Sleep monitoring device for old-age care based on internet of things and monitoring management system

Abstract

This invention discloses an IoT-based sleep monitoring device and management system for elderly healthcare, belonging to the field of sleep monitoring technology. The IoT-based sleep monitoring device for elderly healthcare includes a sleep monitor connected to a sleep monitoring belt via a data cable. The sleep monitoring belt is equipped with a sleep monitoring module, which monitors the user's sleep in real time, collects multi-source sleep data, and transmits the collected data to a cloud platform via an IoT communication module. This invention solves the problem that existing methods cannot achieve real-time monitoring and health risk warning management of the elderly, thus reducing the sleep quality and health level of the elderly population. This invention enables real-time monitoring and health risk warning management of the elderly's sleep, improving their sleep quality and health level.

Description

Technical Field

[0001] This invention relates to the field of sleep monitoring technology, specifically to a sleep monitoring device and monitoring management system for elderly health care based on the Internet of Things. Background Technology

[0002] Elderly healthcare refers to a comprehensive set of measures that guide the elderly to maintain their health and slow down aging through scientific methods. Among these measures, sleep quality is a crucial indicator in the healthcare process for the elderly. Therefore, it is essential to monitor the sleep patterns of the elderly during healthcare care to facilitate effective management.

[0003] The existing methods cannot achieve real-time monitoring of the sleep status of the elderly and early warning management of health risks, which reduces the sleep quality and health level of the elderly population. Summary of the Invention

[0004] The purpose of this invention is to provide an Internet of Things-based sleep monitoring device and management system for elderly health care, which can realize real-time monitoring of the sleep status of the elderly and health risk early warning management, improve the sleep quality and health level of the elderly population, and solve the problems mentioned in the background art.

[0005] To achieve the above objectives, the present invention provides the following technical solution:

[0006] The IoT-based sleep monitoring device for elderly health care includes a sleep monitor connected to a sleep monitoring belt via a data cable. The sleep monitoring belt is equipped with a sleep monitoring module, which is used to monitor the user's sleep in real time and collect multi-source sleep data. The collected multi-source sleep data is then transmitted to a cloud platform via an IoT communication module.

[0007] Preferably, the sleep monitor is connected to a power plug via a power cord, and the power cord is connected to the power plug to provide power. After being powered on, the sleep monitor is charged. The sleep monitor is also equipped with an emergency call button and a motion-activated night light, with the motion-activated night light located to one side of the emergency call button.

[0008] According to another aspect of the present invention, an Internet of Things (IoT)-based sleep monitoring and management system for elderly healthcare is provided, comprising a sleep monitoring module of the IoT-based sleep monitoring device for elderly healthcare as described above, and further comprising:

[0009] The data processing module is used to process multi-source sleep data of users and extract features related to sleep monitoring and management to determine the user's sleep characteristic data;

[0010] The intelligent analysis module is used to build a sleep monitoring anomaly detection model to analyze and identify user sleep characteristic data, judge the user's sleep quality, automatically identify abnormal situations in the user's sleep, and determine the user's sleep anomaly detection results.

[0011] The alarm management module is used to provide early warnings based on the results of abnormal sleep detection, ensuring the safety of the elderly.

[0012] Preferably, the sleep monitoring module collects multi-source sleep data from the user and performs the following operations:

[0013] Based on sensors, the system monitors the user's heart rate, respiratory rate, blood oxygen saturation, and body movement frequency in real time while the user is sleeping, and collects the user's sleep physiological data.

[0014] Based on sensors, the system monitors the temperature, humidity, light, and carbon dioxide concentration in real time while the user is sleeping, and collects data on the user's sleep environment.

[0015] Based on user sleep physiological data and user sleep environment data, multi-source user sleep data is generated.

[0016] Preferably, the user's multi-source sleep data is processed and features related to sleep monitoring and management are extracted, and the following operations are performed:

[0017] The user sleep multi-source data is processed based on Kalman filtering to remove high-frequency noise and clean the user sleep multi-source data to remove outliers.

[0018] Based on Z-score standardization, user sleep multi-source data is processed to normalize user sleep multi-source data with different scales, forming standardized user sleep multi-source data.

[0019] Principal component analysis was used to process multi-source sleep data of users, extracting features related to sleep monitoring and management from the multi-source sleep data of users, and determining user sleep characteristic data, including apnea index, apnea duration, heart rate variability, body movement fluctuation, REM sleep, deep sleep and light sleep.

[0020] Preferably, the system analyzes and identifies user sleep characteristic data to determine user sleep quality and automatically identifies abnormalities during user sleep, performing the following operations:

[0021] A sleep monitoring anomaly detection model was constructed and deployed in the user's sleep monitoring anomaly detection environment to detect anomalies in the user's sleep.

[0022] The system takes user sleep characteristic data as input data and feeds it into the sleep monitoring anomaly detection model. The model analyzes and identifies the user's sleep characteristic data, judges the user's sleep quality, automatically identifies abnormalities in the user's sleep, and outputs the user's sleep anomaly detection results.

[0023] Preferably, a sleep monitoring anomaly detection model is constructed, and the following operations are performed:

[0024] Historical sleep monitoring data was collected and divided into training and testing sets.

[0025] The machine learning model is trained using a training set, enabling the model to autonomously learn sleep monitoring anomaly detection behaviors from the training set, judge the user's sleep quality, and automatically identify abnormal situations in the user's sleep, thus determining the sleep monitoring anomaly detection model.

