion detection unit sends this sensor data as a time-series tensor (e.g., a 30×3 matrix for 30 seconds of movement history) to the central processing unit. The motion detection unit performs noise removal (e.g., moving average filtering, Butterworth filtering), outlier removal, and standardization (e.g., Z-score normalization) in the data preprocessing unit before inputting it into the AI ​​inference module. The AI ​​inference module utilizes convolutional neural networks (CNNs) or recurrent neural networks (RNNs) for feature extraction and anomaly detection. AI input examples include continuous vector sequences such as t=0s: [0.1, 0.2, 0.0], t=1s: [0.15, 0.18, 0.02], and outputs structured data such as anomaly score (0.0–1.0), anomaly type (e.g., fall, wandering, normal), and urgency level (urgent, requiring attention, normal). For example, output examples might include "anomaly score 0.92, type: fall, urgency: urgent" or "anomaly score 0.45, type: wandering, urgency: requiring attention." These outputs are compared with preset benchmark values ​​(e.g., anomaly score above 0.8) in the threshold determination unit. If the score exceeds the threshold, the data is forwarded to the notification or instruction unit. Therefore, this motion detection unit differs from traditional manual monitoring or simple threshold determination methods, achieving a comprehensive improvement in computer technologies such as high-dimensional sensor data AI analysis, multi-module collaboration, real-time anomaly detection, notification, and instruction. The technological benefits include improved anomaly detection accuracy (significantly reducing false positives and false negatives), faster response times (real-time notifications and instructions), improved data management efficiency (automatic recording and historical analysis), and optimized communication load (notifications only when necessary, distributed processing). Specific application areas include a variety of use cases such as care for elderly people living alone, home-based medical support, safety management of nursing facilities, remote health monitoring, support for people with disabilities, and child and pet care.

[0039] A fall detection unit, worn on the body of an elderly person, can immediately detect a fall. "Immediately" means within exponential seconds or minutes. The fall detection unit can detect falls using an accelerometer. This allows for rapid response by detecting falls immediately. Specifically, this fall detection unit uses an accelerometer or gyroscope sensor that can be worn on the waist or wrist of the elderly person to continuously acquire three-axis acceleration data (e.g., X: 0.98, Y: -0.12, Z: 0.05) at a high sampling rate such as 100Hz. The fall detection unit sends this sensor data as a time-series tensor (e.g., a 200×3 matrix for 2 seconds of history) to a central processing unit. The fall detection unit performs noise removal (e.g., low-pass filtering), outlier removal, and normalization in the data preprocessing unit, and then inputs it into a time-series AI model such as LSTM or GRU. The AI ​​input examples are continuous vector sequences such as t=0ms: [0.98, -0.12, 0.05], t=10ms: [1.02, -0.10, 0.03], etc., and the output is structured data such as fall score (0.0~1.0) and urgency level (urgent, need attention, normal). For example, output examples include "fall score 0.95, urgency level: urgent" or "fall score 0.45, urgency level: need attention". These outputs are compared with the preset benchmark values ​​(such as fall score above 0.8) in the threshold judgment unit. If the value exceeds the threshold, the data is forwarded to the notification unit or instruction unit. The fall detection unit uses the AI ​​model to analyze fall-specific patterns (such as rapid acceleration changes and posture changes) in high-dimensional space, realizing complex feature extraction that is different from simple rule judgment by humans. Therefore, compared with traditional manual operations or simple threshold judgment methods, it can achieve technical benefits such as improved fall detection accuracy (significantly reducing false positives and false negatives), faster response speed (real-time notifications and instructions), and improved data management efficiency (automatic recording and historical analysis). Specific application areas include care for elderly people living alone, safety management of nursing facilities, support for people with disabilities, and rehabilitation support.

[0040] The health monitoring department can measure the elderly person's body temperature, heart rate, and blood pressure in real time. "Real time" refers to updates occurring at the second or minute level. The health monitoring department can monitor the elderly person's health status through a medical monitoring system. This system can measure vital signs such as body temperature, heart rate, and blood pressure in real time and send the information to an AI. Therefore, by measuring vital signs in real time, the health status of the elderly person can be monitored at any time. Specifically, this health monitoring department wears medical sensors such as temperature sensors, heart rate sensors, and blood pressure monitors on the elderly person's body, continuously acquiring vital sign data (e.g., body temperature 36.7℃, heart rate 74 bpm, blood pressure 122 / 78 mmHg) at set intervals of 5 minutes or 1 minute. The health monitoring department sends this data as a continuous value array or time-series tensor to a central processing unit. The data preprocessing unit performs noise removal, outlier removal, and standardization, and then inputs it into a multilayer perceptron (MLP) or decision tree ensemble (such as random forest) AI model. Examples of AI inputs include continuous arrays of body temperature, heart rate, and blood pressure values ​​every 5 minutes (e.g., body temperature 36.5℃, heart rate 72 bpm, blood pressure 120 / 80 mmHg) and a 1-day vital sign history (288×3 matrix). AI outputs structured data such as anomaly score (0.0–1.0), anomaly type (e.g., fever, tachycardia, hypotension), and urgency level (urgent, requiring attention, normal). For example, output examples might include "anomaly score 0.88, type: fever, urgency: requiring attention" or "anomaly score 0.95, type: tachycardia, urgency: urgent." These outputs are compared with preset benchmark values ​​in the threshold determination unit; if the threshold is exceeded, the data is forwarded to the notification or instruction unit. The health monitoring unit uses AI's temporal anomaly detection algorithms (e.g., autoregressive models, LSTM-based prediction models) to detect sudden changes or deviations from normal patterns with high precision. Therefore, compared with traditional manual periodic health checks or simple threshold determination methods, this approach achieves technological benefits such as improved anomaly detection accuracy, faster response speed, and enhanced data management efficiency. Specific application areas include health monitoring for elderly people living alone, home-based medical support, chronic disease management, health management of nursing facilities, and telemedicine.

[0041] The analysis unit analyzes information from the motion detection unit, fall detection unit, and health monitoring unit. When an anomaly is detected, it notifies family members or caregivers. Anomalies, for example, refer to deviations from normal movement patterns or exceeding specific thresholds. The analysis unit can utilize AI to analyze motion data and detect anomalies. It can also use machine learning algorithms to analyze motion data and detect anomalies. This allows for rapid response by notifying family members or caregivers when an anomaly is detected. Specifically, this analysis unit integrates temporal motion data from the motion detection unit, acceleration and gyroscope data from the fall detection unit, and vital sign data from the health monitoring unit. After noise removal, outlier removal, and standardization in the data preprocessing unit, the data is input into the AI ​​inference module. The AI ​​inference module uses CNN or RNN in motion analysis, LSTM or GRU in fall detection, and MLP or decision tree ensemble in health status analysis, combining multiple AI architectures. Examples of AI inputs include 30 seconds of action history (30×3 matrix), 2 seconds of acceleration history (200×3 matrix), and 1 day of vital sign history (288×3 matrix). AI outputs structured data such as anomaly detection scores (0.0–1.0), anomaly types (e.g., falls, wandering, fever, tachycardia), and urgency levels (urgent, requiring attention, normal). For example, "anomaly score 0.92, type: fall, urgency: urgent" or "anomaly score 0.45, type: abnormal vital signs, urgency: requiring attention." These outputs are compared to preset benchmark values ​​in the threshold determination unit; if the value exceeds the threshold, the data is forwarded to the notification or instruction unit. The analysis unit databases past anomaly detection history and the response history of family members and caregivers for continuous online learning of the AI ​​model and personalized optimization of notification and instruction content. This not only automates manual operations but also achieves a comprehensive improvement in computer technologies, including high-dimensional analysis of sensor data, multi-module collaboration, real-time anomaly detection, automatic notification, and emergency instructions. The technological benefits include improved anomaly detection accuracy, faster response times, enhanced data management efficiency, and optimized communication load. Specific application areas include care for elderly people living alone, home-based medical support, safety management of nursing facilities, remote health monitoring, and support for people with disabilities.

[0042] The notification department can notify family members or caregivers based on information analyzed by AI. AI refers to technologies such as machine learning and deep learning. The notification department notifies family members or caregivers when an anomaly is detected. Thus, by using AI-analyzed information for notification, the accuracy of information delivery is achieved. Specifically, this notification department receives anomaly judgment results (such as anomaly score, anomaly type, and urgency) from the analysis department, fall detection department, and health monitoring department, compares them with preset benchmark values ​​(such as anomaly score above 0.8) in the threshold judgment department, and when the threshold is exceeded, it takes into account the family member or caregiver's device information (smartphone, tablet, PC, etc.) and geographical location information (GPS coordinates, address), and automatically selects the optimal notification method (push notification, SMS, email, voice call, etc.) to send the information. The notification content includes structured data such as anomaly type, occurrence time, urgency, and recommended coping methods. For example, notifications could include: "Fall detection: Urgency: Urgent, Time of occurrence: 12:34, Recommended response: Call an ambulance" or "Abnormal vital signs: Urgency: Requires attention, Time of occurrence: 15:20, Recommended response: Observe." The notification department can also analyze past notification history and the response history of family members and caregivers to optimize notification methods, content, and timing. Therefore, compared to traditional simple alarm notifications or manual telephone contact, the accuracy, speed, and personalization of information delivery are significantly improved. Technical benefits include reduced false alarms and missed alarms, faster response times, optimized communication load, and improved availability. Specific application areas include care for elderly people living alone, safety management of nursing facilities, support for people with disabilities, and remote health monitoring.

[0043] The instruction unit can provide emergency response instructions based on information analyzed by AI. For example, an emergency response might involve calling an ambulance or contacting family members. When a fall is detected, the instruction unit instructs the unit to call an ambulance or contact family members. Thus, by instructing emergency response based on AI-analyzed information, rapid response is achieved. Specifically, this instruction unit receives anomaly assessment results (such as anomaly score, anomaly type, and urgency) from the analysis unit, fall detection unit, and health monitoring unit. Based on the urgency level, it automatically instructs actions such as calling an ambulance, immediately contacting nearby family members, or notifying nursing service providers. The instructions include structured data such as anomaly type, occurrence time, recommended response method, and contact information, and are sent to the corresponding devices or systems. For example, "Fall detection: Call ambulance, contact family members, occurrence time: 12:34" or "Abnormal vital signs: After observation and family notification, occurrence time: 15:20," etc. The instruction unit can also analyze past response history and the reaction history of family members and nursing staff to optimize response methods, instruction content, and timing. Therefore, compared with traditional manual judgment or simple alarm notifications, the accuracy, speed, and automation of emergency response are significantly improved. Technological benefits include faster response times, fewer mishandling attempts, improved data management efficiency, and enhanced usability. Specific application areas include care for elderly people living alone, safety management of nursing facilities, support for people with disabilities, and remote health monitoring.

[0044] The notification department is equipped with an AI-powered function to periodically report the elderly person's condition to family members or caregivers. "Periodic" here refers to time intervals such as daily or weekly. The notification department can use AI to periodically report the elderly person's condition to family members or caregivers. This provides peace of mind to family members or caregivers through regular reporting. Specifically, this notification department automatically summarizes the elderly person's life and health data (such as a 30×3 matrix of daily movement history, a time series array of vital signs, presence or absence of fall events, and emotional inference results) collected from the analysis department, health monitoring department, movement detection department, and fall detection department according to a predetermined schedule (such as 8:00 AM daily, 9:00 AM every Monday), and inputs it into the AI ​​summary generation module. The AI ​​summary generation module can employ a multilayer perceptron (MLP) or temporal feature extraction model (LSTM, Transformer, etc.) to integrate anomaly detection history, health status changes, emotional trends, and coping history from the past 24 hours or 1 week into a multidimensional vector, generating a natural language summary or structured report (e.g., "No anomalies detected in the past 24 hours, vital signs stable, emotional state: inclined to relax"). Examples of AI inputs include 1 day of action data (1440×3 matrix), vital sign history (288×3 matrix), a list of abnormal events (e.g., 0 falls, 1 vital sign abnormality), and emotional inference scores (e.g., joy 0.2, sadness 0.1, anxiety 0.1, relaxation 0.6). AI outputs include summary text (e.g., "No anomalies detected today. Health status stable."), a list of anomaly occurrence times, recommended coping strategies, and other structured data. These outputs are automatically sent by the notification department scheduler based on the family member's or caregiver's device information (smartphone, tablet, PC, etc.) and the desired notification method (email, push notification, LINE, etc.). The notification department can also analyze past notification history and the response history of family members and caregivers to optimize notification content and timing. Technological benefits include improved accuracy, efficiency, and real-time performance of high-dimensional data summarization compared to traditional simple scheduled reports or manual phone calls, significantly enhancing the accuracy, efficiency, and reassurance of information delivery. Specific application areas include care for elderly people living alone, home-based medical support, family communication in nursing facilities, support for people with disabilities, and remote health monitoring.

[0045] The notification department is equipped with the function of enabling direct communication between family members or caregivers and the elderly in emergency situations. For example, an emergency situation could refer to a fall or the detection of abnormal vital signs. The notification department allows direct communication between family members or caregivers and the elderly in such situations, thus enabling rapid response. Specifically, upon receiving abnormal judgment results (such as a fall score of 0.95, urgency level: urgent, abnormality type: abnormal vital signs, etc.) from the analysis department, fall detection department, and health monitoring department, the notification department immediately activates the family member or caregiver's device (smartphone, tablet, PC, etc.) and the elderly person's home communication terminal (such as a desktop tablet, smart speaker, etc.) to automatically establish a two-way voice call, video call, text chat, or other communication session. AI can automatically generate an emergency description and recommended response (such as "A fall has been detected. Please check the elderly person's consciousness."), and provide a prompt through speech synthesis at the start of the call. Examples of AI inputs include anomaly detection data (anomaly score 0.92, type: fall, urgency: urgent), the elderly person's current location (e.g., living room), emotional presumption score (e.g., anxiety 0.7), and past response history. AI outputs structured data such as call initiation triggers, recommended response messages, and call log records. The notification department records the communication session establishment status and call content, and saves it to a subsequent response history database. Technical benefits include faster, more reliable, and more documented information delivery and communication in emergency situations compared to traditional simple alarm notifications or manual phone calls, thanks to the linkage between AI anomaly detection and automatic communication control. Specific application areas include care for elderly people living alone, emergency contact in nursing facilities, support for people with disabilities, and on-site emergency response in telemedicine.

[0046] The motion detection unit can infer the emotions of elderly individuals and adjust the sensitivity of motion detection based on the inferred emotions. For example, the motion detection unit infers the emotions of elderly individuals and adjusts the sensitivity of motion detection accordingly. Emotions include, for example, joy, sadness, and anger. When an elderly person is anxious, the motion detection unit increases the sensitivity of motion detection to detect abnormal movements early. When an elderly person is relaxed, the sensitivity of motion detection decreases to reduce false detections. When an elderly person is excited, the sensitivity of motion detection is appropriately adjusted to achieve accurate motion detection. Therefore, by adjusting the sensitivity of motion detection based on the emotions of the elderly person, false detections can be reduced and accurate detection can be achieved. Specifically, this motion detection unit integrates data acquired from image sensors, audio sensors, and vital sign sensors, and inputs it into an AI emotion inference module (such as a multimodal neural network, a Transformer-based emotion classifier, etc.). Examples of AI inputs include facial expression feature vectors (such as 68-point marker coordinates), spectral features of audio waveforms (such as MFCC vectors), heart rate variability data (such as a 5-minute RR interval array), and motion history over the past 24 hours (a 1440×3 matrix), etc. The AI ​​outputs structured data such as emotion labels (e.g., joy, sadness, anxiety, relaxation), emotion scores (0.0–1.0), and confidence indices. Based on these emotion inferences, the action detection department dynamically adjusts the threshold parameters of the action detection AI (e.g., CNN, RNN) or the sensitivity coefficient of the anomaly detection algorithm. For example, when the anxiety score is greater than or equal to 0.7, the anomaly detection threshold is lowered from 0.7 to 0.5; when the relaxation score is high, the threshold is raised to 0.8. This achieves optimized false positives reduction and earlier detection based on emotional state. Technical benefits include improved accuracy, reduced false alarms, and increased usability compared to traditional uniform threshold methods, allowing for flexible detection control based on individual states. Specific application areas include care for elderly people living alone, behavioral monitoring of patients with mental illness, safety management of nursing facilities, and support for people with disabilities.

[0047] The motion detection unit analyzes past movement patterns of elderly individuals to improve the accuracy of abnormal movement detection. Abnormal movements, for example, refer to deviations from typical movement patterns or exceeding specific thresholds. The motion detection unit collects past movement data from elderly individuals and learns abnormal movement patterns. It can detect movements that differ from the elderly person's typical movement patterns and identify them as abnormal. The unit can monitor changes in the elderly person's movement patterns in real time, detecting abnormal movements early. Thus, by analyzing past movement patterns, the accuracy of abnormal movement detection is improved. Specifically, this motion detection unit digitizes long-term accumulated historical movement data of the elderly (such as a 1440×3 matrix for one day, a 10080×3 matrix for one week, etc.) and inputs it into an AI anomaly detection model (such as an autoencoder, an LSTM-based prediction model, clustering algorithms, etc.). Examples of AI input include a 30-day historical movement tensor, movement status labels at various times (such as walking, standing still, pre-fall warning signs, etc.), and a list of abnormal event occurrence times. The AI ​​learns the distribution and temporal changes of typical patterns and compares them with newly acquired action data in real time, outputting a deviation score (e.g., 0.0–1.0) and an anomaly type label. For example, output examples might include "Deviation score 0.85, Type: Loitering, Urgency: Attention Required" or "Deviation score 0.95, Type: Fall, Urgency: Urgent." These outputs are compared with the threshold judgment unit's benchmark value, and when an anomaly is detected, they are forwarded to the notification or instruction unit. The action detection unit automatically updates the model parameters through the AI ​​model's online learning function to adapt to changes in individual action characteristics. Technical benefits include personalization, handling of temporal changes, reduced false positives, and improved detection accuracy compared to traditional fixed-rule methods. Specific application areas include care for elderly people living alone, rehabilitation support, behavioral monitoring of patients with chronic diseases, and safety management of nursing facilities.