[0026] The sleep monitoring anomaly detection model was tested using a test set, and the model's performance was evaluated based on the test results to determine whether it could achieve the expected effect of automatically identifying abnormalities in the user's sleep.

[0027] When the sleep monitoring anomaly detection model fails to achieve the expected effect of automatically identifying abnormal situations in the user's sleep, the parameters of the sleep monitoring anomaly detection model are adjusted and optimized, and the optimized sleep monitoring anomaly detection model is tested until the sleep monitoring anomaly detection model can achieve the expected effect of automatically identifying abnormal situations in the user's sleep, and the optimal sleep monitoring anomaly detection model is determined.

[0028] Preferably, based on the user's sleep abnormality detection results, early warning management is implemented, and the following operations are performed:

[0029] When the user's sleep abnormality detection result indicates that the user's sleep is abnormal, an alarm message is automatically sent to family members and medical staff, and a sleep quality analysis report is generated based on the user's multi-source sleep data and displayed in a visual form.

[0030] Medical staff can view the sleep quality analysis reports of the elderly in real time through mobile terminals, and provide suggestions for improving sleep based on the sleep quality analysis reports and adjust the users' lifestyles in a timely manner to ensure the safety of the elderly.

[0031] Preferably, the user's multi-source sleep data is processed and features related to sleep monitoring and management are extracted, including:

[0032] Collect abnormal sleep history records of users, and calculate the sleep risk value of users for each sleep period based on the abnormal sleep history records;

[0033] The sleep risk value is compared with a preset sleep risk threshold. Sleep periods when the sleep risk value is greater than or equal to the preset sleep risk threshold are defined as high-risk sleep periods; sleep periods when the sleep risk value is less than the preset sleep risk threshold are defined as low-risk sleep periods.

[0034] The multi-source user sleep data with millisecond-level timestamp tags collected by the sensors is subjected to differential anomaly removal according to the high-risk sleep period and low-risk sleep period to obtain preprocessed multi-source user sleep data;

[0035] Based on a constructed database of suspected sleep features including typical characteristics of apnea, REM sleep, deep sleep, and light sleep;

[0036] The matching degree between the preprocessed user sleep multi-source data and the suspicious sleep feature database is calculated and used as the first evaluation value.

[0037] The score of the bicorrelation coefficient between the preprocessed user sleep multi-source data and historical normal data of the same period is calculated and used as the second evaluation value;

[0038] The first evaluation value and the second evaluation value are weighted and summed to obtain the feature credibility. Based on the feature credibility, the user's sleep period is divided into core feature period and auxiliary association period. A sleep period feature dataset is generated based on the core feature period, auxiliary association period and preprocessed user sleep multi-source data.

[0039] The standard correlation coefficient is determined based on the mean of the total correlation coefficients of normal data from the same historical period.

[0040] The differential threshold is determined based on the standard correlation coefficient and the preset time period coefficient;

[0041] A weighted score is calculated for the sleep period feature dataset. If the weighted score is greater than or equal to the differential threshold, the sleep apnea index, sleep apnea duration, heart rate variability, body motion fluctuation, REM sleep duration, deep sleep duration, and light sleep duration features are output.

[0042] Preferably, based on the user's sleep abnormality detection results, early warning management is implemented, and the following operations are performed:

[0043] Identify user sleep characteristic data and determine the weighted feature values ​​corresponding to the user sleep characteristic data;

[0044] ;

[0045] in, This represents the weighted feature value corresponding to the user's sleep feature data; This represents the k-th raw data collection value of the i-th type of sleep core indicator; This represents the k-th measurement value of the i-th type of sleep core indicator; This represents the mean of the core indicators for the i-th type of sleep. This represents the standard deviation of the core sleep indicator for the i-th type; The standard deviation of the apnea index; This represents the sensor interference correction coefficient, reflecting the impact of environmental noise and transmission delay on the data; To represent extremely small positive numbers to prevent the denominator from being 0; This represents the total number of core sleep indicator types; it is fixed at 7 (corresponding to: apnea index, apnea duration, heart rate variability, body movement fluctuation, percentage of REM sleep, percentage of deep sleep, and percentage of light sleep). This represents the total number of nighttime data collections for a single indicator.

[0046] The sleep anomaly detection results for users are determined based on a sleep monitoring anomaly detection model and weighted feature values; the sleep monitoring anomaly detection model is as follows:

[0047]

[0048] in, This represents a sleep monitoring abnormality detection model; This represents the multidimensional adaptation coefficient of sleep monitoring; ; , , Indicates scene weight; Represents the scene coefficient; Indicates the detection efficiency coefficient; Indicates the precision coefficient; , These represent adjustment parameters, which control accuracy and efficiency respectively.

[0049] Compared with the prior art, the beneficial effects of the present invention are:

[0050] This invention monitors users' physiological and environmental data in real time during sleep, collects multi-source sleep data, processes this data to extract features relevant to sleep monitoring and management, determines user sleep characteristic data, analyzes and identifies this data using a sleep monitoring anomaly detection model, assesses sleep quality, automatically identifies abnormalities in sleep, determines sleep anomaly detection results, and provides early warning management based on these results. This ensures the safety of the elderly, enabling real-time monitoring of their sleep status and health risk early warning management, and improving the sleep quality and health level of the elderly population. Attached Figure Description

[0051] Figure 1This is a schematic diagram of the Internet of Things-based sleep monitoring device for elderly health care according to the present invention;

[0052] Figure 2 This is a block diagram of the IoT-based sleep monitoring and management system for elderly healthcare of the present invention;

[0053] Figure 3 This is a flowchart of the IoT-based sleep monitoring and management system for elderly healthcare according to the present invention.