[0048] The motion detection unit optimizes detection accuracy based on room environmental conditions during motion detection. For example, during motion detection, the unit considers room environmental conditions (temperature, humidity, lighting, etc.) to optimize detection accuracy. Environmental conditions refer to factors such as temperature, humidity, and lighting. When the room temperature is high, the motion detection unit adjusts the motion detection sensitivity to prevent false detections. When the room humidity is high, the motion detection sensitivity is adjusted to achieve accurate detection. When the room lighting is dim, the motion detection sensitivity is increased to detect abnormal movements earlier. Thus, considering room environmental conditions improves motion detection accuracy. Specifically, this motion detection unit inputs time-series data (such as temperature, humidity, and illuminance values ​​per minute) acquired by environmental sensors (temperature sensors, humidity sensors, illuminance sensors, etc.) as supplementary information into the motion detection AI module. Examples of AI input include environmental vectors such as t=0 minutes: [temperature 25.0℃, humidity 60%, illuminance 300lx], t=1 minutes: [temperature 25.2℃, humidity 61%, illuminance 280lx], and simultaneous motion data (such as labels for walking, stationary, etc.). AI dynamically adjusts anomaly detection thresholds and feature extraction parameters based on environmental conditions, such as suppressing false positives under high temperature and humidity. The output is structured data including anomaly score, anomaly type, and urgency, with environmental optimization reducing false positives and false negatives. These outputs are compared with benchmark values ​​from the threshold determination unit and, if necessary, forwarded to the notification or instruction unit. The technological benefits include improved detection accuracy, reduced false alarms, and increased usability under changing environmental conditions compared to traditional detection methods that do not consider the environment. Specific application areas include care for elderly people living alone, safety management of nursing facilities, remote health monitoring, and support for people with disabilities.

[0049] The motion detection unit can infer the emotions of elderly individuals and determine the priority of motion detection based on the inferred emotions. For example, the motion detection unit infers the emotions of elderly individuals and determines the priority of motion detection based on these inferred emotions. Emotions include, for example, joy, sadness, and anger. When an elderly person is anxious, the motion detection unit increases the priority of motion detection to detect abnormal movements early. When an elderly person is relaxed, the priority of motion detection decreases to reduce false detections. When an elderly person is excited, the priority of motion detection is adjusted appropriately to achieve accurate motion detection. Therefore, by determining the priority of motion detection based on the elderly person's emotions, important abnormalities can be detected early. Specifically, this motion detection unit integrates multimodal data acquired from image sensors, audio sensors, and vital sign sensors, and inputs it into an AI emotion inference module (such as a multimodal neural network or a Transformer-based emotion classifier). Examples of inputs to the AI ​​emotion inference module include facial expression feature vectors (68-point marker coordinates), audio waveform MFCC features, heart rate variability data (5-minute RR interval array), and motion history over the past 24 hours (1440×3 matrix), etc. The AI ​​outputs emotion labels (joy, sadness, anxiety, relaxation, etc.), emotion scores (0.0–1.0), and confidence indices. For example, "Emotion label: Anxiety, Score: 0.75" or "Emotion label: Relaxation, Score: 0.85." Based on the emotion inference results, the action detection unit dynamically controls the priority parameters and detection scheduling of the action detection AI (CNN, RNN, etc.) anomaly detection algorithms. For example, when the anxiety score is greater than or equal to 0.7, the anomaly detection process priority is set to the highest, and the detection frequency is increased to a 1-second interval. When the relaxation score is high, the priority is reduced, and the detection frequency is adjusted to a 5-second interval. In an excited state, to prevent false detections, the feature extraction parameters are optimized, and the detection process priority is set to medium. AI output examples include "Priority: High, Detection Interval: 1 second" or "Priority: Low, Detection Interval: 5 seconds." These priority controls serve as parameters for the action detection unit scheduler and the AI ​​inference module, optimizing resource allocation for the action detection process and the timing of data forwarding to the notification and instruction units in real time. The technical benefits include flexible priority control based on individual emotional states compared to traditional unified detection methods, enabling early detection of important anomalies, reducing false alarms, and optimizing system resources. Specific application areas include care for elderly people living alone, monitoring the behavior of patients with mental illness, safety management of nursing facilities, and support for people with disabilities.

[0050] The motion detection unit can adjust its detection range based on the furniture layout of a room during motion detection. For example, the motion detection unit considers the furniture layout of the room when adjusting its detection range. Furniture layout refers to, for example, the position and size of the furniture. The motion detection unit can optimize the motion detection range based on the room's furniture layout. When the furniture layout changes, the motion detection range is reset. The motion detection unit can also adjust the motion detection sensitivity based on the furniture layout. Thus, the motion detection range is optimized by considering the room's furniture layout. Specifically, to obtain furniture layout information within the room, this motion detection unit uses image sensors for spatial scanning or collects furniture layout data (such as structured data like the coordinates, dimensions, and height of each piece of furniture) through user input. The motion detection unit stores this furniture layout data as a 3D spatial map (e.g., room dimensions 6m × 4m × 2.5m, furniture A: coordinates [1.2, 0.8, 0.0], dimensions [1.0, 0.5, 0.8], etc.) in an internal database. The AI ​​inference module combines the furniture layout map with the motion sensor layout information to automatically calculate the detection range (field of view, blind spots, occluded areas, etc.) of each sensor. For example, when furniture movement or addition is detected, the AI ​​identifies new blind spots and resets the detection range. AI input examples include furniture layout vectors (coordinates and dimensions of each piece of furniture), sensor layout vectors (installation location and orientation of each sensor), and past motion detection history (such as a list of missed detection times). AI output includes structured data such as a map of the effective detection range of each sensor, a list of blind spots, and detection sensitivity parameters (e.g., Sensor 1: sensitivity 0.9, Sensor 2: sensitivity 0.7). For example, "Sensor 1: Detection range [0, 2m], blind spot: behind furniture A," and "Sensor 2: Sensitivity 0.8, detection range [2, 4m]." These outputs are reflected in the sensor control module of the motion detection unit, optimizing the detection range and sensitivity in real time. When the furniture layout changes, it can be reset via the user interface or automatically scanned and relearned. Technical benefits include greater flexibility in responding to changes in furniture layout compared to traditional fixed detection range methods, reducing blind spots and false detections, and improving detection accuracy and usability. Specific application areas include care for elderly people living alone, safety management of nursing facilities, support for people with disabilities, and automatic optimization of smart home environments.

[0051] The motion detection unit can filter pet movements during motion detection to prevent false detections. For example, the motion detection unit filters pet movements during motion detection to prevent false detections. Pet movements refer to, for example, movement patterns and size. The motion detection unit can detect pet movements and exclude them from the motion detection list. The motion detection unit can distinguish between the movements of pets and elderly people to prevent false detections. By filtering pet movements, the motion detection unit improves motion detection accuracy. Specifically, this motion detection unit inputs motion data acquired by image sensors, infrared sensors, and ultrasonic sensors into the AI ​​motion recognition module to distinguish between the movements of pets and elderly people. Examples of AI inputs include object detection feature vectors from image sensors (such as bounding box coordinates, object size, and movement speed), infrared sensor response patterns, and past motion history (such as pet movement path labels, body height and length information). The AI ​​uses convolutional neural networks (CNN) or object detection algorithms (YOLO, SSD, etc.) to classify the type of motion object (elderly person, pet, others) and distinguish between the movements of pets and elderly people. The AI ​​output consists of structured data such as action type labels (e.g., elderly, dog, cat), confidence scores (0.0–1.0), and exclusion flags. For example, "Action type: dog, confidence: 0.92, exclusion: True" or "Action type: elderly, confidence: 0.98, exclusion: False". The action detection department excludes data identified as pets from the anomaly detection AI input to prevent false detections. Furthermore, the AI ​​continuously learns pet movement patterns (e.g., small movements, low body height, specific movement speed ranges) to improve recognition accuracy. Therefore, compared to traditional simple action detection methods, false detections caused by pets are significantly reduced, improving anomaly detection accuracy and usability. Specific application areas include care for elderly people living alone, security management of pet-friendly residential areas, and animal-assisted care environments in nursing facilities.

[0052] The analysis unit can infer the emotions of elderly individuals and adjust the analysis algorithm based on the inferred emotions. For example, the analysis unit infers the emotions of elderly individuals and adjusts the analysis algorithm accordingly. Emotions include, for example, joy, sadness, and anger. When an elderly person feels uneasy, the analysis unit increases the sensitivity of the analysis algorithm to detect anomalies early. When an elderly person is relaxed, the sensitivity of the analysis algorithm is decreased to reduce false positives. When an elderly person is excited, the sensitivity of the analysis algorithm is appropriately adjusted to achieve accurate analysis. Thus, by adjusting the analysis algorithm based on the elderly person's emotions, the analysis accuracy is improved. Specifically, this analysis unit inputs multimodal data obtained from the motion detection unit, health monitoring unit, etc., into the AI ​​emotion inference module to obtain emotion inference results (such as emotion labels and emotion scores). Examples of AI inputs include facial image feature vectors, audio feature quantities, heart rate variability data, and past action history. AI outputs structured data such as emotion labels (e.g., uneasy, relaxed, excited), emotion scores (0.0–1.0), and confidence indices. Based on the emotion inference results, the analysis unit dynamically adjusts the sensitivity parameters or thresholds of the anomaly detection AI (CNN, RNN, LSTM, etc.). For example, when the anxiety score is greater than or equal to 0.7, the anomaly detection threshold is lowered from 0.7 to 0.5; when the relaxation score is high, the threshold is raised to 0.8. In an excited state, feature extraction parameters are optimized to strengthen filtering and prevent false alarms. AI output examples include "Anomaly score 0.92, Type: Fall, Urgency: Urgent" or "Anomaly score 0.45, Type: Abnormal vital signs, Urgency: Requires attention." These outputs are compared with the threshold determination unit's benchmark value and, if necessary, forwarded to the notification or instruction unit. The technical benefits include improved accuracy, reduced false alarms, and increased usability compared to traditional uniform threshold methods, allowing for flexible analysis and control based on individual states. Specific application areas include care for elderly people living alone, behavioral monitoring of patients with mental illness, safety management of nursing facilities, and support for people with disabilities.

[0053] The analysis unit can improve analysis accuracy by referring to past anomaly detection data during analysis. For example, the analysis unit can refer to past anomaly detection data during analysis to improve accuracy. Anomaly detection data refers to past anomaly cases, databases, etc. The analysis unit can collect past anomaly detection data to optimize the analysis algorithm. The analysis unit can learn anomaly patterns by referring to past anomaly detection data. The analysis unit can improve the accuracy of the analysis algorithm based on past anomaly detection data. Therefore, by referring to past anomaly detection data, analysis accuracy is improved. Specifically, this analysis unit databases historical anomaly detection data (such as anomaly occurrence time, anomaly type, detection score, response result, etc.) collected by the action detection unit, fall detection unit, health monitoring unit, etc., and inputs it into the AI ​​anomaly detection model (such as an autoencoder, LSTM-based prediction model, clustering algorithm, etc.). Examples of AI input include a historical anomaly detection tensor of the past 30 days, a list of anomaly event occurrence times, and response result labels (such as emergency response, observed, etc.). The AI ​​learns the distribution and temporal changes of anomaly patterns and compares them with newly acquired data in real time, outputting a deviation score and anomaly type label. For example, outputs such as "Deviation score 0.85, Type: Loitering, Urgency: Attention Required" or "Deviation score 0.95, Type: Fall, Urgency: Urgent" are provided. These outputs are compared with the benchmark value of the threshold judgment department, and when an anomaly is detected, they are forwarded to the notification or instruction department. The analysis department automatically updates the model parameters through the online learning function of the AI ​​model to adapt to individual changes in anomaly trends. Technical benefits include personalized and temporal response, reduced false positives, and improved detection accuracy compared to traditional fixed-rule methods. Specific application areas include care for elderly people living alone, rehabilitation support, behavioral monitoring of patients with chronic diseases, and safety management of nursing facilities.

[0054] The parsing unit can adjust the level of detail in its analysis based on the frequency of anomaly detection. For example, the parsing unit adjusts the level of detail based on the frequency of anomaly detection. The frequency of anomaly detection refers to, for example, the number of detections or the time interval between detections. When the anomaly detection frequency is high, the parsing unit increases the level of detail to pinpoint the cause of the anomaly. When the anomaly detection frequency is low, the parsing unit decreases the level of detail to conserve resources. The parsing unit can dynamically adjust the level of detail based on the anomaly detection frequency. Thus, resource optimization is achieved by adjusting the level of detail based on the anomaly detection frequency. Specifically, this parsing unit continuously calculates the frequency of anomaly detection events received by the motion detection unit, fall detection unit, and health monitoring unit (e.g., the number of anomaly detections per hour, the interval between anomaly occurrences in the last 24 hours), and inputs this data into the resolution control module. When the anomaly detection frequency is high, the resolution control module automatically selects a detailed parsing mode (e.g., high-dimensional feature extraction, anomaly cause estimation, time-series pattern parsing), and the AI ​​inference module uses deeper neural networks (e.g., deep CNN, LSTM, Transformer) or a combination of multiple algorithms (e.g., anomaly clustering + autoencoder) for parsing. Examples of AI inputs include a list of anomaly occurrence times (e.g., t1=10:05, t2=10:12, t3=10:18), anomaly type history (e.g., 2 falls, 1 abnormal vital sign), and time-series tensors of recent actions, vital signs, and environmental data (e.g., a 3600×3 matrix of 60-minute action history, a 288×3 matrix of vital signs). AI outputs structured data such as anomaly cause inference labels (e.g., cause of fall: unsteady gait, cause of abnormal vital signs: fever trend), anomaly pattern classification (e.g., periodicity, suddenness, etc.), and detailed anomaly score distribution (e.g., anomaly score of 0.0–1.0 at each time point). On the other hand, when the anomaly detection frequency is low, a simplified parsing mode (e.g., extracting only key features, simple threshold judgment, lightweight MLP, etc.) is selected to save computational resources and power consumption. These level of detail controls serve as parameters for the parsing unit scheduler and AI inference module, optimizing the overall system load distribution and real-time performance. The technological benefits include, compared to traditional uniform detail resolution methods, flexible resource allocation based on anomaly occurrences, optimized resolution accuracy, and improved computational efficiency and system stability. Specific application areas include care for elderly people living alone, safety management of nursing facilities, remote health monitoring, support for people with disabilities, and home-based medical support.

[0055] The analysis unit can infer the emotions of elderly individuals and adjust the display method of the analysis results based on the inferred emotions. For example, the analysis unit can infer the emotions of elderly individuals and adjust the display method of the analysis results based on the inferred emotions. Emotions include, for example, joy, sadness, and anger. For instance, when an elderly person is feeling uneasy, the analysis unit displays the analysis results in a concise manner to provide reassurance; when the elderly person is relaxed, it displays the analysis results in a detailed manner to deepen understanding; and when the elderly person is excited, it displays the analysis results in a visually easy-to-understand manner. By adjusting the display method of the analysis results according to the elderly person's emotions, an easy-to-understand display can be achieved. Specifically, this analysis unit inputs multimodal data obtained from image sensors, voice sensors, vital sign sensors, etc., into the AI ​​emotion inference module (such as multimodal neural networks, Transformer-based emotion classifiers, etc.) to infer emotion labels (such as uneasy, relaxed, excited, etc.) and emotion scores (0.0 to 1.0). Examples of AI inputs include facial expression feature vectors (68-point marker coordinates) from facial images, MFCC features from speech waveforms, heart rate variability data (5-minute RR interval sequence), and activity history from the past 24 hours (1440×3 matrix). AI outputs structured data such as emotion labels (e.g., anxiety, relaxation, excitement), emotion scores, and confidence indices. Based on the emotion inference results, this analysis unit automatically selects the display mode (e.g., concise display, detailed display, visually enhanced display) for the analysis result display module. For example, when the anxiety score is higher than 0.7, the anomaly detection results are displayed with concise information such as "No anomaly" or "Safe" and large icons to provide reassurance; when the relaxation score is high, a detailed time-series graph of the anomaly detection score, a detailed list of anomaly types, and past response history are displayed; in an excited state, visual enhancements such as color differentiation and animation are used to easily draw attention or explain the situation. These display controls are also reflected in the layout of the user interface module and the output content of the notification section. The technical effect is that, compared to the previous uniform display method, information can be flexibly provided according to individual emotional states, improving comprehension, reassurance, preventing misunderstandings, and enhancing usability. Specific application areas include care for elderly people living alone, provision of information on elderly care facilities, support for people with disabilities, and remote health monitoring.