[0054] In the picture: 1. Sleep monitor; 11. Data cable; 12. Power cord; 13. Emergency call button; 14. Sensor night light; 2. Sleep monitoring belt; 3. Power plug. Detailed Implementation

[0055] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0056] To address the current limitations in real-time monitoring and health risk management of the elderly, which negatively impacts their sleep quality and overall health, please refer to [link to relevant documentation]. Figures 1-3 This embodiment provides the following technical solution:

[0057] The IoT-based sleep monitoring device for elderly healthcare includes a sleep monitor 1, which is connected to a sleep monitoring belt 2 via a data cable 11. The sleep monitoring belt 2 is equipped with a sleep monitoring module, which monitors the user's sleep in real time and collects multi-source sleep data. Based on an IoT communication module, the collected multi-source sleep data is transmitted to a cloud platform. The cloud platform stores, analyzes, and visualizes the collected multi-source sleep data, providing users with real-time monitoring and health management services. Specifically, storing the multi-source sleep data in the cloud utilizes big data and artificial intelligence technologies for trend analysis and anomaly detection. Once an anomaly (such as abnormal heart rate or sleep apnea) is detected, an alarm message is automatically sent to family members or medical personnel.

[0058] In this embodiment, the sleep monitor 1 is connected to a power plug 3 via a power cord 12. The power cord 12 is connected to the power plug 3 to provide power and charge the sleep monitor 1. The sleep monitor 1 is also equipped with an emergency call button 13 and a motion-activated night light 14. The motion-activated night light 14 is located to one side of the emergency call button 13. In case of an emergency, an emergency call can be made through the emergency call button 13.

[0059] To better demonstrate the IoT-based sleep monitoring and management process for elderly healthcare, this embodiment provides an IoT-based sleep monitoring and management system for elderly healthcare, including a sleep monitoring module, a data processing module, an intelligent analysis module, and an alarm management module of the IoT-based sleep monitoring device for elderly healthcare as described above.

[0060] In this embodiment, the sleep monitoring module collects multi-source sleep data from the user and performs the following operations:

[0061] Based on sensors, the system monitors the user's heart rate, respiratory rate, blood oxygen saturation, and body movement frequency in real time while the user is sleeping, and collects the user's sleep physiological data.

[0062] Specifically, the system collects user sleep physiological data based on sensors, as shown in Table 1:

[0063]

[0064] Based on sensors, the system monitors the temperature, humidity, light, and carbon dioxide concentration in real time while the user is sleeping, and collects data on the user's sleep environment.

[0065] Specifically, the system collects user sleep environment data based on sensors, as shown in Table 2:

[0066]

[0067] Based on user sleep physiological data and user sleep environment data, multi-source user sleep data is generated.

[0068] It should be noted that by monitoring users' physiological and environmental data in real time during sleep, multi-source sleep data is collected to facilitate subsequent analysis of users' sleep patterns and to promptly detect any abnormal sleep conditions.

[0069] The data processing module is used to process multi-source sleep data of users and extract features related to sleep monitoring and management to determine the user's sleep characteristic data.

[0070] In this embodiment, the user's multi-source sleep data is processed and features related to sleep monitoring and management are extracted. The following operations are performed:

[0071] The user sleep multi-source data is processed based on Kalman filtering to remove high-frequency noise and clean the user sleep multi-source data to remove outliers.

[0072] Based on Z-score standardization, user sleep multi-source data is processed to normalize user sleep multi-source data with different scales, forming standardized user sleep multi-source data.

[0073] Principal component analysis was used to process multi-source sleep data of users, extracting features related to sleep monitoring and management from the multi-source sleep data of users, and determining user sleep characteristic data, including apnea index, apnea duration, heart rate variability, body movement fluctuation, REM sleep, deep sleep and light sleep.

[0074] It should be noted that the apnea index is the number of apneas per hour, used to assess the severity of sleep apnea syndrome; the duration of apnea reflects the severity of the apnea event; heart rate variability is used to assess the regulatory function of the autonomic nervous system, and decreased heart rate variability may be associated with decreased sleep quality; body movement fluctuations are used to analyze the continuity and stability of sleep, and frequent body movements may indicate poor sleep quality; REM sleep refers to high-frequency brain activity and rapid eye movements; deep sleep refers to low-frequency, low-amplitude brain activity, with stable heart rate and respiration; light sleep refers to brain activity between REM sleep and deep sleep, with larger heart rate fluctuations.

[0075] The intelligent analysis module is used to analyze and identify user sleep characteristic data, determine the user's sleep quality, automatically identify abnormal situations during the user's sleep, and determine the user's sleep abnormality detection results.

[0076] In this embodiment, user sleep characteristic data is analyzed and identified to determine the user's sleep quality and automatically identify abnormalities during the user's sleep, and the following operations are performed:

[0077] A sleep monitoring anomaly detection model was constructed and deployed in the user's sleep monitoring anomaly detection environment to detect anomalies in the user's sleep.

[0078] The system takes user sleep characteristic data as input data and feeds it into the sleep monitoring anomaly detection model. The model analyzes and identifies the user's sleep characteristic data, judges the user's sleep quality, automatically identifies abnormalities in the user's sleep, and outputs the user's sleep anomaly detection results.

[0079] In this embodiment, a sleep monitoring anomaly detection model is constructed, and the following operations are performed:

[0080] Historical sleep monitoring data was collected and divided into training and testing sets.

[0081] The machine learning model is trained using a training set, enabling the model to autonomously learn sleep monitoring anomaly detection behaviors from the training set, judge the user's sleep quality, and automatically identify abnormal situations in the user's sleep, thus determining the sleep monitoring anomaly detection model.

[0082] The sleep monitoring anomaly detection model was tested using a test set, and the model's performance was evaluated based on the test results to determine whether it could achieve the expected effect of automatically identifying abnormalities in the user's sleep.

[0083] When the sleep monitoring anomaly detection model fails to achieve the expected effect of automatically identifying abnormal situations in the user's sleep, the parameters of the sleep monitoring anomaly detection model are adjusted and optimized, and the optimized sleep monitoring anomaly detection model is tested until the sleep monitoring anomaly detection model can achieve the expected effect of automatically identifying abnormal situations in the user's sleep, and the optimal sleep monitoring anomaly detection model is determined.

[0084] The alarm management module is used to provide early warnings based on the results of abnormal sleep detection, ensuring the safety of the elderly.

[0085] In this embodiment, early warning management is performed based on the user's sleep abnormality detection results, and the following operations are performed:

[0086] When the user's sleep abnormality detection result indicates that the user's sleep is abnormal, an alarm message is automatically sent to family members and medical staff, and a sleep quality analysis report is generated based on the user's multi-source sleep data and displayed in a visual form.

[0087] Medical staff can view the sleep quality analysis reports of the elderly in real time through mobile terminals, and provide suggestions for improving sleep based on the sleep quality analysis reports and adjust the users' lifestyles in a timely manner to ensure the safety of the elderly.

[0088] Among the suggestions for improving user sleep are:

[0089] 1) Lifestyle adjustments:

[0090] Try to go to bed and wake up at the same time every day, avoid excessive daytime naps, and keep naps to within 30 minutes. Keep the bedroom quiet and comfortable, use blackout curtains and soundproofing equipment, and adjust the room temperature to 18-22℃ and the humidity to 40%-60%.

[0091] 2) Dietary therapy:

[0092] Avoid greasy, spicy, and irritating foods for dinner, and avoid eating or drinking two hours before bedtime to prevent frequent nighttime urination. Drinking warm milk or honey milk before bed can help relax your mind and body.

[0093] 3) Moderate exercise:

[0094] Engaging in moderate exercise during the day (such as walking, Tai Chi, yoga, etc.) can help relax the mind and body and improve sleep quality. Avoid exercising close to bedtime to prevent over-excitement of the body.

[0095] 4) Psychological adjustment:

[0096] Relieve stress and anxiety by listening to soothing music and meditation, help the body relax, encourage older adults to participate in social activities, enrich their spiritual life, and reduce feelings of loneliness.

[0097] 5) Adjunctive therapies:

[0098] Acupuncture and cupping can promote blood circulation, relieve muscle tension, and improve sleep. Soaking your feet in warm water for 20 minutes can relax muscles and relieve fatigue.

[0099] In summary, by real-time monitoring of users' physiological and environmental data during sleep, collecting multi-source sleep data, processing this data and extracting features relevant to sleep monitoring and management, identifying user sleep characteristic data, and using a sleep monitoring anomaly detection model to analyze and identify user sleep characteristic data, the system can determine user sleep quality, automatically identify abnormalities in the user's sleep, determine the user's sleep anomaly detection results, and implement early warning management based on the user's sleep anomaly detection results. This ensures the safety of the elderly, enables real-time monitoring of the elderly's sleep status and health risk early warning management, and can improve the sleep quality and health level of the elderly population.

[0100] Furthermore, the IoT-based sleep monitoring and management system for elderly care can be applied to home-based care, nursing homes, and medical and rehabilitation settings. In home-based care, it can help seniors manage their health independently while reducing the caregiving burden on their children; for example, children can view their parents' sleep data anytime via a mobile app to stay informed about their health status. In nursing homes, it can monitor seniors' vital signs in real time, improving care efficiency; for example, real-time monitoring of heart rate, respiration, and bed ambulation data can provide caregivers with precise care suggestions and optimize resource allocation. In medical and rehabilitation settings, a remote monitoring system can provide patients with personalized sleep management plans; for example, doctors can adjust treatment plans based on patients' sleep data to improve rehabilitation outcomes.

[0101] Therefore, IoT-based sleep monitoring devices and management systems for elderly healthcare provide efficient and convenient health management services for seniors through real-time monitoring, data analysis, and intelligent feedback. They can be widely used in homes, nursing homes, and medical settings, not only improving the quality of life for seniors but also providing strong technical support for caregivers.

[0102] The system processes multi-source sleep data from users and extracts features relevant to sleep monitoring and management, including:

[0103] Collect abnormal sleep history records of users, and calculate the sleep risk value of users for each sleep period based on the abnormal sleep history records;

[0104] The sleep risk value is compared with a preset sleep risk threshold. Sleep periods when the sleep risk value is greater than or equal to the preset sleep risk threshold are defined as high-risk sleep periods; sleep periods when the sleep risk value is less than the preset sleep risk threshold are defined as low-risk sleep periods.