[0056] The parsing unit can apply different parsing algorithms based on the type of anomaly detection during parsing. For example, the parsing unit applies different parsing algorithms depending on the type of anomaly detection. Types of anomaly detection include, for example, abnormal movement and abnormal vital signs. For instance, in fall detection, a fall-specific parsing algorithm is applied; in health anomaly detection, a health-specific parsing algorithm is applied; and in movement anomaly detection, a movement-specific parsing algorithm is applied. By applying the optimal parsing algorithm based on the type of anomaly detection, parsing accuracy can be improved. Specifically, this parsing unit determines the type of anomaly detection event (such as fall, abnormal vital signs, abnormal movement, etc.) and selects a dedicated AI parsing module for each type. For example, fall detection uses a temporal deep learning model such as LSTM or GRU, inputting triaxial acceleration data (e.g., 100Hz sampling, 2-second 200×3 matrix), and outputting a fall score and urgency level. Health anomaly detection employs a multilayer perceptron (MLP) or decision tree ensemble (such as a random forest), inputting a continuous sequence of vital signs (e.g., a 288×3 matrix of body temperature, heart rate, and blood pressure), and outputting anomaly scores and anomaly types. Motion anomaly detection uses a convolutional neural network (CNN) or recurrent neural network (RNN), inputting a motion history tensor (e.g., a 30×3 matrix for 30 seconds), and outputting anomaly severity and anomaly type. Examples of AI outputs include "Fall score 0.95, urgency: urgent," "Anomaly score 0.88, type: fever, urgency: need attention," and "Deviance score 0.85, type: loitering, urgency: need attention," etc. These outputs are compared with baseline values ​​in the threshold determination unit and forwarded to the notification or instruction unit when necessary. The analysis unit optimizes feature extraction methods and judgment criteria for each anomaly type and significantly improves analysis accuracy and response speed through multi-AI model collaboration. The technical benefits are that, compared to previous single-algorithm approaches, anomaly type optimization reduces false detections, improves detection accuracy, and enhances system flexibility. Specific application areas include care for elderly people living alone, safety management of elderly care facilities, remote health monitoring, support for people with disabilities, and home-based medical support.

[0057] The parsing unit can determine the parsing priority based on the time period of anomaly detection. For example, the parsing unit determines the parsing priority based on the time period of anomaly detection, such as daytime or nighttime. For instance, if anomalies are detected at night, the parsing unit increases the parsing priority for rapid response; if anomalies are detected during the day, the parsing priority decreases to conserve resources. The parsing unit can also dynamically adjust the parsing priority based on the time period of anomaly detection. By determining the parsing priority based on the time period of anomaly detection, rapid response can be achieved. Specifically, this parsing unit records the occurrence time of anomaly detection events (such as 24-hour clock time, day of the week information, etc.) as a timestamp and inputs it into the priority control module. When anomalies are detected at night (e.g., 10 PM to 6 AM) or during specific time periods such as holidays, the priority control module sets the parsing process priority to the highest, immediately allocates resources to the AI ​​inference module, and transmits the data to the notification and instruction units in real time. During daytime or normal times, the priority is decreased, and batch processing or simplified parsing modes are selected to optimize system load and energy consumption. Examples of AI input include a list of anomaly occurrence times (e.g., t1=23:15, t2=02:30), anomaly types (e.g., falls, abnormal vital signs), and past response history. AI output consists of structured data such as priority labels (high, medium, low), recommended response times (immediate, normal, delayed, etc.), and parsed scheduling information. This priority control is reflected in the timing of actions taken by the scheduler in the parsing department and the notification and instruction departments, enabling immediate response to urgent anomalies at night and resource conservation for minor anomalies during the day. The technical benefits are that, compared to previous uniform priority methods, flexible priority control over time periods achieves rapid response, resource optimization, and improved system stability. Specific application areas include care for elderly people living alone, nighttime security management of elderly care facilities, remote health monitoring, and support for people with disabilities.

[0058] The notification unit can infer the emotions of elderly individuals and adjust the notification content accordingly. For example, it can infer emotions such as joy, sadness, and anger. The notification unit can send reassuring messages when an elderly person is anxious, detailed messages when they are relaxed, and concise messages when they are excited. By adjusting the notification content based on the elderly person's emotions, appropriate information delivery can be achieved. Specifically, this notification unit inputs multimodal data acquired from image sensors, voice sensors, and vital sign sensors into the AI ​​emotion inference module to infer emotion tags (such as anxiety, relaxation, excitement, etc.) and emotion scores (0.0–1.0). Examples of AI inputs include facial image feature vectors, voice feature quantities, heart rate variability data, and past action history. The AI ​​output is structured data such as emotion tags, emotion scores, and confidence indices. Based on the emotion inference results, this notification unit automatically selects the template and expression method for the notification content generation module. For example, when the anxiety score is above 0.7, a reassuring message such as "The current situation is safe, please rest assured" is generated; when the relaxation score is high, a detailed notification is generated, including the anomaly detection score, health status details, and recommended coping strategies; and when the alertness score is high, a concise and clear expression is used (e.g., "A fall has been detected, please be careful"). The notification department also analyzes notification history and the reaction history of family members and caregivers to achieve personalized optimization of notification content. The technical benefits are that, compared to previous uniform notification methods, information can be flexibly conveyed according to individual emotional states, improving reassurance, comprehension, preventing misunderstandings, and enhancing usability. Specific application areas include care for elderly people living alone, provision of information on elderly care facilities, support for people with disabilities, and remote health monitoring.

[0059] The notification department can optimize its notification methods by referencing the past reaction records of family members or caregivers. For example, the department can analyze these records to select the optimal notification method; prioritize notification methods that family members or caregivers have responded to quickly in the past; and adjust the content and timing of notifications based on the reaction records. Specifically, the department builds a notification history database for each family member or caregiver, recording the reaction content (e.g., notification reception time, app launch, response button click, phone call, on-site visit), response time, notification method (push notification, SMS, email, voice call), and notification content (abnormality type, urgency, recommended response, etc.) for each notification event in chronological order. This historical data is then input into an AI notification optimization module (e.g., gradient boosting decision tree, MLP, temporal clustering) to learn the reaction tendencies of family members or caregivers and the optimal notification method, timing, and content. Examples of AI inputs include notification event history over the past 30 days (e.g., a 288×5 matrix of notification type, method, response time, and response outcome), family members' device usage tendencies (e.g., smartphone usage rate 0.9, PC usage rate 0.1), and past emergency response history (e.g., 10 emergency notifications, average response time 3 minutes). AI outputs structured data such as notification method recommendation tags (e.g., push notification, SMS, voice call), recommended notification timing (e.g., immediate, 5 minutes later, avoid at night), and notification content templates (e.g., detailed notification, simple notification). For example, "Family Member A: Push notification, immediate, detailed content," "Nurse B: SMS, 5 minutes later, simple content," etc. Based on the AI ​​output, the notification department controls the notification scheduler and content generation module to automatically select the optimal notification method, timing, and content for each family member or caregiver and send it. It also continuously collects post-notification response data and improves optimization accuracy through online learning of the AI ​​model. Therefore, compared to previous unified notification methods or manual selection, notifications can be optimized based on individual response preferences, resulting in improved response speed, reduced false notifications and missed notifications, improved availability, and optimized communication load. Specific application areas include care for elderly people living alone, family communication in elderly care facilities, support for people with disabilities, and remote health monitoring.

[0060] The notification department can adjust the urgency of a notification based on the type of anomaly. For example, it can adjust the urgency based on the type of anomaly, such as abnormal movement or abnormal vital signs. For instance, it might send a high-urgency notification upon detecting a fall; adjust the notification content based on urgency upon detecting a health abnormality; and adjust the timing of the notification based on urgency upon detecting an abnormal movement. By adjusting the urgency based on the type of anomaly, rapid response can be achieved. Specifically, the notification department inputs anomaly judgment data (such as anomaly type labels, anomaly scores, and urgency judgment values) received from the parsing department, fall detection department, and health monitoring department into the notification judgment module, and applies urgency judgment rules or AI classification models (such as decision trees, MLP, rule-based threshold judgment, etc.) to each anomaly type. Examples of AI input include structured data such as "Anomaly type: fall, anomaly score: 0.95", "Anomaly type: abnormal vital signs, anomaly score: 0.88", and "Anomaly type: abnormal movement, anomaly score: 0.75". The AI ​​output includes urgency labels (e.g., urgent, require attention, normal), recommended notification timing (e.g., immediate, after 5 minutes, scheduled notification), and notification content templates (e.g., detailed notification, simple notification). For example, "Fall: Urgent, Immediate Notification," "Abnormal Vital Signs: Require Attention, Detailed Notification," and "Abnormal Movement: Normal, Scheduled Notification." Based on the AI ​​output, the notification department controls the notification scheduler and content generation module to automatically select the optimal urgency, notification timing, and content for each type of anomaly and send it. It also analyzes past response history for each anomaly type and the reaction history of family members and caregivers to continuously optimize the urgency determination rules and AI model parameters. Therefore, compared to previous uniform urgency notification methods, this anomaly type optimization achieves improved response speed, reduced false alarms and missed alarms, improved availability, and increased system resource efficiency. Specific application areas include care for elderly people living alone, safety management of elderly care facilities, support for people with disabilities, and remote health monitoring.

[0061] The notification unit can infer the emotions of elderly individuals and adjust the timing of notifications accordingly. For example, the unit can infer the emotions of elderly individuals and adjust the timing of notifications based on these inferred emotions, such as joy, sadness, and anger. The unit can, for instance, promptly notify when an elderly person is anxious, delay notification when they are relaxed, and notify at an appropriate time when they are excited. By adjusting the timing of notifications based on the elderly person's emotions, notifications can be delivered at the appropriate time. Specifically, this notification unit inputs multimodal data acquired from image sensors, voice sensors, and vital sign sensors into an AI emotion inference module (such as a multimodal neural network or a Transformer-based emotion classifier) ​​to infer emotion labels (such as anxiety, relaxation, excitement, etc.) and emotion scores (0.0–1.0). Examples of AI inputs include facial expression feature vectors (68-point marker coordinates) from facial images, MFCC features from speech waveforms, heart rate variability data (5-minute RR interval sequences), and activity history from the past 24 hours (a 1440×3 matrix). The AI ​​output consists of structured data such as emotion labels, emotion scores, and confidence indices. This notification system dynamically controls the timing parameters of the notification scheduler based on emotion inference results. For example, when the anxiety score is higher than 0.7, an anomaly is detected and notification is sent immediately; when the relaxation score is high, the notification timing is delayed by 5-10 minutes; and when the person is in an excited state, notification is sent after the situation has stabilized to prevent misunderstandings and confusion. AI output examples include "Notification Timing: Immediate," "Notification Timing: 5 minutes later," and "Notification Timing: After the situation has stabilized." These timing controls are reflected in the notification system's scheduler and content generation module, achieving real-time optimization of the notification process. The technical benefits are that, compared to previous uniform notification timing methods, notification timing can be flexibly controlled based on individual emotional states, improving reassurance, reducing false alarms, and enhancing usability. Specific application areas include care for elderly people living alone, behavioral monitoring of patients with mental illness, information provision for elderly care facilities, and support for people with disabilities.

[0062] The notification department can select the optimal notification method by considering the geographical location information of family members or caregivers when issuing notifications. For example, the notification department considers the geographical location information of family members or caregivers when issuing notifications. Geographical location information includes, for example, GPS information and addresses. For instance, when family members or caregivers are nearby, the notification department sends a notification encouraging direct visit; when they are far away, it sends a notification via telephone or text message. The notification department selects the optimal notification method based on the geographical location information of family members or caregivers. By selecting the optimal notification method based on the geographical location information of family members or caregivers, rapid response can be achieved. Specifically, this notification department establishes an up-to-date geographical location information database (such as GPS coordinates, address, movement history, etc.) for each family member or caregiver, and obtains this information in real time through the location information acquisition module when a notification event occurs. This notification department inputs the current location of the recipient, the location of the anomaly (e.g., room information of the elderly person's residence), and past response history (e.g., number of on-site visits, number of remote responses, etc.) into the AI ​​notification method selection module (e.g., decision tree, MLP, rule-based judgment, etc.). Examples of AI input include "Family member A: GPS coordinates [35.6, 139.7], Elderly person's residence: GPS coordinates [35.6, 139.8], Distance: 1.2km" and "Caregiver B: Address: Tokyo, Distance: 10km". The AI ​​output consists of structured data such as recommended notification methods (e.g., direct access promotion, telephone, SMS, email), notification content templates (e.g., on-site visit requests, remote response requests), and notification priorities (high, medium, low). For example, "Family member A: Recommended direct access, immediate notification" and "Caregiver B: Telephone notification, detailed content". Based on the AI ​​output, the notification department controls the notification scheduler and content generation module to automatically select the optimal notification method, content, and timing for sending. It also continuously collects post-notification response history and improves optimization accuracy through online learning of the AI ​​model. Therefore, compared to previous unified notification methods, optimizing notifications using geographic location information achieves technical effects such as improved response speed, reduced false notifications and missed notifications, improved availability, and optimized communication load. Specific application areas include care for elderly people living alone, family communication in elderly care facilities, support for people with disabilities, and remote health monitoring.

[0063] The notification department can adjust the notification format based on the device information of family members or caregivers. For example, device information could refer to smartphones, tablets, etc. For instance, when a family member or caregiver uses a smartphone, a push notification might be sent; when using a computer, an email notification might be sent. The department selects the optimal notification format based on the family member's or caregiver's device information. By adjusting the notification format based on this information, appropriate information delivery can be achieved. Specifically, the notification department builds a device usage history database for each family member or caregiver, recording the device type (e.g., smartphone, tablet, PC, feature phone), operating system type, application usage, and notification receiving settings for each notification event. The department inputs the latest device information, past notification receiving history, and response history of the recipient into the AI ​​notification format selection module (e.g., decision tree, MLP, rule-based judgment). Examples of AI input include "Family Member A: Smartphone usage rate 0.95%, PC usage rate 0.05%" and "Caregiver B: Tablet usage rate 0.8%, notification application installed," etc. The AI ​​outputs structured data such as recommended notification formats (e.g., push notifications, SMS, emails, voice calls), notification content templates (e.g., detailed notifications, simple notifications), and notification priorities (high, medium, low). For example, "Family member A: Push notification, detailed content," "Caregiver B: Email notification, simple content," etc. Based on the AI ​​output, this notification department controls the notification scheduler and content generation module to automatically select the optimal notification format, content, and timing before sending it. It also continuously collects the reception and response history after notifications, using the AI ​​model to learn online and improve optimization accuracy. Therefore, compared to previous methods of uniformly formatting notifications, optimizing notifications using device information achieves improved accuracy, speed, and availability of information delivery, as well as optimized communication load. Specific application areas include care for elderly people living alone, family communication in elderly care facilities, support for people with disabilities, and remote health monitoring.

[0064] The fall detection unit can infer the emotions of elderly individuals and adjust the sensitivity of fall detection based on the inferred emotions. For example, the fall detection unit can infer the emotions of elderly individuals and adjust the sensitivity of fall detection accordingly. Emotions include, for example, joy, sadness, and anger. For instance, when an elderly person is anxious, the fall detection unit increases the sensitivity of fall detection to detect abnormalities early; when an elderly person is relaxed, the sensitivity decreases to reduce false positives; and when an elderly person is excited, the sensitivity is appropriately adjusted to achieve accurate detection. By adjusting the sensitivity of fall detection based on the elderly person's emotions, false positives can be reduced and accurate detection can be achieved. Specifically, this fall detection unit inputs data obtained from image sensors, voice sensors, vital sign sensors, etc., into the AI ​​emotion inference module to obtain emotion inference results (such as emotion labels and emotion scores). Examples of AI inputs include facial image feature vectors, voice feature quantities, heart rate variability data, and past movement history. The AI ​​output is structured data such as emotion labels (e.g., anxiety, relaxation, excitement), emotion scores (0.0–1.0), and confidence indices. The fall detection unit dynamically adjusts the sensitivity parameters and thresholds of the fall detection AI (LSTM, GRU, etc.) based on emotion inference results. For example, when the anxiety score is higher than 0.7, the fall detection threshold is lowered from 0.7 to 0.5; when the relaxation score is high, the threshold is raised to 0.8. During an excited state, feature extraction parameters are optimized, and filtering is strengthened to prevent false detections. Examples of AI outputs include "Fall score 0.95, urgency: urgent" or "Fall score 0.45, urgency: need attention," etc. These outputs are compared with a benchmark value in the threshold determination unit and forwarded to the notification or instruction unit when necessary. The technical benefits are that, compared to the previous uniform threshold method, it allows for flexible detection based on individual states, improving accuracy, reducing false alarms, and enhancing usability. Specific application areas include care for elderly people living alone, behavioral monitoring of patients with mental illness, safety management of elderly care facilities, and support for people with disabilities.