[0105] The multi-source user sleep data with millisecond-level timestamp tags collected by the sensors is subjected to differential anomaly removal according to the high-risk sleep period and low-risk sleep period to obtain preprocessed multi-source user sleep data;

[0106] Based on a constructed database of suspected sleep features including typical characteristics of apnea, REM sleep, deep sleep, and light sleep;

[0107] The matching degree between the preprocessed user sleep multi-source data and the suspicious sleep feature database is calculated and used as the first evaluation value.

[0108] The score of the bicorrelation coefficient between the preprocessed user sleep multi-source data and historical normal data of the same period is calculated and used as the second evaluation value;

[0109] The first evaluation value and the second evaluation value are weighted and summed to obtain the feature credibility. Based on the feature credibility, the user's sleep period is divided into core feature period and auxiliary association period. A sleep period feature dataset is generated based on the core feature period, auxiliary association period and preprocessed user sleep multi-source data.

[0110] The standard correlation coefficient is determined based on the mean of the total correlation coefficients of normal data from the same historical period.

[0111] The differential threshold is determined based on the standard correlation coefficient and the preset time period coefficient;

[0112] A weighted score is calculated for the sleep period feature dataset. If the weighted score is greater than or equal to the differential threshold, the sleep apnea index, sleep apnea duration, heart rate variability, body motion fluctuation, REM sleep duration, deep sleep duration, and light sleep duration features are output.

[0113] In this embodiment, the sleep risk value is calculated as: Sleep Risk Value = k × (Number of Historical Abnormalities / Total Number of Monitoring Sessions), where k is the severity coefficient of the abnormality type; Sleep apnea events: k = 2.0; Heart rate abnormalities: k = 1.5; Body movement abnormalities: k = 1.0; Other abnormalities: k = 0.8. Number of historical abnormalities: The total number of various abnormal events occurring during this sleep period in the same historical period (e.g., the same month, day of the week); Total number of monitoring sessions: The total number of monitoring days during this sleep period in the same historical period.

[0114] In this embodiment, the multi-source user sleep data with millisecond-level timestamps collected by the sensors undergoes differentiated anomaly removal based on the high-risk and low-risk sleep periods to obtain preprocessed multi-source user sleep data, including:

[0115] When the sensor collects heart rate, respiratory rate, blood oxygen saturation, body movement frequency (physiological data) and temperature, humidity, light intensity, carbon dioxide concentration (environmental data), it automatically carries a millisecond-level timestamp tag.

[0116] High-risk periods: Core physiological data such as heart rate, respiration, body movement, and blood oxygen saturation undergo triple filtering to strictly ensure accuracy. Time-series data are divided into groups of 5 minutes each. If the variance within a group is greater than 1.2 times the mean variance of normal data for the same historical period, the standard score is deleted. , For data where s is the group mean and s is the group standard deviation, if s is greater than 3 (e.g., respiratory rate with a group mean of 16 breaths / min and a standard deviation of 2, and a certain data is 10 breaths / min), then... If the absolute value is 3, do not delete; if the data is 9 times / minute, z-score = -3.5, and the absolute value is > 3, delete); ② After merging all groups, calculate the "global mean" and delete data where "Euclidean distance (the absolute difference between the data and the global mean) is greater than 1.5 times the mean of all distances"; ③ Use the median absolute deviation (MAD) method to calculate Where M is the median of the data, data with |x-M|>3×MAD are deleted to avoid abnormalities such as apnea and sudden heart rate changes being misjudged as normal.

[0117] During low-risk periods: Implement double-selection to balance efficiency and accuracy.

[0118] During low-risk periods: Implement double-selection to balance efficiency and accuracy.

[0119] ① Only delete data with variance exceeding the threshold and z-score > 2.5 (relax the z-score threshold to reduce the accidental deletion of small fluctuations in body movement and heart rate during normal sleep).

[0120] ② Only delete data where the Euclidean distance is greater than 2.0 times the mean of all distances to reduce the calculation cost during low-risk periods.

[0121] In this embodiment, the matching degree between the preprocessed user sleep multi-source data and the suspicious sleep feature database is calculated as the first evaluation value. That is, the matching degree between the preprocessed user sleep multi-source data and the features in the suspicious sleep feature database is as follows: matching one core feature earns 0.5 points, matching two or more features earns 1 point, and this is the first evaluation value. For example, if data for a certain period of time meets the criteria of "a sudden drop in respiratory amplitude of 95% for 12 seconds and blood oxygen saturation of 88%", then it matches the "apnea" characteristic. .

[0122] In this embodiment, the score of the bicorrelation coefficient between the preprocessed user sleep multi-source data and historical normal data from the same period is calculated as the second evaluation value. That is, the standardized correlation coefficient between the current physiological data and the mean of the historical normal data from the same period is calculated and converted into a score: Standard z-score calculation: ; This is the current data; This is the normal average value for the same historical period; Standard deviation (dispersion) of historical data for the same period; first correlation coefficient. : Example reflecting baseline deviation: Current heart rate 65 beats / min, historical average 70 beats / min. ,but Second correlation coefficient :like , ;like , ;like , Quantifying the degree of extreme deviation, for example: current heart rate 55 beats / minute, then... , Total correlation coefficient (R_total): ,make sure Second evaluation value Assignment rules: ( ) 0 points; ) 0.6 points; ) 1 point; Example: If , ,but , .

[0123] In this embodiment, the first evaluation value and the second evaluation value are weighted and summed to obtain the feature credibility.