[0065] The fall detection department can analyze the past fall history of elderly individuals to improve the accuracy of fall detection. For example, the fall detection department can analyze the past fall history of elderly individuals to improve the accuracy of fall detection. Fall history includes past fall cases, databases, etc. The fall detection department collects past fall data of elderly individuals to learn fall patterns; based on past fall history, it identifies high-risk fall situations; and analyzes the fall history of elderly individuals to optimize the fall detection algorithm. By analyzing past fall history, the accuracy of fall detection can be improved. Specifically, this fall detection department stores detailed data of past fall events (such as the time of occurrence, location of occurrence, accelerometer values, changes in vital signs, and the history of actions before the fall) in a database in the form of time-series tensors or structured event lists. This fall detection department inputs this historical data into AI anomaly detection models (such as LSTM-based time-series prediction models, autoencoders, clustering algorithms, etc.) to learn characteristic patterns and risk factors before falls (such as decreased walking speed, postural instability, and sudden changes in vital signs). Examples of AI inputs include historical tensors of fall events over the past 30 days (e.g., 200×3 acceleration matrices per event, vital sign history, environmental condition vectors), a list of fall occurrence times, and action state labels before and after a fall (e.g., walking, stationary, pre-fall signs, etc.). The AI ​​outputs structured data such as a fall risk score (0.0–1.0), risk factor labels (e.g., unsteady gait, environmental factors, physical discomfort, etc.), and fall pattern classifications (e.g., sudden, gradual, etc.). For example, "Fall risk score 0.85, factor: unsteady gait" or "Fall risk score 0.92, factor: physical discomfort." Based on these AI outputs, our fall detection department personalizes and adjusts the threshold parameters and feature extraction methods of the real-time fall detection AI (LSTM, GRU, etc.) to improve sensitivity when detecting actions or situations similar to past fall patterns, enabling early detection. Furthermore, the fall history database has an online learning function, reflecting daily new fall events and near misses, automatically updating AI model parameters to adapt to changes in individual risk tolerance. Therefore, compared with previous methods based on fixed rules or simple thresholds, this approach achieves personalization, adaptability to changing timeframes, reduced false positives, and improved detection accuracy. Technical benefits include improved fall detection accuracy (significantly reducing false positives and false negatives), faster response times (real-time notifications and instructions), improved data management efficiency (automatic recording and historical analysis), and enhanced system flexibility. Specific application areas include care for elderly people living alone, safety management of elderly care facilities, rehabilitation support, fall risk management for patients with chronic diseases, and support for people with disabilities.

[0066] The fall detection unit can optimize detection accuracy based on the elderly person's physical condition or clothing during fall detection. For example, the fall detection unit considers the elderly person's physical condition or clothing to optimize detection accuracy. Physical condition includes factors such as fatigue level and medical history. Clothing includes factors such as shoe type and clothing thickness. For example, the fall detection unit increases the sensitivity of fall detection when the elderly person is in poor physical condition; adjusts the sensitivity when clothing restricts movement; and optimizes the fall detection algorithm based on physical condition or clothing. By considering the elderly person's physical condition or clothing, the accuracy of fall detection can be improved. Specifically, this fall detection unit obtains physical condition data (such as subjective fatigue score, history of abnormal vital signs, past medical history, medication status, etc.) and clothing data (such as shoe type labels, clothing thickness / mobility convenience indicators, wearable sensor information, etc.) from the health monitoring unit or user interface, collecting them in structured data form. This fall detection unit inputs this physical condition and clothing data into an AI fall risk prediction module (such as multilayer perceptron MLP, decision tree ensemble, rule-based judgment, etc.) to calculate a fall risk score and detection sensitivity adjustment parameters. Examples of AI inputs include vital sign history (288×3 matrix) over the past 24 hours, fatigue score (0.0–1.0), shoe type (e.g., sneakers, slippers, sandals), clothing thickness (mm), and ease-of-use index (0.0–1.0). The AI ​​output consists of structured data such as a fall risk score (0.0–1.0), sensitivity adjustment values ​​(e.g., +0.2, -0.1), and recommended detection algorithms (e.g., high-sensitivity mode, normal mode). For example, "fall risk 0.85, sensitivity +0.2" or "fall risk 0.45, sensitivity -0.1." Based on the AI ​​output, this fall detection department dynamically adjusts the thresholds and feature extraction parameters of the fall detection AI (LSTM, GRU, etc.). Sensitivity is increased to achieve early detection when physical condition is poor or clothing is inconvenient; conversely, false detections are suppressed when physical condition is good and clothing is convenient. Furthermore, physical condition and clothing data are stored as a daily historical database, which is then used by the AI ​​model to learn online and adapt to changes in individual preferences. Therefore, compared to previous methods using uniform thresholds or subjective human judgment, flexible detection and control that considers physical condition and clothing improves accuracy, reduces false alarms, and enhances usability. Technical benefits include improved fall detection accuracy, reduced false and false detections, increased system flexibility, and personalized optimization for users. Specific application areas include care for elderly people living alone, safety management of elderly care facilities, rehabilitation support, fall risk management for patients with chronic diseases, and support for people with disabilities.

[0067] The fall detection unit can infer the emotions of elderly individuals and determine the priority of fall detection based on the inferred emotions. For example, the fall detection unit can infer the emotions of elderly individuals and determine the priority of fall detection based on the inferred emotions. Emotions include, for example, joy, sadness, and anger. The fall detection unit, for example, increases the priority of fall detection when the elderly person is anxious; decreases the priority when the elderly person is relaxed; and appropriately adjusts the priority when the elderly person is excited. By determining the priority of fall detection based on the elderly person's emotions, important abnormalities can be detected early. Specifically, this fall detection unit inputs multimodal data (such as facial image feature vectors, MFCC features of speech waveforms, heart rate variability data, and movement history over the past 24 hours) obtained from image sensors, voice sensors, and vital sign sensors into an AI emotion inference module (such as a multimodal neural network or a Transformer-based emotion classifier) ​​to infer emotion labels (such as anxiety, relaxation, and excitement) and emotion scores (0.0–1.0). Examples of AI inputs include 68-point marker coordinates of facial images, speech features, heart rate variability data (5-minute RR interval sequence), and movement history over the past 24 hours (1440×3 matrix). AI outputs structured data such as emotion labels, emotion scores, and confidence indices. Examples include "Emotion Label: Anxiety, Score: 0.75" and "Emotion Label: Relaxation, Score: 0.85." Based on emotion inference results, this fall detection unit dynamically controls the priority parameters and detection scheduling of the fall detection AI (LSTM, GRU, etc.). For example, when the anxiety score is higher than 0.7, the fall detection process priority is set to the highest, and the detection frequency is increased to a 1-second interval; when the relaxation score is high, the priority is reduced, and the detection frequency is adjusted to a 5-second interval; in an excited state, to prevent false detections, the feature extraction parameters are optimized, and the detection process priority is set to medium. Examples of AI outputs include "Priority: High, Detection Interval: 1 second" and "Priority: Low, Detection Interval: 5 seconds." These priority controls serve as parameters for the fall detection scheduler and AI inference module, optimizing resource allocation and data transmission timing for notifications and instructions in real time during the fall detection process. The technical benefits include flexible priority control based on individual emotional states compared to previous unified detection methods, enabling early detection of significant anomalies, reducing false alarms, and optimizing system resources. Specific application areas include care for elderly people living alone, behavioral monitoring of patients with mental illness, safety management of elderly care facilities, and support for people with disabilities.

[0068] The fall detection unit can adjust its detection accuracy based on the material or condition of the floor during fall detection. For example, the fall detection unit considers the material or condition of the floor when adjusting its accuracy. Floor materials include, for example, carpets and wooden floors. Floor conditions include, for example, dampness and slipperiness. The fall detection unit, for example, increases the sensitivity of fall detection when the floor is slippery; adjusts the sensitivity when the floor is hard; and optimizes the fall detection algorithm based on the material or condition of the floor. By considering the material or condition of the floor, the accuracy of fall detection can be improved. Specifically, this fall detection unit acquires floor material data (such as labels for carpets, wooden floors, tatami mats, etc.) and floor condition data (such as humidity sensor values, slipperiness indicators, temperature and humidity values, etc.) from environmental sensors or the user interface, collecting them in structured data form. This fall detection unit inputs this floor information into an AI fall risk estimation module (such as decision trees, MLP, rule-based judgment, etc.) to calculate a fall risk score and detection sensitivity adjustment parameters. Examples of AI inputs include floor material labels (e.g., carpet = 0, wood flooring = 1), floor state vectors (humidity 0.8, slip resistance 0.7), and environmental conditions (temperature, humidity). The AI ​​outputs structured data such as a fall risk score (0.0–1.0), sensitivity adjustment values ​​(e.g., +0.2, -0.1), and recommended detection algorithms (e.g., high sensitivity mode, normal mode). For example, "fall risk 0.9, sensitivity +0.3" or "fall risk 0.4, sensitivity -0.1." Based on the AI ​​output, this fall detection department dynamically adjusts the thresholds and feature extraction parameters of the fall detection AI (LSTM, GRU, etc.) to increase sensitivity for early detection on slippery or damp floors and suppress false detections on hard or safe floors. Floor material and state data are stored as a historical database, and the AI ​​model learns online to adapt to environmental changes. Therefore, compared to previous detection methods that did not consider the environment, this flexible detection control that considers floor conditions improves accuracy, reduces false alarms, and enhances usability. The technological benefits include improved fall detection accuracy, reduced false and false detections, increased system flexibility, and user-optimized personalization. Specific applications include care for elderly people living alone, safety management of elderly care facilities, rehabilitation support, support for people with disabilities, and automatic optimization of smart home environments.

[0069] The fall detection unit can indicate different response measures based on the impact level of a fall during fall detection. For example, the fall detection unit indicates different response measures based on the impact level of the fall, such as the value of the accelerometer sensor or the impact intensity. For example, if the impact level is high, the fall detection unit indicates an emergency response; if the impact level is low, it indicates a mild response; and so on, indicating an appropriate response based on the impact level. By indicating different response measures based on the impact level, appropriate responses can be achieved. Specifically, this fall detection unit inputs three-axis acceleration data (e.g., 100Hz sampling, 2-second 200×3 matrix) obtained from the accelerometer and gyroscope sensors at the time of the fall event into an AI impact level estimation module (e.g., convolutional neural network CNN, decision tree, rule-based judgment, etc.) to calculate a fall impact level score (0.0~1.0) and a classification label (e.g., high impact, low impact, etc.). Examples of AI inputs include peak acceleration at the time of a fall (e.g., 2.5G), impact duration (e.g., 0.3 seconds), changes in vital signs (e.g., a sudden increase in heart rate), and the state of motion before the fall. The AI ​​outputs structured data such as an impact score (0.0–1.0), impact classification (high, medium, low), and recommended response tags (e.g., emergency response, observation, mild response). For example, "Impact 0.95, high, emergency response" or "Impact 0.45, low, mild response." Based on the AI ​​output, the fall detection department automatically generates and sends impact-based response instructions (e.g., call an ambulance, contact family, observe) to the instruction or notification department's equipment and systems. Furthermore, a historical database of past fall impact scores and response results is created, and the AI ​​model learns online to improve the accuracy of optimal responses. Therefore, compared to previous uniform response methods or subjective human judgment, flexible response instructions considering impact score improve response speed, accuracy, and reduce misresponses. Technical benefits include increased response speed, reduced misresponses, improved usability, and increased system flexibility. Specific application areas include care for elderly people living alone, safety management of elderly care facilities, rehabilitation support, support for people with disabilities, and remote health monitoring.

[0070] The instruction unit can infer the emotions of the elderly and adjust the emergency response instructions accordingly. For example, the instruction unit can infer the emotions of the elderly and adjust the emergency response instructions accordingly. Emotions include, for example, joy, sadness, and anger. The instruction unit can provide reassuring instructions when the elderly person is anxious; detailed instructions when the elderly person is relaxed; and concise instructions when the elderly person is excited. By adjusting the emergency response instructions according to the elderly person's emotions, appropriate responses can be achieved. Specifically, this instruction unit inputs multimodal data (such as facial expression feature vectors, MFCC feature quantities of speech waveforms, heart rate variability data, and the past 24 hours' movement history) acquired from image sensors, voice sensors, and vital sign sensors into an AI emotion inference module (such as multimodal neural networks, Transformer-based emotion classifiers, etc.). Examples of inputs to the AI ​​emotion inference module include 68-point marker coordinates of a facial image, a speech MFCC feature vector, a 5-minute RR interval sequence, and a 1440×3 matrix of movement history. The AI ​​outputs structured data such as emotion labels (e.g., anxiety, relaxation, excitement), emotion scores (0.0–1.0), and confidence indices. For example, "Emotion label: anxiety, score: 0.78" or "Emotion label: relaxation, score: 0.85." Based on the emotion inference results, this instruction unit automatically selects the template and expression method for the emergency response instruction generation module. For example, when the anxiety score is higher than 0.7, it generates reassuring information such as "Current situation is safe, please rest assured, an ambulance has been called"; when the relaxation score is high, it generates detailed instructions including anomaly detection score, health status details, and recommended coping actions (e.g., "A fall has been detected, an ambulance has been called and family members have been notified, please remain quiet and wait"); when the state of excitement is high, a concise and clear expression is used (e.g., "A fall has been detected, an ambulance has been called, please remain quiet"). AI output examples include "Instruction content: reassuring, detailed, concise," etc. These instructions are sent to the devices of family members, caregivers, or the elderly person's residence terminal through the instruction unit's output module. Furthermore, by analyzing the historical content of instructions and the reactions of the elderly and their families, AI models are used to learn online and personalize the instructions. This allows for more flexible instruction generation based on individual emotional states, compared to previous uniform methods, improving reassurance, comprehension, preventing misunderstandings, and enhancing usability. Technological benefits include improved accuracy, speed, and personalization in emergency responses, reduced mishandling, and increased system flexibility. Specific applications include care for elderly people living alone, emergency response in elderly care facilities, support for people with disabilities, and remote health monitoring.

[0071] The command center can select the optimal response method during emergency situations by referring to past response history. For example, the command center can select the optimal response method by referring to past response history. Past response history includes, for example, response methods and outcomes. The command center can analyze past emergency response history to select the optimal response method; recommend rapid response methods based on past response history; and select appropriate response methods by referring to past response history. Specifically, this command center stores historical data (such as the time of occurrence, response method, time required to complete the response, outcome labels, and the reactions of family members and caregivers) of past emergency response events (such as calling an ambulance, contacting family members, and observation) in the form of a time-series database. This command center inputs this historical data into an AI response optimization module (such as gradient boosting decision trees, MLP, and time-series clustering) to learn past response patterns, success rates, response speeds, and the response tendencies of family members and caregivers. Examples of AI inputs include response history over the past 30 days (e.g., a 288×5 matrix of response type, method, time, and result), family response success rates (e.g., 0.95% success rate for calling an ambulance, 0.9% success rate for contacting family members), and past emergency response history (e.g., 10 emergency responses, average response time 5 minutes). AI outputs structured data such as recommended response methods (e.g., calling an ambulance, contacting family members, observation), recommended response methods (e.g., phone call, SMS, on-site visit), and recommended response timing (e.g., immediately, after 5 minutes). For example, "Recommended response: Call an ambulance, immediately" or "Recommended response: Contact family members, phone call." Based on the AI ​​output, this instruction unit controls the instruction content generation module to automatically select the optimal response method, method, and timing based on history and output the instruction. It also continuously collects post-response result data and improves optimization accuracy through online learning of the AI ​​model. Therefore, compared to previous uniform response methods or manual selection, optimizing responses using individual and contextual history achieves technical effects such as improved response speed, reduced false alarms and missed reports, improved availability, and increased system resource efficiency. Specific application areas include care for elderly people living alone, emergency response in elderly care facilities, support for people with disabilities, and remote health monitoring.

[0072] The instruction unit can adjust the level of detail in its response based on the type of anomaly during an emergency. For example, during an emergency, the instruction unit adjusts the level of detail based on the type of anomaly. Types of anomalies include, for example, abnormal movement or abnormal vital signs. For instance, when a fall is detected, the instruction unit provides detailed response instructions; when a health abnormality is detected, the level of detail is adjusted based on the type of anomaly; when an abnormal movement is detected, the level of detail is adjusted based on the type of anomaly. By adjusting the level of detail based on the type of anomaly, appropriate responses can be achieved. Specifically, this instruction unit inputs anomaly judgment data (such as anomaly type labels, anomaly scores, urgency judgment values, etc.) received from the analysis unit, fall detection unit, health monitoring unit, etc., into the instruction content generation module, and applies a level of detail judgment rule or AI classification model (such as decision tree, MLP, rule-based judgment, etc.) to each anomaly type. Examples of AI input include structured data such as "Anomaly type: fall, anomaly score: 0.95", "Anomaly type: abnormal vital signs, anomaly score: 0.88", and "Anomaly type: abnormal movement, anomaly score: 0.75". The AI ​​outputs response detail labels (e.g., detailed, normal, simple), recommended instruction templates (e.g., detailed instructions, simple instructions), and recommended response methods (e.g., call an ambulance, contact family, observe). For example, "Fall: detailed instructions, call an ambulance," "Abnormal vital signs: normal instructions, contact family," "Abnormal movement: simple instructions, observe," etc. Based on the AI ​​output, this instruction department controls the instruction content generation module to automatically select and output the optimal detail, instruction content, and response method for each type of anomaly. It also analyzes the response history of each anomaly type and the reaction history of family members and caregivers, continuously optimizing the detail determination rules and AI model parameters. Therefore, compared to the previous uniform detail instruction method, this anomaly type optimization achieves improved response speed, reduced false alarms and missed alarms, improved availability, and improved system resource efficiency. Specific application areas include care for elderly people living alone, emergency response in elderly care facilities, support for people with disabilities, and remote health monitoring.