[0124] High-risk periods: (Prioritize trusting typical features such as sleep apnea and sleep stages to avoid missed detections during high-risk periods); Low-risk periods: (Prioritize trusting data deviation to reduce feature mismatches during low-risk periods).

[0125] In this embodiment, user sleep periods are divided into core feature periods and auxiliary associated periods based on feature credibility, and a sleep period feature dataset is generated based on the core feature periods, auxiliary associated periods, and preprocessed multi-source user sleep data.

[0126] Core feature time period: A continuous time period with feature confidence S ≥ 0.7 (e.g., 02:30-02:45 when apnea feature is detected). The heart rate, respiration, blood oxygen, and body movement preprocessing data of this time period are directly used to extract the apnea duration and apnea index.

[0127] Auxiliary correlation period: The 15-minute period before and after the core period is used to extract auxiliary data (e.g., if the core period is apnea, the auxiliary period is used to extract blood oxygen change trend and heart rate recovery pattern data) to verify the continuity of the apnea index and the comparison of heart rate variability before and after.

[0128] Special case handling: If there is no time period with feature confidence S≥0.7, it is determined to be normal sleep, no core feature time period is generated, and the feature extraction process ends directly;

[0129] The final result is a sleep period feature dataset consisting of core feature period data and auxiliary related period data.

[0130] In this embodiment, the standard correlation coefficient is determined based on the mean of the total correlation coefficients of normal data from the same historical period; that is: standard correlation coefficient = mean of the total correlation coefficients of normal data from the same historical period. Mean; Note: Normal data for the same historical period refers to sleep data without abnormal markers, ensuring... .

[0131] In this embodiment, the differentiation threshold is determined based on the standard correlation coefficient and the preset time period coefficient, including:

[0132] Preset time period coefficients: For example: High-risk core time period: Time period coefficient = 1.3; High-risk auxiliary time period: Time period coefficient = 1.1; Low-risk core time period: Time period coefficient = 0.8; Low-risk auxiliary time period: Time period coefficient = 0.6.

[0133] Differentiation threshold = standard correlation coefficient × time period weight coefficient; Example: if the standard correlation coefficient is 0.4, then the differentiation threshold for the high-risk core time period is 0.4 × 1.3 = 0.52.

[0134] In this embodiment, a weighted score is calculated for the sleep period feature dataset. If the weighted score is greater than or equal to the differential threshold, the following features are output: apnea index, apnea duration, heart rate variability, body motion fluctuation, REM sleep duration, deep sleep duration, and light sleep duration, including:

[0135] Weighted score calculation:

[0136] Core period data score = average feature confidence S within the core feature period

[0137] Auxiliary period data score = average feature confidence S within the auxiliary correlation period

[0138] Weighted score = (Core period data score × Core weight) + (Auxiliary period data score × Auxiliary weight)

[0139] Core weight = 0.6, auxiliary weight = 0.4 (can be adjusted to 0.7 / 0.3 during high-risk periods).

[0140] Judgment criteria: If the weighted score is greater than or equal to the difference threshold, it is judged as a valid sleep feature, and the target feature is output:

[0141] Apnea Index: During the core characteristic period, the number of apnea events (meeting the criteria of a sudden drop in breathing amplitude ≥90% + duration ≥10 seconds + blood oxygen <90%) is counted. Combined with blood oxygen verification during auxiliary related periods (excluding false triggers), the number of apnea events per hour is calculated to obtain the apnea index.

[0142] Apnea duration: Within the core characteristic period, the start and end timestamps of each apnea event are detected using time-series data of respiratory amplitude and blood oxygenation. The time difference is calculated to obtain the duration of each apnea.

[0143] Heart rate variability: Heart rate time series data are extracted during the core and auxiliary time periods, and time domain analysis or frequency domain analysis (calculating the ratio of high-frequency power to low-frequency power) is performed to obtain heart rate variability characteristics.

[0144] Body movement fluctuations: During the core + auxiliary period, the body movement frequency (number of body movements per unit time) and body movement amplitude (variance of the acceleration sensor values) are statistically analyzed to obtain body movement fluctuation characteristics.

[0145] REM sleep / deep sleep / light sleep: Based on heart rate, respiration, body movement, and environmental data during core and auxiliary sleep periods, the model matches patterns from each stage in a suspected sleep feature database (e.g., high heart rate variability + little body movement + dark environment during REM sleep; stable heart rate + even breathing + very little body movement during deep sleep; fluctuating heart rate + more body movement + sensitivity to ambient light during light sleep). Combining weighted confidence, the model divides the duration of each sleep stage and outputs the percentage of deep / light / REM sleep and the total duration.

[0146] Based on the user's sleep abnormality detection results, implement early warning management and perform the following operations:

[0147] Identify user sleep characteristic data and determine the weighted feature values ​​corresponding to the user sleep characteristic data;

[0148] ;

[0149] in, This represents the weighted feature value corresponding to the user's sleep feature data; This represents the k-th raw data collection value of the i-th type of sleep core indicator; This represents the k-th measurement value of the i-th type of sleep core indicator; This represents the mean of the core indicators for the i-th type of sleep. This represents the standard deviation of the core sleep indicator for the i-th type; The standard deviation of the apnea index; This represents the sensor interference correction coefficient, reflecting the impact of environmental noise and transmission delay on the data; To represent extremely small positive numbers to prevent the denominator from being 0; This represents the total number of core sleep indicator types; it is fixed at 7 (corresponding to: apnea index, apnea duration, heart rate variability, body movement fluctuation, percentage of REM sleep, percentage of deep sleep, and percentage of light sleep). This represents the total number of nighttime data collections for a single indicator.

[0150] The sleep anomaly detection results for users are determined based on a sleep monitoring anomaly detection model and weighted feature values; the sleep monitoring anomaly detection model is as follows:

[0151]

[0152] in, This represents a sleep monitoring abnormality detection model; This represents the multidimensional adaptation coefficient of sleep monitoring; ; , , Indicates scene weight; Represents the scene coefficient; Indicates the detection efficiency coefficient; Indicates the precision coefficient; , These represent adjustment parameters, which control accuracy and efficiency respectively.

[0153] In this embodiment, ;

[0154] in, Indicates the number of interference types; Represents the environmental noise figure of the i-th type of interference; This represents the transmission delay coefficient for the i-th type of interference.

[0155] In this embodiment, TP indicates a true positive (number of correctly identified abnormal events); TN indicates a true negative (number of correctly identified normal events); FP indicates a false positive (number of incorrectly identified abnormal events); FN indicates a false negative (number of missed abnormal events).

[0156] In this embodiment, ;in, This indicates the average processing latency.

[0157] In this embodiment, .

[0158] In this embodiment, .

[0159] The working principle and beneficial effects of the above technical solution are as follows: By identifying and calculating weighted feature values ​​from user sleep characteristic data, the system comprehensively considers factors such as the original collected values, measured values, mean, standard deviation, and sensor interference correction coefficients of various core sleep indicators. This enables the system to comprehensively and accurately measure user sleep characteristics, fully considering the importance of different sleep indicators and the impact of environmental interference on the data. For example, changes in key indicators such as the sleep apnea index can be accurately reflected in the weighted feature values, which helps to analyze the user's sleep status more meticulously. A multi-dimensional adaptation coefficient is introduced into the sleep monitoring anomaly detection model, integrating multiple factors such as scene weight, scene coefficient, detection efficiency coefficient, accuracy coefficient, and correction parameters. This allows the model to flexibly adjust the detection focus according to different application scenarios and needs, emphasizing both detection efficiency and accuracy. For example, when monitoring the sleep of the elderly, parameters can be adjusted according to the actual situation to better adapt to the sleep monitoring needs under different environments and health conditions.

[0160] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0161] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A sleep monitoring device for elderly healthcare based on the Internet of Things, comprising a sleep monitor (1), characterized in that, The sleep monitor (1) is connected to a sleep monitoring belt (2) via a data cable (11). The sleep monitoring belt (2) is equipped with a sleep monitoring module. The sleep monitoring module is used to monitor the user's sleep in real time and collect multi-source sleep data of the user. Based on the Internet of Things communication module, the collected multi-source sleep data of the user is transmitted to the cloud platform. The system processes multi-source sleep data from users and extracts features relevant to sleep monitoring and management, including: Collect abnormal sleep history records of users, and calculate the sleep risk value of users for each sleep period based on the abnormal sleep history records; The sleep risk value is compared with a preset sleep risk threshold. Sleep periods when the sleep risk value is greater than or equal to the preset sleep risk threshold are defined as high-risk sleep periods; sleep periods when the sleep risk value is less than the preset sleep risk threshold are defined as low-risk sleep periods. Differential anomaly removal is performed on the multi-source user sleep data with millisecond-level timestamp tags collected by the sensors according to the high-risk sleep period and low-risk sleep period to obtain preprocessed multi-source user sleep data; Based on a constructed database of suspected sleep features including typical characteristics of apnea, REM sleep, deep sleep, and light sleep; The matching degree between the preprocessed user sleep multi-source data and the suspicious sleep feature database is calculated and used as the first evaluation value. The score of the bicorrelation coefficient between the preprocessed user sleep multi-source data and historical normal data of the same period is calculated and used as the second evaluation value; The first evaluation value and the second evaluation value are weighted and summed to obtain the feature credibility. Based on the feature credibility, the user's sleep period is divided into core feature period and auxiliary association period. A sleep period feature dataset is generated based on the core feature period, auxiliary association period and preprocessed user sleep multi-source data. The standard correlation coefficient is determined based on the mean of the total correlation coefficients of normal data from the same historical period. The differential threshold is determined based on the standard correlation coefficient and the preset time period coefficient; A weighted score is calculated for the sleep period feature dataset. If the weighted score is greater than or equal to the differential threshold, the sleep apnea index, sleep apnea duration, heart rate variability, body motion fluctuation, REM sleep duration, deep sleep duration, and light sleep duration features are output.

2. The sleep monitoring device for elderly health care based on the Internet of Things according to claim 1, characterized in that, The sleep monitor (1) is connected to a power plug (3) via a power cord (12). The power cord (12) is connected to the power plug (3) to be powered on. After being powered on, the sleep monitor (1) is charged. The sleep monitor (1) is also equipped with an emergency call button (13) and a sensor night light (14). The sensor night light (14) is located on one side of the emergency call button (13).

3. A sleep monitoring and management system for elderly healthcare based on the Internet of Things, characterized in that: The sleep monitoring module of the IoT-based sleep monitoring device for elderly healthcare as described in claim 2, and the IoT-based sleep monitoring management system for elderly healthcare further include: The data processing module is used to process multi-source sleep data of users and extract features related to sleep monitoring and management to determine the user's sleep characteristic data; The intelligent analysis module is used to build a sleep monitoring anomaly detection model to analyze and identify user sleep characteristic data, judge the user's sleep quality, automatically identify abnormal situations in the user's sleep, and determine the user's sleep anomaly detection results. The alarm management module is used to provide early warnings based on the results of abnormal sleep detection, ensuring the safety of the elderly.