[0073] The instruction unit can infer the emotions of elderly individuals and determine the priority of emergency responses based on the inferred emotions. For example, the instruction unit can infer the emotions of elderly individuals and determine the priority of emergency responses based on the inferred emotions. Emotions include, for example, joy, sadness, and anger. The instruction unit, for example, increases the priority of emergency responses when the elderly person is anxious; decreases the priority of emergency responses when the elderly person is relaxed; and appropriately adjusts the priority of emergency responses when the elderly person is excited. By determining the priority of emergency responses based on the elderly person's emotions, important anomalies can be addressed earlier. Specifically, this instruction unit inputs multimodal data (such as facial image feature vectors, MFCC features of speech waveforms, heart rate variability data, and the past 24 hours' movement history) acquired from image sensors, voice sensors, and vital sign sensors into an AI emotion inference module (such as multimodal neural networks, Transformer-based emotion classifiers, etc.) to infer emotion labels (such as anxiety, relaxation, excitement, etc.) and emotion scores (0.0–1.0). Examples of AI inputs include 68-point marker coordinates of facial images, speech features, 5-minute RR interval sequences, and movement history in a 1440×3 matrix. The AI ​​outputs structured data such as emotion tags, emotion scores, and confidence indices. Examples include "Emotion tag: Anxiety, Score: 0.75" and "Emotion tag: Relaxation, Score: 0.85." Based on the emotion inference results, the instruction unit dynamically controls the priority parameters and response scheduling of the emergency response priority control module. For example, when the anxiety score is higher than 0.7, the emergency response process priority is set to the highest, and the frequency of response instructions is increased to a 1-minute interval; when the relaxation score is high, the priority is lowered, and the frequency of response instructions is adjusted to a 5-minute interval; in an excited state, to prevent mis-response, the feature extraction parameters are optimized, and the response process priority is set to medium. AI output examples include "Priority: High, Response Interval: 1 minute" and "Priority: Low, Response Interval: 5 minutes." These priority controls serve as parameters for the instruction unit scheduler and the AI ​​inference module, optimizing resource allocation and notification timing for the emergency response process and the instruction unit's data transmission in real time. The technical effect is that, compared to previous uniform response methods, priorities can be flexibly controlled according to individual emotional states, enabling early response to important anomalies, reducing mis-response, and optimizing system resources. Specific application areas include care for elderly people living alone, emergency response for patients with mental illness, safety management of elderly care facilities, and support for people with disabilities.

[0074] The instruction department can select the optimal response method during emergency situations by considering the geographical location information of family members or caregivers. For example, when family members or caregivers are nearby, the instruction department selects the optimal response method based on their geographical location information, such as GPS information and addresses. For instance, when family members or caregivers are nearby, the instruction department selects a response method that encourages direct visitation; when they are far away, it instructs them to respond via telephone or text message; and it selects the optimal response method based on the geographical location information of family members or caregivers. By selecting the optimal response method based on the geographical location information of family members or caregivers, rapid response can be achieved. Specifically, this instruction department establishes an up-to-date geographical location information database (such as GPS coordinates, address, movement history, etc.) for each family member or caregiver, and obtains this information in real time through the location information acquisition module when an emergency occurs. This instruction department inputs the current location of the person being addressed, the location of the anomaly (e.g., room information in the elderly person's residence), and past response history (e.g., number of on-site visits, number of remote responses, etc.) into the AI ​​response method selection module (e.g., decision tree, MLP, rule-based judgment, etc.). Examples of AI input include "Family member A: GPS coordinates [35.6, 139.7], Elderly person's residence: GPS coordinates [35.6, 139.8], Distance: 1.2km" and "Caregiver B: Address: Tokyo, Distance: 10km". The AI ​​output consists of structured data such as recommended response methods (e.g., direct access promotion, telephone, SMS, email), response content templates (e.g., on-site visit request, remote response request), and response priorities (high, medium, low). For example, "Family member A: Recommended direct access, immediate response" and "Caregiver B: Telephone response, detailed content". Based on the AI ​​output, this instruction unit controls the response scheduler and content generation module to automatically select the optimal response method, content, and timing and output instructions. It also continuously collects historical data after responses and improves optimization accuracy through online learning of the AI ​​model. Therefore, compared to previous unified response methods, optimizing responses using geographic location information achieves technical effects such as improved response speed, reduced false alarms and missed reports, improved availability, and increased system resource efficiency. Specific application areas include care for elderly people living alone, emergency response in elderly care facilities, support for people with disabilities, and remote health monitoring.

[0075] The instruction department can consider the device information of family members or caregivers during emergency responses and adjust the response accordingly. For example, device information refers to smartphones, tablets, etc. The instruction department may send push notifications to instruct family members or caregivers to respond when they are using smartphones, or via email when they are using computers. The instruction department can also select the optimal response based on the device information of family members or caregivers. By adjusting the response based on the device information of family members or caregivers, appropriate information delivery is achieved. Specifically, this instruction department builds a device usage history database for each family member or caregiver, recording the device type (e.g., smartphone, tablet, PC, feature phone), operating system type, application usage, notification receiving settings, etc., for each response event. This instruction department inputs the latest device information, past response history, and other relevant data into the AI ​​response method selection module (e.g., decision tree, MLP, rule-based judgment). Examples of AI input include "Family member A: Smartphone usage 0.95%, PC usage 0.05%" and "Caregiver B: Tablet usage 0.8%, notification app installed." The AI ​​output consists of structured data such as recommended response formats (e.g., push notifications, SMS, email, voice calls), response content templates (e.g., detailed instructions, simple instructions), and response priorities (high, medium, low). For example, "Family member A: Push notification, detailed content" and "Caregiver B: Email notification, simple content." Based on the AI ​​output, this instruction unit controls the response scheduler and content generation module to automatically select and output the optimal response format, content, and timing suitable for the device information. Furthermore, it continuously collects the reception and response history after each response, improving optimization accuracy through online learning of the AI ​​model. Therefore, compared to traditional uniform response methods, optimizing responses using device information achieves improved accuracy, speed, and usability in information delivery, as well as increased system resource efficiency. Specific application areas include care for elderly people living alone, emergency response in nursing facilities, support for people with disabilities, and remote health monitoring.

[0076] The health monitoring department can infer the emotions of elderly individuals and adjust the sensitivity of health monitoring accordingly. For example, the department infers the emotions of elderly individuals, such as joy, sadness, and anger. The department can increase the sensitivity of health monitoring when an elderly person is anxious to detect abnormalities early; decrease the sensitivity when an elderly person is relaxed to reduce false positives; and appropriately adjust the sensitivity when an elderly person is excited to achieve accurate monitoring. By adjusting the sensitivity of health monitoring based on the emotions of the elderly, false positives can be reduced and accurate monitoring can be achieved. Specifically, this health monitoring department inputs multimodal data (such as facial expression feature vectors, MFCC feature vectors of speech waveforms, heart rate variability data, and action records from the past 24 hours) acquired from image sensors, voice sensors, and vital sign sensors into an AI emotion inference module (such as multimodal neural networks and Transformer-based emotion classifiers). Examples of inputs to the AI ​​emotion inference module include the coordinates of 68 markers on a facial image, the MFCC feature vector of speech, a 5-minute RR interval sequence, and a 1440×3 matrix of action records. The AI ​​outputs structured data such as emotion labels (e.g., anxiety, relaxation, excitement), emotion scores (0.0–1.0), and confidence indices. Examples include "Emotion Label: Anxiety, Score: 0.78" and "Emotion Label: Relaxation, Score: 0.85." Based on the emotion inference results, the health monitoring department dynamically adjusts the sensitivity parameters and thresholds of the health monitoring AI (e.g., LSTM, GRU, CNN). For example, when the anxiety score is greater than 0.7, the health monitoring threshold is lowered from 0.7 to 0.5; when the relaxation score is high, the threshold is raised to 0.8. During excitement, feature extraction parameters are optimized, and filtering is strengthened to prevent false positives. AI output examples include "Abnormal Score 0.92, Type: Abnormal Vital Signs, Urgency: Urgent" or "Abnormal Score 0.45, Type: Mild Abnormality, Urgency: Requires Attention." These outputs are compared with baseline values ​​in the threshold determination department and forwarded to the notification or instruction department when necessary. The health monitoring department uses the AI ​​model's online learning function to reflect daily new data, automatically updating model parameters to adapt to changes in individual emotions and health status. The technological advantages include improved accuracy, reduced false alarms, and enhanced usability compared to traditional uniform threshold methods. Specific applications include care for elderly people living alone, health monitoring of patients with mental illness, safety management of nursing facilities, support for people with disabilities, and home-based medical support.

[0077] The health monitoring department can analyze the past health data of elderly individuals to improve the accuracy of anomaly detection. For example, the health monitoring department analyzes the past health data of elderly individuals to improve the accuracy of anomaly detection. Health data includes, for example, past health records and vital sign data. The health monitoring department collects past health data of elderly individuals to learn anomaly patterns. The health monitoring department can also identify specific high-risk conditions based on the past health data of elderly individuals. The health monitoring department can also analyze the health data of elderly individuals to optimize anomaly detection algorithms. By analyzing past health data, the accuracy of anomaly detection is improved. Specifically, this health monitoring department accumulates past vital sign data (such as a 288×4 matrix / day of body temperature, heart rate, blood pressure, respiratory rate, etc.), health records (such as medication history, past medical history, subjective health scores, etc.), and abnormal event history (such as the time of occurrence of anomalies, the type of anomaly, and the response results, etc.) in the form of time-series tensors or structured databases. This health monitoring department inputs this historical data into AI anomaly detection models (such as LSTM-based time-series prediction models, autoencoders, clustering algorithms, etc.) to learn characteristic patterns and risk factors (such as periodic increases in heart rate, sudden changes in body temperature, abnormal fluctuations in blood pressure, etc.) before anomalies occur in a high-dimensional space. Examples of AI input include a 30-day history tensor of vital signs (288×4×30), a list of anomaly occurrence times, and health status labels (such as good, requiring attention, abnormal, etc.). The AI ​​output consists of structured data such as anomaly risk scores (0.0–1.0), risk factor labels (such as arrhythmia, fever tendency, blood pressure fluctuations, etc.), and anomaly pattern classifications (such as periodic, sudden, etc.). For example, "Anomaly risk score 0.85, factor: arrhythmia" or "Anomaly risk score 0.92, factor: fever tendency." Based on these AI outputs, this health monitoring department personalizes and optimizes the threshold parameters and feature extraction methods of the real-time health monitoring AI (LSTM, GRU, etc.) to improve sensitivity and achieve early detection when similar anomaly patterns are detected. Furthermore, the health database possesses online learning capabilities, reflecting daily new health events and near-miss cases, and automatically updating AI model parameters to adapt to changes in individual risk trends. Therefore, compared to traditional fixed-rule or simple threshold-based methods, it achieves personalized, time-varying responses, reduced false positives, and improved detection accuracy. Technical benefits include improved health monitoring accuracy (significantly reducing false positives and false negatives), faster response times (real-time notifications and instructions), improved data management efficiency (automatic recording and historical analysis), and enhanced system flexibility. Specific application areas include care for elderly people living alone, health management of nursing facilities, rehabilitation support, health risk management for patients with chronic diseases, and support for people with disabilities.

[0078] The Health Monitoring Department can optimize monitoring accuracy based on the influence of seasons and weather during health monitoring. For example, the department considers the impact of seasons and weather to optimize monitoring accuracy. Seasons include summer and winter, and weather includes sunny and rainy days. The department adjusts the sensitivity of health monitoring during seasonal transitions to detect anomalies earlier, and improves sensitivity during inclement weather. The department can also optimize health monitoring algorithms based on the influence of seasons and weather, thereby improving the accuracy of health monitoring. Specifically, the department collects seasonal information (such as month, date, average temperature, and sunshine duration) and weather information (such as time-series data on sunny and rainy days, humidity, air pressure, and outdoor temperature) as structured data from environmental sensors or external meteorological APIs. This environmental data is then input into the health monitoring AI (such as decision trees, MLP, and rule-based decision-making) as supplementary information, dynamically adjusting anomaly detection thresholds and feature extraction parameters for different seasons and weather conditions. Examples of AI input include environmental vectors such as "Season: Summer, Outdoor Temperature: 32℃, Humidity: 80%" and "Weather: Rain, Air Pressure: 990hPa, Indoor Temperature: 28℃," as well as vital sign data (such as body temperature, heart rate, and blood pressure) at the same time. The AI ​​output is structured data such as anomaly score, anomaly type, and urgency level. Through seasonal and weather optimization, false positives and false negatives are reduced. For example, "Anomaly score 0.88, Type: Heatstroke risk, Urgency: Requires attention" and "Anomaly score 0.92, Type: Hypothermia risk, Urgency: Urgent." These outputs are compared with baseline values ​​in the threshold determination department and forwarded to the notification or instruction department when necessary. The health monitoring department optimizes AI model parameters through online learning based on environmental changes, flexibly responding to seasonal diseases and weather-related risks. The technical effect is that, compared to traditional monitoring that does not consider the environment, it achieves improved monitoring accuracy, reduced false alarms, and improved usability under environmental changes. Specific application areas include care for elderly people living alone, health management of nursing facilities, remote health monitoring, support for people with disabilities, and home medical support.

[0079] The health monitoring department can infer the emotions of elderly individuals and determine the priority of health monitoring based on these inferred emotions. For example, the department infers the emotions of elderly individuals and determines the priority of health monitoring accordingly. Emotions include, for instance, joy, sadness, and anger. The department may increase the priority of health monitoring when an elderly person feels anxious, decrease the priority when they are relaxed, and adjust the priority appropriately when they are excited. By prioritizing health monitoring based on the elderly person's emotions, important abnormalities can be detected early. Specifically, this health monitoring department inputs multimodal data (such as facial image feature vectors, MFCC features of speech waveforms, heart rate variability data, and action records from the past 24 hours) obtained from image sensors, voice sensors, and vital sign sensors into an AI emotion inference module (such as a multimodal neural network or a Transformer-based emotion classifier) ​​to infer emotion labels (such as anxiety, relaxation, and excitement) and emotion scores (0.0–1.0). Examples of AI inputs include facial expression feature vectors (68-point marker coordinates), speech features, heart rate variability data (5-minute RR interval sequence), and 24-hour action records (1440×3 matrix). AI outputs structured data such as emotion labels, emotion scores, and confidence indices. Examples include "Emotion Label: Anxiety, Score: 0.75" and "Emotion Label: Relaxation, Score: 0.85." Based on emotion inference results, this health monitoring department dynamically controls the priority parameters and monitoring schedule of the health monitoring AI (LSTM, GRU, etc.). For example, when the anxiety score is greater than 0.7, the health monitoring process priority is set to the highest, and the monitoring frequency is increased to a 1-minute interval; when the relaxation score is high, the priority is reduced, and the monitoring frequency is adjusted to a 5-minute interval; in an excited state, to prevent false detections, the feature extraction parameters are optimized, and the monitoring process priority is set to medium. Examples of AI outputs include "Priority: High, Monitoring Interval: 1 minute" and "Priority: Low, Monitoring Interval: 5 minutes." These priority controls serve as parameters for the health monitoring department's scheduler and AI inference module, optimizing resource allocation and data transmission timing for notifications and instructions in real time during the health monitoring process. The technical benefits include, compared to traditional unified monitoring methods, early detection of significant anomalies, reduced false alarms, and optimal system resource optimization by flexibly controlling priorities based on individual emotional states. Specific application areas include care for elderly people living alone, health monitoring of patients with mental illness, safety management of nursing facilities, and support for people with disabilities.

[0080] The Health Monitoring Department can improve the accuracy of anomaly detection during health monitoring by referencing the health data of family members or caregivers. For example, during health monitoring, the department can improve the accuracy of anomaly detection by referring to the health data of family members or caregivers. This health data includes, for example, past health records and vital sign data. The department can learn anomaly patterns by referring to this data. The department can also identify specific high-risk conditions based on the health data of family members or caregivers. Furthermore, the department can analyze the health data of family members or caregivers to optimize anomaly detection algorithms. Specifically, the department accumulates the health records of family members or caregivers (such as vital sign history, past medical history, genetic risk information, and lifestyle data) in the form of time-series tensors or structured databases. This data is then input into an AI anomaly detection model (such as decision trees, MLP, clustering algorithms, etc.) to learn common anomaly patterns and genetic risk factors (such as hypertension tendency, heart disease risk, etc.) among family members or caregivers. The AI's input examples include the vital sign history of three family members over the past year (a 365×4×3 matrix), past medical history tags (such as hypertension, diabetes, etc.), and lifestyle habit scores (such as exercise frequency, balanced diet, etc.). The AI's output consists of structured data such as an abnormal risk score (0.0–1.0), risk factor tags (such as hereditary hypertension, lifestyle-related disease risk, etc.), and abnormal pattern classifications (such as familial common patterns, individual specific patterns, etc.). For example, "Abnormal risk 0.85, factor: familial hypertension" or "Abnormal risk 0.92, factor: lifestyle-related disease." Based on these AI outputs, our health monitoring department optimizes the threshold parameters and feature extraction methods of the real-time health monitoring AI according to the risk trends of family members or caregivers. This improves sensitivity and enables early detection when hereditary or lifestyle-related risks are high. Furthermore, the health database of family members or caregivers has an online learning function, reflecting daily new data and automatically updating the AI ​​model parameters to adapt to changes in risk trends. Therefore, compared to traditional individual monitoring methods, utilizing the health information of family members or caregivers achieves risk prediction, reduces false positives, and improves detection accuracy. The technological benefits include improved accuracy in health monitoring, reduced false positives and false negatives, increased system flexibility, and family-based health risk management. Specific applications include care for elderly people living alone, family health management, health risk assessment in nursing facilities, early detection of hereditary diseases, and support for people with disabilities.