4. The sleep monitoring and management system for elderly healthcare based on the Internet of Things according to claim 3, characterized in that, The sleep monitoring module collects multi-source sleep data from the user and performs the following operations: Based on sensors, the system monitors the user's heart rate, respiratory rate, blood oxygen saturation, and body movement frequency in real time while the user is sleeping, and collects the user's sleep physiological data. Based on sensors, the system monitors the temperature, humidity, light, and carbon dioxide concentration in real time while the user is sleeping, and collects data on the user's sleep environment. Based on user sleep physiological data and user sleep environment data, multi-source user sleep data is generated.

5. The sleep monitoring and management system for elderly healthcare based on the Internet of Things according to claim 4, characterized in that, Process user sleep data from multiple sources and extract features relevant to sleep monitoring and management, performing the following operations: The user sleep multi-source data is processed based on Kalman filtering to remove high-frequency noise and clean the user sleep multi-source data to remove outliers. Based on Z-score standardization, user sleep multi-source data is processed and normalized to form standardized user sleep multi-source data with different scales. Principal component analysis was used to process multi-source sleep data of users, extracting features related to sleep monitoring and management from the multi-source sleep data of users, and determining user sleep characteristic data, including apnea index, apnea duration, heart rate variability, body movement fluctuation, REM sleep, deep sleep and light sleep.

6. The sleep monitoring and management system for elderly healthcare based on the Internet of Things according to claim 5, characterized in that, The system analyzes and identifies user sleep characteristic data to determine sleep quality and automatically identifies abnormalities during sleep, performing the following operations: A sleep monitoring anomaly detection model was constructed and deployed in the user's sleep monitoring anomaly detection environment to detect anomalies in the user's sleep. The system takes user sleep characteristic data as input data and feeds it into the sleep monitoring anomaly detection model. The model analyzes and identifies the user's sleep characteristic data, judges the user's sleep quality, automatically identifies abnormal situations in the user's sleep, and outputs the user's sleep anomaly detection results.

7. The sleep monitoring and management system for elderly healthcare based on the Internet of Things according to claim 6, characterized in that, Construct a sleep monitoring anomaly detection model and perform the following operations: Historical sleep monitoring data was collected and divided into training and testing sets. The machine learning model is trained using a training set, enabling the model to autonomously learn sleep monitoring anomaly detection behavior from the training set, judge the user's sleep quality, and automatically identify abnormal situations in the user's sleep, thus determining the sleep monitoring anomaly detection model. The sleep monitoring anomaly detection model was tested using a test set, and the model's performance was evaluated based on the test results to determine whether it could achieve the expected effect of automatically identifying abnormalities in the user's sleep. When the sleep monitoring anomaly detection model fails to achieve the expected effect of automatically identifying abnormal situations in the user's sleep, the parameters of the sleep monitoring anomaly detection model are adjusted and optimized, and the optimized sleep monitoring anomaly detection model is tested until the sleep monitoring anomaly detection model can achieve the expected effect of automatically identifying abnormal situations in the user's sleep, and the optimal sleep monitoring anomaly detection model is determined.

8. The sleep monitoring and management system for elderly healthcare based on the Internet of Things according to claim 7, characterized in that, Based on the user's sleep abnormality detection results, implement early warning management and perform the following operations: When the user's sleep abnormality detection result indicates that the user's sleep is abnormal, an alarm message is automatically sent to family members and medical staff, and a sleep quality analysis report is generated based on the user's multi-source sleep data and displayed in a visual form. Medical staff can view the sleep quality analysis reports of the elderly in real time through mobile terminals, and provide suggestions for improving sleep based on the sleep quality analysis reports and adjust the users' lifestyles in a timely manner to ensure the safety of the elderly.

9. The sleep monitoring and management system for elderly healthcare based on the Internet of Things according to claim 7, characterized in that, Based on the user's sleep abnormality detection results, implement early warning management and perform the following operations: Identify user sleep characteristic data and determine the weighted feature values ​​corresponding to the user sleep characteristic data; ； in, This represents the weighted feature value corresponding to the user's sleep feature data; This represents the k-th raw data collection value of the i-th type of sleep core indicator; This represents the k-th measurement value of the i-th type of sleep core indicator; This represents the mean of the core indicators for the i-th type of sleep. This represents the standard deviation of the core sleep indicator for the i-th type; The standard deviation of the apnea index; This represents the sensor interference correction coefficient, reflecting the impact of environmental noise and transmission delay on the data; To represent extremely small positive numbers to prevent the denominator from being 0; This represents the total number of core sleep indicator types; it is fixed at 7. This represents the total number of nighttime data collections for a single indicator. The sleep anomaly detection results for users are determined based on a sleep monitoring anomaly detection model and weighted feature values; the sleep monitoring anomaly detection model is as follows: in, This represents a sleep monitoring abnormality detection model; This represents the multidimensional adaptation coefficient of sleep monitoring; ; , , Indicates scene weight; Represents the scene coefficient; Indicates the detection efficiency coefficient; Indicates the precision coefficient; , These represent adjustment parameters, which control accuracy and efficiency respectively.

Images