[0081] The health monitoring department can apply different monitoring algorithms based on the type of abnormality during health monitoring. For example, during health monitoring, the department applies different monitoring algorithms depending on the type of abnormality. Types of abnormalities include, for example, abnormal movements, abnormal vital signs, etc. For instance, when an abnormal body temperature is detected, a specific monitoring algorithm for body temperature is applied; when an abnormal heart rate is detected, a specific monitoring algorithm for heart rate is applied; and when an abnormal blood pressure is detected, a specific monitoring algorithm for blood pressure is applied. By applying the optimal monitoring algorithm based on the type of abnormality, monitoring accuracy is improved. Specifically, this health monitoring department determines the type of abnormal event detected (such as abnormal body temperature, abnormal heart rate, abnormal blood pressure, abnormal movements, etc.) and selects a dedicated AI monitoring module for each type. For example, abnormal body temperature detection uses temporal LSTM or decision trees, inputting body temperature data (e.g., 288 points per minute / day) and outputting an abnormality score and urgency level; abnormal heart rate detection uses CNN or RNN, inputting heart rate variability data (e.g., 5-minute RR interval sequences, 288×1 matrices) and outputting an abnormality score and abnormality type; abnormal blood pressure detection uses multilayer perceptron (MLP) or random forest, inputting blood pressure data (e.g., 288×2 matrices) and outputting an abnormality score and abnormality type. Examples of AI outputs include "abnormal body temperature score 0.95, urgency level: urgent," "abnormal heart rate score 0.88, type: arrhythmia, urgency level: require attention," and "abnormal blood pressure score 0.85, type: hypertension, urgency level: require attention," etc. These outputs are compared with baseline values ​​in the threshold determination department and forwarded to the notification or instruction department when necessary. The health monitoring department optimizes feature extraction methods and judgment criteria for each abnormality type, significantly improving monitoring accuracy and response speed through the collaboration of multiple AI models. The technical benefits are significant: compared to traditional single-algorithm methods, optimization based on anomaly types reduces false positives, improves detection accuracy, and enhances system flexibility. Specific application areas include care for elderly people living alone, health management of nursing facilities, remote health monitoring, support for people with disabilities, and home-based medical support.

[0082] The system involved in this embodiment is not limited to the above example. For example, various changes can be made as follows.

[0083] The motion detection unit can monitor room temperature and humidity in real time and automatically adjust its sensitivity when abnormal environmental conditions are detected. For example, when the room temperature rises sharply, the motion detection unit increases its sensitivity to detect abnormal movements earlier. Furthermore, when humidity is high, the sensitivity of the motion detection can be adjusted to prevent false detections. Thus, motion detection based on environmental conditions achieves more accurate monitoring.

[0084] Fall detection systems can predict the fall risk of older adults and adjust the sensitivity of fall detection based on the predicted risk. For example, if an older adult has a history of falls, the system increases sensitivity to detect a potential recurrence earlier. Furthermore, when an older adult is in poor physical condition, the fall risk increases, allowing for further increases in sensitivity. Therefore, fall detection based on fall risk provides a more reliable guarantee of safety for older adults.

[0085] The health monitoring department can collect dietary and exercise data from older adults and assess their health status based on this data. For example, it can monitor whether older adults are consuming a balanced diet; if nutritional deficiencies are detected, family members or caregivers will be notified; if insufficient exercise is detected, notifications can be issued to promote appropriate exercise. Thus, by utilizing dietary and exercise data, the health status of older adults can be comprehensively monitored.

[0086] The analysis unit can infer the emotions of elderly individuals and adjust the display of the analysis results accordingly. For example, when an elderly person is anxious, the analysis results are displayed concisely to provide reassurance; when an elderly person is relaxed, detailed analysis results are displayed to enhance understanding; and when an elderly person is excited, the analysis results can be displayed in a visually easy-to-understand manner. Thus, by displaying analysis results based on the elderly person's emotions, easily understandable information is provided.

[0087] The notification department can optimize notification methods by referencing past response records of family members or caregivers. For example, it can prioritize notification methods that allow for a rapid response from family members or caregivers, thus enabling swift action in emergencies. Furthermore, the content and timing of notifications can be adjusted based on past response records. Therefore, by utilizing past response records, the optimal notification method can be selected to achieve rapid and appropriate information delivery.

[0088] The instruction system can anticipate the emotions of elderly individuals and adjust emergency response instructions accordingly. For example, when an elderly person is anxious, reassuring instructions are provided; when they are relaxed, detailed instructions are offered to enhance understanding; and when they are excited, concise and easy-to-understand instructions are provided. Thus, by providing emergency response instructions based on the elderly person's emotions, appropriate responses are achieved.

[0089] The health monitoring department can optimize monitoring accuracy based on seasonal and weather effects. For example, the sensitivity of health monitoring can be adjusted during seasonal transitions to detect anomalies earlier; and the sensitivity can be increased during inclement weather. This makes it possible to consider the impact of seasons and weather on health monitoring, resulting in more accurate anomaly detection.

[0090] The motion detection unit can filter pet movements to prevent false detections. For example, it can distinguish pet movements from those of elderly people, excluding pet movements from the detection list; it can also learn pet movement patterns to reduce false detections. Thus, by filtering pet movements, the accuracy of motion detection is improved, and false detections are prevented.

[0091] The notification department can anticipate the emotions of elderly individuals and adjust notification content accordingly. For example, when an elderly person is anxious, notifications can be sent with reassuring content; when they are relaxed, notifications containing detailed information can be sent; and when they are excited, notifications can be concise and easy to understand. Thus, by tailoring notifications to the elderly person's emotions, appropriate information delivery is achieved.

[0092] The command center can consider the geographical location of family members or caregivers during emergency responses and select the optimal response method. For example, when family members or caregivers are nearby, the command center chooses a response method that encourages them to visit directly; when family members or caregivers are far away, the command center can provide instructions via telephone or text message. Thus, selecting the optimal response method based on the geographical location of family members or caregivers enables rapid response.

[0093] The following is a brief description of the processing flow of the implementation method.

[0094] Step 1: The motion detection unit detects the elderly person's movements. For example, motion detection units are installed throughout the room, using motion sensors to monitor the elderly person's walking, standing, sitting, and other movements in real time. Step 2: The analysis unit analyzes the information detected by the motion detection unit. For example, AI or machine learning algorithms are used to analyze the motion data and detect anomalies. Step 3: The notification unit issues notifications based on the information analyzed by the analysis unit. For example, when an anomaly is detected, family members or caregivers are notified. Step 4: The fall detection unit detects falls in the elderly person. For example, accelerometers worn on the elderly person's body are used to detect falls immediately. Step 5: The instruction unit instructs emergency responses based on the information analyzed by the analysis unit. For example, when a fall is detected, instructions are given to call an ambulance or contact family members. Step 6: The health monitoring unit monitors the elderly person's health. For example, the unit measures the elderly person's vital signs such as body temperature, heart rate, and blood pressure in real time and sends the information to the AI ​​through a medical monitoring system. Specifically, in step 1, this system installs multiple motion sensors (infrared sensors, ultrasonic sensors, image sensors, etc.) on the ceiling or wall to acquire the elderly person's position coordinates and movement status (still, walking, pre-fall signs, etc.) at 1-second intervals as three-dimensional vector data. In step 2, this sensor data is sent to the central processing unit as a time-series tensor (e.g., a 30×3 matrix for 30 seconds of movement recording). After denoising (moving average filtering, outlier removal) and standardization (Z-score normalization) in the data preprocessing unit, feature extraction and anomaly detection are performed using a convolutional neural network (CNN) or a recurrent neural network (RNN). AI input examples include continuous vector sequences such as t=0s: [0.1, 0.2, 0.0] and t=1s: [0.15, 0.18, 0.02]. The output is structured data such as anomaly score (0.0~1.0), anomaly type (fall, wandering, normal), and urgency level (urgent, requires attention, normal). In step 3, the notification unit receives the anomaly judgment results from the analysis unit, fall detection unit, and health monitoring unit. It compares these results with a preset benchmark value (anomaly score above 0.8) in the threshold judgment unit. If the score exceeds the threshold, it automatically selects the optimal notification method (push notification, SMS, email, voice call, etc.) and sends the message, referencing the device information (smartphone, tablet, PC, etc.) and geographical location information (GPS coordinates, address) of family members or caregivers. In step 4, the fall detection unit samples three-axis acceleration data at 100Hz using an accelerometer and gyroscope worn on the elderly person's waist or wrist. It then uses a time-series AI model such as LSTM or GRU to analyze fall-specific patterns. AI output examples include "Fall score 0.95, urgency: urgent" or "Fall score 0.45, urgency: caution required," etc. In step 5, the instruction unit automatically instructs actions such as calling an ambulance, immediately contacting nearby family members, and notifying nursing service providers based on the urgency level.In step 6, the health monitoring department retrieves vital signs such as body temperature, heart rate, and blood pressure from the medical monitoring system every 5 minutes, inputting these as a continuous value array into the AI, and outputting anomaly scores and anomaly types (fever, tachycardia, hypotension, etc.). This series of processes differs from traditional manual care or simple threshold judgment methods, achieving comprehensive improvements in computer technology, including AI analysis of high-dimensional sensor data, multi-module collaboration, real-time anomaly detection, notification, and indication. Technical benefits include improved anomaly detection accuracy (significantly reducing false positives and false negatives), faster response speed (real-time notifications and indications), improved data management efficiency (automatic recording and historical analysis), and optimized communication load (notifications only when necessary, distributed processing). Specific application areas include diverse use cases such as care for elderly people living alone, home medical support, safety management of nursing facilities, remote health monitoring, support for people with disabilities, and child and pet care.

[0095] The specific processing unit 290 sends the result of the specific processing to the smart device 14. In the smart device 14, the control unit 46A causes the output device 40 to output the result of the specific processing. The microphone 38B acquires voice representing the user's input to the result of the specific processing. The control unit 46A sends the voice data representing the user's input acquired by the microphone 38B to the data processing device 12. In the data processing device 12, the specific processing unit 290 acquires the voice data.

[0096] Data generation model 58 is what is known as generative AI (Artificial Intelligence). An example of data generation model 58 includes ChatGPT (registered trademark) (Internet search).<URL: https: / / openai.com / blog / chatgpt> Generative AI, such as data generation model 58, is obtained by deep learning through a neural network. The data generation model 58 is input with a prompt containing instructions, and with inference data such as speech data representing speech, text data representing text, and image data representing images (e.g., data of still images or data of moving images). The data generation model 58 infers the inference data based on the instructions shown in the prompt and outputs the inference result in one or more data forms such as speech data, text data, and image data. The data generation model 58 includes, for example, text generation AI, image generation AI, and multimodal generation AI. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization. The specific processing unit 290 performs the above-described specific processing while using the data generation model 58. The data generation model 58 can also be a fine-tuned model to output inference results from prompts without instructions; in this case, the data generation model 58 can output inference results from prompts without instructions. The data processing device 12, etc., includes multiple data generation models 58, including AI other than generative AI. AI other than generative AI includes, but is not limited to, linear regression, logistic regression, decision trees, random forests, support vector machines (SVM), k-means clustering, convolutional neural networks (CNN), recurrent neural networks (RNN), generative adversarial networks (GAN), or Naive Bayes. Furthermore, AI can also act as an AI agent. Additionally, when the processing described above is performed by AI, this processing may be partially or entirely performed by AI, but is not limited to this example. Moreover, processing performed by AI, including generative AI, can be replaced by rule-based processing, and vice versa.

[0097] Furthermore, the processing performed by the aforementioned data processing system 10 is executed by the specific processing unit 290 of the data processing device 12 or the control unit 46A of the smart device 14, but it can also be executed jointly by the specific processing unit 290 of the data processing device 12 and the control unit 46A of the smart device 14. Additionally, the specific processing unit 290 of the data processing device 12 acquires or collects information required for processing from the smart device 14 or external devices, and the smart device 14 acquires or collects information required for processing from the data processing device 12 or external devices.

[0098] Each of the aforementioned elements, including a motion detection unit, an analysis unit, a notification unit, a fall detection unit, an instruction unit, and a health monitoring unit, can be implemented, for example, in at least one of the smart device 14 and the data processing device 12. For example, the motion detection unit uses the motion sensor of the smart device 14 to detect the elderly person's movements. The analysis unit, for example, is implemented by the specific processing unit 290 of the data processing device 12, which analyzes the motion data. The notification unit, for example, is implemented by the control unit 46A of the smart device 14, which notifies family members or caregivers when an anomaly is detected. The fall detection unit, for example, uses the accelerometer of the smart device 14 to detect falls in the elderly person. The instruction unit, for example, is implemented by the specific processing unit 290 of the data processing device 12, which instructs emergency response. The health monitoring unit, for example, uses the medical monitoring system of the smart device 14 to monitor the elderly person's health status. The correspondence between the various units and the device or control unit is not limited to the above examples and can be modified in various ways.

[0099] [Second Implementation] Figure 3 An example of the configuration of the data processing system 210 according to the second embodiment is shown.

[0100] like Figure 3 As shown, the data processing system 210 includes a data processing device 12 and smart glasses 214. One example of the data processing device 12 is a server.

[0101] The data processing device 12 includes a computer 22, a database 24, and a communication I / F 26. The computer 22 includes a processor 28, RAM 30, and memory 32. The processor 28, RAM 30, and memory 32 are connected to a bus 34. Furthermore, the database 24 and the communication I / F 26 are also connected to the bus 34. The communication I / F 26 is connected to a network 54. An example of the network 54 includes a WAN and / or LAN, etc.

[0102] The smart glasses 214 include a computer 36, a microphone 238, a speaker 240, a camera 42, and a communication I / F 44. The computer 36 includes a processor 46, RAM 48, and memory 50. The processor 46, RAM 48, and memory 50 are connected to a bus 52. Furthermore, the microphone 238, speaker 240, and camera 42 are also connected to the bus 52.

[0103] Microphone 238 receives user commands by receiving the user's voice. Microphone 238 captures the user's voice, converts the captured sound into speech data, and outputs it to processor 46. Speaker 240 outputs sound according to commands from processor 46.

[0104] Camera 42 is a small digital camera equipped with an optical system such as a lens, aperture and shutter, and an imaging element such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor, used to capture the user's surroundings (e.g., the shooting range defined by an angle of view equivalent to the field of vision of an average healthy person).

[0105] Communication I / F 44 is connected to network 54. Communication I / F 44 and 26 are responsible for the transmission and reception of various information between processor 46 and processor 28 via network 54. The transmission and reception of various information between processor 46 and processor 28 using communication I / F 44 and 26 is performed in a secure state.

[0106] Figure 4 An example of the main functions of the data processing device 12 and the smart glasses 214 is shown. Figure 4 As shown, in the data processing device 12, specific processing is performed by the processor 28. The memory 32 stores a specific processing program 56.

[0107] The processor 28 reads a specific processing program 56 from the memory 32 and executes the read specific processing program 56 on the RAM 30. The specific processing is implemented by the processor 28 operating as a specific processing unit 290 based on the specific processing program 56 executed on the RAM 30.

[0108] The memory 32 stores a data generation model 58 and an emotion-specific model 59. The data generation model 58 and the emotion-specific model 59 are used by the specific processing unit 290. The specific processing unit 290 can use the emotion-specific model 59 to infer the user's emotions and perform specific processing based on the user's emotions. The emotion inference function using the emotion-specific model 59 (emotion-specific function) includes various inferences and predictions related to the user's emotions, but is not limited to this example. Furthermore, emotion inference and prediction also include, for example, emotion analysis (analysis).

[0109] In the smart glasses 214, specific processing is performed by the processor 46. A specific processing program 60 is stored in the memory 50. The processor 46 reads the specific processing program 60 from the memory 50 and executes the read specific processing program 60 on the RAM 48. Specific processing is implemented by the processor 46 operating as a control unit 46A based on the specific processing program 60 executed on the RAM 48. Furthermore, the smart glasses 214 may also have the same data generation model and emotion-specific model as the data generation model 58 and the emotion-specific model 59, and use these models to perform the same processing as the specific processing unit 290.

[0110] Furthermore, other devices besides the data processing device 12 may also have the data generation model 58. For example, a server device may have the data generation model 58. In this case, the data processing device 12 obtains the processing results (prediction results, etc.) using the data generation model 58 by communicating with the server device that has the data generation model 58. In addition, the data processing device 12 may be a server device or a user-owned terminal device (e.g., a mobile phone, robot, home appliance, etc.).

[0111] The specific processing unit 290 sends the result of the specific processing to the smart glasses 214. In the smart glasses 214, the control unit 46A causes the speaker 240 to output the result of the specific processing. The microphone 238 acquires voice input representing the user's input to the specific processing result. The control unit 46A sends the voice data representing the user's input acquired by the microphone 238 to the data processing device 12. In the data processing device 12, the specific processing unit 290 acquires the voice data.

[0112] The data generation model 58 is a so-called generative AI. An example of the data generation model 58 includes generative AIs such as ChatGPT. The data generation model 58 is obtained by performing deep learning on a neural network. The data generation model 58 is input with a prompt containing instructions, and also with inference data such as speech data representing speech, text data representing text, and image data representing images (e.g., data of still images or data of moving images). The data generation model 58 infers the inference data based on the instructions shown in the prompt and outputs the inference result in one or more data forms such as speech data, text data, and image data. The data generation model 58 includes, for example, text generation AI, image generation AI, and multimodal generation AI. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization. The specific processing unit 290 performs the aforementioned specific processing while using the data generation model 58. The data generation model 58 can also be a fine-tuned model to output inference results from prompts that do not contain instructions; in this case, the data generation model 58 is able to output inference results from prompts that do not contain instructions. The data processing device 12, etc., includes various data generation models 58, which include AI other than generative AI. These AIs include, for example, linear regression, logistic regression, decision trees, random forests, support vector machines (SVM), k-means clustering, convolutional neural networks (CNN), recurrent neural networks (RNN), generative adversarial networks (GAN), or Naive Bayes, and are capable of various processing methods, but are not limited to these examples. Furthermore, AI can also be an AI agent. Additionally, when the processing described above is performed by AI, this processing may be partially or entirely performed by AI, but is not limited to these examples. Moreover, processing performed by AI including generative AI can be replaced by rule-based processing, and vice versa.

[0113] The data processing system 210 of the second embodiment performs the same processing as the data processing system 10 of the first embodiment. The processing performed by the data processing system 210 is executed by the specific processing unit 290 of the data processing device 12 or the control unit 46A of the smart glasses 214, but it can also be executed jointly by the specific processing unit 290 of the data processing device 12 and the control unit 46A of the smart glasses 214. Furthermore, the specific processing unit 290 of the data processing device 12 acquires or collects information required for processing from the smart glasses 214 or external devices, and the smart glasses 214 acquires or collects information required for processing from the data processing device 12 or external devices.

[0114] Each of the aforementioned elements, including a motion detection unit, an analysis unit, a notification unit, a fall detection unit, an instruction unit, and a health monitoring unit, can be implemented, for example, in at least one of the smart glasses 214 and the data processing device 12. For example, the motion detection unit uses the motion sensor of the smart glasses 214 to detect the elderly person's movements. The analysis unit, for example, is implemented by the specific processing unit 290 of the data processing device 12, which analyzes the motion data. The notification unit, for example, is implemented by the control unit 46A of the smart glasses 214, which notifies family members or caregivers when an anomaly is detected. The fall detection unit, for example, uses the accelerometer of the smart glasses 214 to detect falls in the elderly person. The instruction unit, for example, is implemented by the specific processing unit 290 of the data processing device 12, which instructs emergency response. The health monitoring unit, for example, uses the medical monitoring system of the smart glasses 214 to monitor the elderly person's health status. The correspondence between the various units and the device or control unit is not limited to the above examples and can be modified in various ways.

[0115] [Third Implementation] Figure 5 An example of the configuration of the data processing system 310 according to the third embodiment is shown.

[0116] like Figure 5 As shown, the data processing system 310 includes a data processing device 12 and a head-mounted terminal 314. An example of the data processing device 12 is a server.

[0117] The data processing device 12 includes a computer 22, a database 24, and a communication I / F 26. The computer 22 includes a processor 28, RAM 30, and memory 32. The processor 28, RAM 30, and memory 32 are connected to a bus 34. Furthermore, the database 24 and the communication I / F 26 are also connected to the bus 34. The communication I / F 26 is connected to a network 54. An example of the network 54 includes a WAN and / or LAN, etc.

[0118] The head-mounted terminal 314 includes a computer 36, a microphone 238, a speaker 240, a camera 42, a communication I / F 44, and a display 343. The computer 36 includes a processor 46, RAM 48, and memory 50. The processor 46, RAM 48, and memory 50 are connected to a bus 52. Furthermore, the microphone 238, speaker 240, camera 42, and display 343 are also connected to the bus 52.

[0119] Microphone 238 receives user commands by receiving the user's voice. Microphone 238 captures the user's voice, converts the captured sound into speech data, and outputs it to processor 46. Speaker 240 outputs sound according to commands from processor 46.

[0120] Camera 42 is a small digital camera equipped with an optical system such as a lens, aperture and shutter, and an imaging element such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor, used to capture the user's surroundings (e.g., the shooting range defined by an angle of view equivalent to the field of vision of an average healthy person).

[0121] Communication I / F 44 is connected to network 54. Communication I / F 44 and 26 are responsible for the transmission and reception of various information between processor 46 and processor 28 via network 54. The transmission and reception of various information between processor 46 and processor 28 using communication I / F 44 and 26 is performed in a secure state.

[0122] Figure 6 An example of the main functions of the data processing device 12 and the head-mounted terminal 314 is shown. Figure 6 As shown, in the data processing device 12, specific processing is performed by the processor 28. The memory 32 stores a specific processing program 56.

[0123] The processor 28 reads a specific processing program 56 from the memory 32 and executes the read specific processing program 56 on the RAM 30. The specific processing is implemented by the processor 28 operating as a specific processing unit 290 based on the specific processing program 56 executed on the RAM 30.

[0124] The memory 32 stores a data generation model 58 and an emotion-specific model 59. The data generation model 58 and the emotion-specific model 59 are used by the specific processing unit 290. The specific processing unit 290 can use the emotion-specific model 59 to infer the user's emotions and perform specific processing based on the user's emotions. The emotion inference function using the emotion-specific model 59 (emotion-specific function) includes various inferences and predictions related to the user's emotions, but is not limited to this example. Furthermore, emotion inference and prediction also include, for example, emotion analysis (analysis).

[0125] In the head-mounted terminal 314, specific processing is performed by the processor 46. A specific program 60 is stored in the memory 50. The processor 46 reads the specific program 60 from the memory 50 and executes the read specific program 60 on the RAM 48. Specific processing is implemented by the processor 46 operating as a control unit 46A based on the specific program 60 executed on the RAM 48. Furthermore, the head-mounted terminal 314 may also have the same data generation model and emotion-specific model as the data generation model 58 and the emotion-specific model 59, and use these models to perform the same processing as the specific processing unit 290.

[0126] Furthermore, other devices besides the data processing device 12 may also have the data generation model 58. For example, a server device may have the data generation model 58. In this case, the data processing device 12 obtains the processing results (prediction results, etc.) using the data generation model 58 by communicating with the server device that has the data generation model 58. In addition, the data processing device 12 may be a server device or a user-owned terminal device (e.g., a mobile phone, robot, home appliance, etc.).

[0127] The specific processing unit 290 sends the result of the specific processing to the head-mounted terminal 314. In the head-mounted terminal 314, the control unit 46A causes the speaker 240 and display 343 to output the result of the specific processing. The microphone 238 acquires voice representing the user's input to the specific processing result. The control unit 46A sends the voice data representing the user's input acquired by the microphone 238 to the data processing device 12. In the data processing device 12, the specific processing unit 290 acquires the voice data.

[0128] The data generation model 58 is a so-called generative AI. An example of the data generation model 58 includes generative AIs such as ChatGPT. The data generation model 58 is obtained by performing deep learning on a neural network. The data generation model 58 is input with a prompt containing instructions, and also with inference data such as speech data representing speech, text data representing text, and image data representing images (e.g., data of still images or data of moving images). The data generation model 58 infers the inference data based on the instructions shown in the prompt and outputs the inference result in one or more data forms such as speech data, text data, and image data. The data generation model 58 includes, for example, text generation AI, image generation AI, and multimodal generation AI. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization. The specific processing unit 290 performs the aforementioned specific processing while using the data generation model 58. The data generation model 58 can also be a fine-tuned model to output inference results from prompts that do not contain instructions; in this case, the data generation model 58 is able to output inference results from prompts that do not contain instructions. The data processing device 12, etc., includes various data generation models 58, which include AI other than generative AI. These AIs include, for example, linear regression, logistic regression, decision trees, random forests, support vector machines (SVM), k-means clustering, convolutional neural networks (CNN), recurrent neural networks (RNN), generative adversarial networks (GAN), or Naive Bayes, and are capable of various processing methods, but are not limited to these examples. Furthermore, AI can also be an AI agent. Additionally, when the processing described above is performed by AI, this processing may be partially or entirely performed by AI, but is not limited to these examples. Moreover, processing performed by AI including generative AI can be replaced by rule-based processing, and vice versa.

[0129] The data processing system 310 of the third embodiment performs the same processing as the data processing system 10 of the first embodiment. The processing performed by the data processing system 310 is executed by the specific processing unit 290 of the data processing device 12 or the control unit 46A of the head-mounted terminal 314, but it can also be executed jointly by the specific processing unit 290 of the data processing device 12 and the control unit 46A of the head-mounted terminal 314. Furthermore, the specific processing unit 290 of the data processing device 12 acquires or collects information required for processing from the head-mounted terminal 314 or external devices, and the head-mounted terminal 314 acquires or collects information required for processing from the data processing device 12 or external devices.

[0130] Each of the aforementioned elements, including a motion detection unit, an analysis unit, a notification unit, a fall detection unit, an instruction unit, and a health monitoring unit, can be implemented, for example, in at least one of the head-mounted terminal 314 and the data processing device 12. For example, the motion detection unit uses the motion sensor of the head-mounted terminal 314 to detect the elderly person's movements. The analysis unit, for example, is implemented by the specific processing unit 290 of the data processing device 12, which analyzes the motion data. The notification unit, for example, is implemented by the control unit 46A of the head-mounted terminal 314, which notifies family members or caregivers when an abnormality is detected. The fall detection unit, for example, uses the accelerometer of the head-mounted terminal 314 to detect falls in the elderly person. The instruction unit, for example, is implemented by the specific processing unit 290 of the data processing device 12, which instructs emergency response. The health monitoring unit, for example, uses the medical monitoring system of the head-mounted terminal 314 to monitor the elderly person's health status. The correspondence between the various units and the device or control unit is not limited to the above examples and can be modified in various ways.

[0131] [Fourth Implementation] Figure 7 An example of the configuration of the data processing system 410 according to the fourth embodiment is shown.

[0132] like Figure 7 As shown, the data processing system 410 includes a data processing device 12 and a robot 414. An example of the data processing device 12 is a server.

[0133] The data processing device 12 includes a computer 22, a database 24, and a communication I / F 26. The computer 22 includes a processor 28, RAM 30, and memory 32. The processor 28, RAM 30, and memory 32 are connected to a bus 34. Furthermore, the database 24 and the communication I / F 26 are also connected to the bus 34. The communication I / F 26 is connected to a network 54. An example of the network 54 includes a WAN and / or LAN, etc.

[0134] Robot 414 includes a computer 36, a microphone 238, a speaker 240, a camera 42, a communication I / F 44, and a control object 443. The computer 36 includes a processor 46, RAM 48, and memory 50. The processor 46, RAM 48, and memory 50 are connected to a bus 52. Furthermore, the microphone 238, speaker 240, camera 42, and control object 443 are also connected to the bus 52.

[0135] Microphone 238 receives user commands by receiving the user's voice. Microphone 238 captures the user's voice, converts the captured sound into speech data, and outputs it to processor 46. Speaker 240 outputs sound according to commands from processor 46.

[0136] The camera 42 is a small digital camera equipped with an optical system such as a lens, aperture and shutter, and an imaging element such as a CMOS image sensor or a CCD image sensor, used to photograph the user's surroundings (e.g., the shooting range defined by an angle of view equivalent to the field of vision of an average healthy person).

[0137] Communication I / F 44 is connected to network 54. Communication I / F 44 and 26 are responsible for the transmission and reception of various information between processor 46 and processor 28 via network 54. The transmission and reception of various information between processor 46 and processor 28 using communication I / F 44 and 26 is performed in a secure state.

[0138] The controlled object 443 includes a display device, LEDs for the eyes, and motors for driving the arms, hands, and feet. The posture and movements of the robot 414 are controlled by controlling the motors for the arms, hands, and feet. Some of the robot 414's emotions can be expressed by controlling these motors. In addition, facial expressions of the robot 414 can also be expressed by controlling the illumination state of the LEDs for the robot 414's eyes.

[0139] Figure 8 An example of the main functions of the data processing device 12 and the robot 414 is shown. For example... Figure 8 As shown, in the data processing device 12, specific processing is performed by the processor 28. The memory 32 stores a specific processing program 56.

[0140] The processor 28 reads a specific processing program 56 from the memory 32 and executes the read specific processing program 56 on the RAM 30. The specific processing is implemented by the processor 28 operating as a specific processing unit 290 based on the specific processing program 56 executed on the RAM 30.

[0141] The memory 32 stores a data generation model 58 and an emotion-specific model 59. The data generation model 58 and the emotion-specific model 59 are used by the specific processing unit 290. The specific processing unit 290 can use the emotion-specific model 59 to infer the user's emotions and perform specific processing based on the user's emotions. The emotion inference function using the emotion-specific model 59 (emotion-specific function) includes various inferences and predictions related to the user's emotions, but is not limited to this example. Furthermore, emotion inference and prediction also include, for example, emotion analysis (analysis).

[0142] In robot 414, specific processing is performed by processor 46. A specific program 60 is stored in memory 50. Processor 46 reads the specific program 60 from memory 50 and executes the read specific program 60 on RAM 48. Specific processing is achieved by processor 46 acting as control unit 46A based on the specific program 60 executed on RAM 48. Furthermore, robot 414 may also have the same data generation model and emotion-specific model as data generation model 58 and emotion-specific model 59, and use these models to perform the same processing as specific processing unit 290.

[0143] Furthermore, other devices besides the data processing device 12 may also have the data generation model 58. For example, a server device may have the data generation model 58. In this case, the data processing device 12 obtains the processing results (prediction results, etc.) using the data generation model 58 by communicating with the server device that has the data generation model 58. In addition, the data processing device 12 may be a server device or a user-owned terminal device (e.g., a mobile phone, robot, home appliance, etc.).

[0144] The specific processing unit 290 sends the result of the specific processing to the robot 414. In the robot 414, the control unit 46A causes the speaker 240 and the controlled object 443 to output the result of the specific processing. The microphone 238 acquires voice representing the user's input regarding the result of the specific processing. The control unit 46A sends the voice data representing the user input acquired by the microphone 238 to the data processing device 12. In the data processing device 12, the specific processing unit 290 acquires the voice data.

[0145] The data generation model 58 is a so-called generative AI. An example of the data generation model 58 includes generative AIs such as ChatGPT. The data generation model 58 is obtained by performing deep learning on a neural network. The data generation model 58 is input with a prompt containing instructions, and also with inference data such as speech data representing speech, text data representing text, and image data representing images (e.g., data of still images or data of moving images). The data generation model 58 infers the inference data based on the instructions shown in the prompt and outputs the inference result in one or more data forms such as speech data, text data, and image data. The data generation model 58 includes, for example, text generation AI, image generation AI, and multimodal generation AI. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization. The specific processing unit 290 performs the aforementioned specific processing while using the data generation model 58. The data generation model 58 can also be a fine-tuned model to output inference results from prompts that do not contain instructions; in this case, the data generation model 58 is able to output inference results from prompts that do not contain instructions. The data processing device 12, etc., includes various data generation models 58, which include AI other than generative AI. These AIs include, for example, linear regression, logistic regression, decision trees, random forests, support vector machines (SVM), k-means clustering, convolutional neural networks (CNN), recurrent neural networks (RNN), generative adversarial networks (GAN), or Naive Bayes, and are capable of various processing methods, but are not limited to these examples. Furthermore, AI can also be an AI agent. Additionally, when the processing described above is performed by AI, this processing may be partially or entirely performed by AI, but is not limited to these examples. Moreover, processing performed by AI including generative AI can be replaced by rule-based processing, and vice versa.

[0146] The data processing system 410 of the fourth embodiment performs the same processing as the data processing system 10 of the first embodiment. The processing performed by the data processing system 410 is executed by the specific processing unit 290 of the data processing device 12 or the control unit 46A of the robot 414, but it can also be executed jointly by the specific processing unit 290 of the data processing device 12 and the control unit 46A of the robot 414. Furthermore, the specific processing unit 290 of the data processing device 12 acquires or collects information required for processing from the robot 414 or external devices, and the robot 414 acquires or collects information required for processing from the data processing device 12 or external devices.

[0147] Each of the aforementioned elements, including a motion detection unit, an analysis unit, a notification unit, a fall detection unit, an instruction unit, and a health monitoring unit, can be implemented, for example, in at least one of the robot 414 and the data processing device 12. For example, the motion detection unit uses the motion sensors of the robot 414 to detect the elderly person's movements. The analysis unit, for example, is implemented by a specific processing unit 290 of the data processing device 12, which analyzes the motion data. The notification unit, for example, is implemented by the control unit 46A of the robot 414, which notifies family members or caregivers when an anomaly is detected. The fall detection unit, for example, uses the accelerometer of the robot 414 to detect falls in the elderly person. The instruction unit, for example, is implemented by the specific processing unit 290 of the data processing device 12, which instructs emergency response. The health monitoring unit, for example, uses the medical monitoring system of the robot 414 to monitor the elderly person's health status. The correspondence between the various units and the device or control unit is not limited to the above examples and can be modified in various ways.

[0148] Furthermore, the emotion-specific model 59, serving as an emotion engine, can determine the user's emotion based on a specific mapping. Specifically, the emotion-specific model 59 can determine the user's emotion based on an emotion graph that serves as a specific mapping (see...). Figure 9 The robot's emotions can be determined by the emotion-specific model 59. In addition, the emotion-specific model 59 can also determine the robot's emotions in the same way, and the specific processing unit 290 can also perform specific processing using the robot's emotions.

[0149] Figure 9 This is a diagram representing an emotion map 400 that maps various emotions. In the emotion map 400, emotions are arranged radially from the center in concentric circles. The closer to the center of the concentric circles, the more primitive the emotion is. Further out on the concentric circles, emotions are arranged representing states or actions arising from mood. Emotion is a concept that includes both feelings and mental states. To the left of the concentric circles, emotions generated by reactions occurring in the brain are arranged roughly. To the right of the concentric circles, emotions guided by situational judgments are arranged roughly. Above and below the concentric circles, emotions generated by reactions occurring in the brain and guided by situational judgments are arranged roughly. Furthermore, the emotion of "pleasure" is arranged above the concentric circles, and the emotion of "unpleasantness" is arranged below. Thus, in the emotion map 400, various emotions are mapped according to the structure of emotion generation, while easily generated emotions are mapped nearby.

[0150] These emotions are distributed at the 3 o'clock position on the Emotion Chart 400, and usually fluctuate between peace and unease. In the right half of the Emotion Chart 400, because situational awareness is more dominant than internal feelings, it gives a sense of calm.

[0151] The inner side of the emotion diagram 400 represents the mind, and the outer side of the emotion diagram 400 represents actions. Therefore, the further you go to the outer side of the emotion diagram 400, the more the emotion can be seen (manifested in actions).

[0152] Here, human emotions are based on a balance of various factors such as posture and blood sugar levels. When these balances deviate from the ideal, it indicates unhappiness; when they approach the ideal, it indicates pleasure. In robots, cars, and motorcycles, emotions can also be created based on a balance of factors such as posture and remaining battery power. When these balances deviate from the ideal, it indicates unhappiness; when they approach the ideal, it indicates pleasure. Emotion maps can be generated, for example, based on Dr. Mitsuyoshi's emotion map (Research on Speech Emotion Recognition and Brain Physiological Signal Analysis Systems for Emotion, Tokushima University, Doctoral Dissertation: https: / / ci.nii.ac.jp / naid / 500000375379). In the left half of the emotion map, emotions belonging to the "Reaction" domain, where sensation is dominant, are arranged. Furthermore, in the right half of the emotion map, emotions belonging to the "Situation" domain, where situational cognition is dominant, are arranged.

[0153] The emotion map defines two types of emotions that promote learning. One is a negative emotion located near the middle of "repentance" or "reflection" on the situation side. That is, when the robot experiences negative emotions such as "I never want to feel this way again" or "I never want to be scolded again." The other is a positive emotion located near "desire" on the response side. That is, when the robot experiences positive feelings such as "wanting more" or "wanting to know more."

[0154] The emotion-specific model 59 feeds user input into a pre-learned neural network to obtain emotion values ​​representing each emotion shown in the emotion graph 400, and determines the user's emotion. This neural network is pre-learned based on multiple learning data sets that combine user input with emotion values ​​representing each emotion shown in the emotion graph 400. Furthermore, this neural network is learned to... Figure 10 As shown in sentiment graph 900, sentiment values ​​in nearby configurations are similar to each other. Figure 10 Examples show that multiple emotions such as "peace of mind", "stability", and "reassurance" have similar emotional values.

[0155] In the above embodiments, a specific processing is described by a single computer 22, but the technology disclosed herein is not limited to this, and distributed processing by multiple computers, including computer 22, is also possible.

[0156] In the above embodiments, an example of storing a specific processing program 56 in memory 32 is illustrated, but the technology disclosed herein is not limited thereto. For example, the specific processing program 56 may also be stored in a portable computer-readable non-transitory storage medium such as a USB (Universal Serial Bus) memory. The specific processing program 56 stored in the non-transitory storage medium is installed in the computer 22 of the data processing device 12. The processor 28 performs specific processing according to the specific processing program 56.

[0157] Alternatively, the specific processing program 56 can be stored in a storage device such as a server connected to the data processing device 12 via a network 54, and the specific processing program 56 can be downloaded and installed into the computer 22 upon request from the data processing device 12.

[0158] Furthermore, it is not necessary to store the entire specific process 56 in a storage device such as a server connected to the data processing device 12 via the network 54, nor is it necessary to store the entire specific process 56 in the memory 32; a portion of the specific process 56 may also be stored.

[0159] As a hardware resource for performing specific processing, various processors can be used. For example, a CPU is a general-purpose processor that functions as a hardware resource for performing specific processing by executing software, i.e., programs. Additionally, a dedicated circuit can be listed as a processor; it is a processor with a circuit structure specifically designed for performing specific processing, such as a FPGA (Field-Programmable Gate Array), PLD (Programmable Logic Device), or ASIC (Application Specific Integrated Circuit). Every processor has built-in or connected memory, and every processor executes specific processing by using that memory.

[0160] The hardware resources for performing a specific process can consist of one of these various processors, or a combination of two or more processors of the same or different types (e.g., a combination of multiple FPGAs, or a combination of a CPU and an FPGA). Furthermore, the hardware resources for performing a specific process can also be a single processor.

[0161] As an example of a single processor, the first type consists of a combination of one or more CPUs and software, which functions as a hardware resource to perform specific processing. The second type uses a processor, such as a System-on-a-chip (SoC), which implements the entire system functionality, including multiple hardware resources performing specific processing, using a single IC chip. In this case, the specific processing is implemented using one or more of the aforementioned processors that serve as hardware resources.

[0162] Furthermore, as the hardware architecture of these various processors, more specifically, circuits combining semiconductor elements and other circuit components can be used. Moreover, the specific process described above is merely an example. Therefore, it goes without saying that, without departing from the main point, unnecessary steps can be removed, new steps can be added, or the processing order can be changed.

[0163] Furthermore, although the above examples have been described in terms of first to fourth embodiments, some or all of these embodiments can also be combined. Additionally, the smart device 14, smart glasses 214, head-mounted terminal 314, and robot 414 are just examples; they can be combined separately or are other devices.

[0164] The foregoing descriptions and illustrations are detailed explanations of the parts covered by this disclosure and are merely one example of this disclosure. For instance, the descriptions of the above-described structure, function, role, and effect are just one example of the structure, function, role, and effect of the parts covered by this disclosure. Therefore, it goes without saying that, without departing from the spirit of this disclosure, unnecessary parts can be deleted, new elements can be added, or replacements can be made to the foregoing descriptions and illustrations. Furthermore, to avoid confusion and facilitate understanding of the parts covered by this disclosure, explanations of technical common sense that does not require special explanation for implementing this disclosure have been omitted from the foregoing descriptions and illustrations.

[0165] All documents, patent applications and technical standards described in this specification are incorporated herein by reference as if they were specifically and individually described as incorporated by reference.

[0166] (Note 1) A system, characterized in that it comprises: The motion detection unit is used to detect motion. The analysis unit is used to analyze the information detected by the motion detection unit; The notification unit is used to send notifications based on the information parsed by the parsing unit. Fall detection unit, used to detect falls; The analysis unit is also used to analyze the information detected by the fall detection unit; The instruction unit is used to instruct emergency response based on the information parsed by the parsing unit; The health monitoring department is used to monitor health conditions. The analysis unit is also used to analyze the information monitored by the health monitoring unit; The notification unit is also used to send notifications based on the information parsed by the parsing unit.

[0167] (Note 2) The system as described in Appendix 1 is characterized in that the motion detection unit is located in various parts of the room and is capable of monitoring the movements of the elderly in real time.

[0168] (Note 3) The system as described in Appendix 1 is characterized in that the fall detection unit is worn on the body of the elderly person and can detect a fall immediately.

[0169] (Note 4) The system as described in Appendix 1 is characterized in that the health monitoring unit is capable of measuring the elderly person's body temperature, heart rate, and blood pressure in real time.

[0170] (Note 5) The system as described in Appendix 1 is characterized in that the parsing unit parses information from the motion detection unit, the fall detection unit, and the health monitoring unit, and notifies family members or caregivers when an abnormality is detected.

[0171] (Note 6) The system as described in Appendix 1 is characterized in that the notification unit notifies family members or caregivers based on information analyzed by AI.

[0172] (Note 7) The system as described in Appendix 1 is characterized in that the instruction unit instructs emergency response based on information analyzed by AI.

[0173] (Note 8) The system as described in Appendix 1 is characterized in that the notification unit has the function of periodically reporting the elderly person's condition to family members or caregivers via AI.

[0174] (Note 9) The system as described in Appendix 1 is characterized in that the notification unit has the function of enabling family members or caregivers to communicate directly with the elderly in emergency situations.

[0175] (Postscript 10) The system as described in Appendix 1 is characterized in that the motion detection unit is able to infer the emotions of the elderly person and adjust the sensitivity of motion detection according to the inferred emotions.

[0176] (Postscript 11) The system as described in Appendix 1 is characterized in that the motion detection unit analyzes the past motion patterns of the elderly to improve the detection accuracy of abnormal motions.

[0177] (Postscript 12) The system as described in Appendix 1 is characterized in that the motion detection unit optimizes the detection accuracy based on the environmental conditions of the room during motion detection.

[0178] (Postscript 13) The system as described in Appendix 1 is characterized in that the motion detection unit is able to infer the emotions of the elderly person and determine the priority of motion detection based on the inferred emotions.

[0179] (Postscript 14) The system as described in Appendix 1 is characterized in that, during motion detection, the motion detection unit adjusts the detection range based on the furniture layout of the room.

[0180] (Postscript 15) The system as described in Appendix 1 is characterized in that the motion detection unit filters the pet's motion during motion detection to prevent false detection.

[0181] (Postscript 16) The system as described in Appendix 1 is characterized in that the parsing unit is able to infer the emotions of the elderly and adjust the parsing algorithm according to the inferred emotions.

[0182] (Postscript 17) The system as described in Appendix 1 is characterized in that the parsing unit refers to past anomaly detection data during parsing to improve parsing accuracy.

[0183] (Postscript 18) The system as described in Appendix 1 is characterized in that, during parsing, the parsing unit adjusts the level of detail of the parsing according to the frequency of anomaly detection.

[0184] (Postscript 19) The system as described in Appendix 1 is characterized in that the analysis unit is able to infer the emotions of the elderly and adjust the display method of the analysis results according to the inferred emotions.

[0185] (Postscript 20) The system as described in Appendix 1 is characterized in that, during the parsing process, the parsing unit applies different parsing algorithms according to the type of anomaly detection.

[0186] (Postscript 21) The system as described in Appendix 1 is characterized in that, during parsing, the parsing unit determines the parsing priority based on the time period of anomaly detection.

[0187] (Postscript 22) The system as described in Appendix 1 is characterized in that the notification unit is able to infer the emotions of the elderly and adjust the notification content according to the inferred emotions.

[0188] (Postscript 23) The system as described in Appendix 1 is characterized in that the notification unit optimizes the notification method by referring to the past reaction records of family members or caregivers when issuing notifications.

[0189] (Postscript 24) The system as described in Appendix 1 is characterized in that the notification unit adjusts the urgency of the notification according to the type of abnormality when issuing the notification.

[0190] (Postscript 25) The system as described in Appendix 1 is characterized in that the notification unit is able to infer the emotions of the elderly person and adjust the timing of the notification based on the inferred emotions.

[0191] (Postscript 26) The system as described in Appendix 1 is characterized in that, when making a notification, the notification unit considers the geographical location information of the family member or caregiver to select the optimal notification method.

[0192] (Postscript 27) The system as described in Appendix 1 is characterized in that the notification unit adjusts the form of the notification based on the device information of family members or caregivers when issuing the notification.

[0193] (Postscript 28) The system as described in Appendix 1 is characterized in that the fall detection unit is able to infer the emotions of the elderly person and adjust the sensitivity of the fall detection based on the inferred emotions.

[0194] (Postscript 29) The system as described in Appendix 1 is characterized in that the fall detection unit analyzes the elderly person's past fall records to improve the accuracy of fall detection.

[0195] (Note 30) The system as described in Appendix 1 is characterized in that the fall detection unit optimizes the detection accuracy based on the elderly person's physical condition or clothing during fall detection.

[0196] (Postscript 31) The system as described in Appendix 1 is characterized in that the fall detection unit is able to infer the emotions of the elderly person and determine the priority of fall detection based on the inferred emotions.

[0197] (Note 32) The system as described in Appendix 1 is characterized in that the fall detection unit adjusts the detection accuracy based on the material or condition of the floor during fall detection.

[0198] (Postscript 33) The system as described in Appendix 1 is characterized in that, during fall detection, the fall detection unit indicates different response measures based on the degree of impact of the fall.

[0199] (Postscript 34) The system as described in Appendix 1 is characterized in that the instruction unit is capable of inferring the emotions of the elderly person and adjusting the instructions for emergency response based on the inferred emotions.

[0200] (Postscript 35) The system as described in Appendix 1 is characterized in that, in the event of an emergency, the instruction unit selects the optimal response method by referring to past response records.

[0201] (Postscript 36) The system as described in Appendix 1 is characterized in that, in emergency response, the instruction unit adjusts the level of detail of the response according to the type of anomaly.

[0202] (Postscript 37) The system as described in Appendix 1 is characterized in that the instruction unit is able to infer the emotions of the elderly person and determine the priority of emergency response based on the inferred emotions.

[0203] (Postscript 38) The system as described in Appendix 1 is characterized in that, in the event of an emergency, the instruction unit considers the geographical location information of family members or caregivers to select the optimal response method.

[0204] (Postscript 39) The system as described in Appendix 1 is characterized in that, in the event of an emergency, the instruction unit adjusts the response method based on the equipment information of family members or caregivers.

[0205] (Postscript 40) The system as described in Appendix 1 is characterized in that the health monitoring unit is able to infer the emotions of the elderly and adjust the sensitivity of the health monitoring based on the inferred emotions.

[0206] (Postscript 41) The system as described in Appendix 1 is characterized in that the health monitoring unit analyzes the elderly person's past health data to improve the accuracy of anomaly detection.

[0207] (Postscript 42) The system as described in Appendix 1 is characterized in that the health monitoring unit optimizes the monitoring accuracy based on the influence of seasons or weather during health monitoring.

[0208] (Postscript 43) The system as described in Appendix 1 is characterized in that the health monitoring unit is able to infer the emotions of the elderly and determine the priority of health monitoring based on the inferred emotions.

[0209] (Postscript 44) The system as described in Appendix 1 is characterized in that, during health monitoring, the health monitoring unit improves the accuracy of abnormality detection based on the health data of family members or caregivers.

[0210] (Postscript 45) The system as described in Appendix 1 is characterized in that the health monitoring unit applies different monitoring algorithms according to the type of abnormality during health monitoring.

Claims

1. A system, characterized in that, include: The motion detection unit is used to detect motion. The analysis unit is used to analyze the information detected by the motion detection unit; The notification unit is used to send notifications based on the information parsed by the parsing unit. Fall detection unit, used to detect falls; The analysis unit is also used to analyze the information detected by the fall detection unit; The instruction unit is used to instruct emergency response based on the information parsed by the parsing unit; The health monitoring department is used to monitor health conditions. The analysis unit is also used to analyze the information monitored by the health monitoring unit; The notification unit is also used to send notifications based on the information parsed by the parsing unit.

2. The system as described in claim 1, characterized in that, The motion detection unit is installed in various parts of the room and can monitor the elderly person's movements in real time.

3. The system as described in claim 1, characterized in that, The fall detection device is worn on the body of the elderly and can detect falls immediately.

4. The system as described in claim 1, characterized in that, The health monitoring unit can measure the elderly person's body temperature, heart rate, and blood pressure in real time.

5. The system as described in claim 1, characterized in that, The analysis unit analyzes information from the motion detection unit, the fall detection unit, and the health monitoring unit, and notifies family members or caregivers when an abnormality is detected.

6. The system as described in claim 1, characterized in that, The notification department uses AI-analyzed information to notify family members or caregivers.

7. The system as described in claim 1, characterized in that, The instruction unit provides instructions for emergency response based on information analyzed by AI.

8. The system as described in claim 1, characterized in that, The notification system has the function of using AI to regularly report the elderly person's condition to family members or caregivers.

9. The system as described in claim 1, characterized in that, The notification system enables family members or caregivers to communicate directly with the elderly in emergency situations.

10. The system as claimed in claim 1, characterized in that, The motion detection unit can infer the emotions of the elderly and adjust the sensitivity of motion detection based on the inferred emotions.

Images