Inpatient wearing monitoring equipment integrating omnibearing monitoring and automatic reminding and conversation

Abstract

The invention discloses hospitalized patient wearing monitoring equipment integrating all-directional monitoring and automatic reminding and conversation, and belongs to the technical field of medical monitoring equipment. The problems that an existing device is limited in monitoring range and cannot achieve mobile monitoring, a single threshold value alarm mode is adopted, and invalid alarm is likely to be generated are solved, and by collecting vital signs and body position data of a patient, through data processing and intelligent analysis, based on a preset multi-parameter association rule and in combination with weight coefficient calculation, the vital signs and the body position data of the patient are obtained. Comprehensively judging an abnormal level; grading response is triggered according to the abnormal grade, mild reminding is carried out only through the patient interaction terminal, a patient is guided to adjust autonomously, synchronous early warning is carried out on the medical care terminal, and equipment end sound-light alarm, medical care end emergency pop-up windows and central monitoring system alarm are triggered immediately when serious abnormity occurs. Therefore, the function from patient self-intervention to medical care rapid intervention is realized, and the monitoring efficiency and the patient safety are effectively improved.

Description

Technical Field

[0001] This invention relates to the field of medical monitoring equipment technology, specifically to a monitoring device for inpatients that integrates comprehensive monitoring, automatic reminders, and dialogue. Background Technology

[0002] Monitoring vital signs of hospitalized patients is a core part of clinical diagnosis and treatment and nursing work, especially for critically ill patients, postoperative rehabilitation patients, and patients with chronic diseases. It is necessary to continuously monitor key vital signs parameters such as heart rate, blood oxygen saturation, and body temperature to detect abnormalities in a timely manner and take appropriate measures to reduce the risk of disease deterioration.

[0003] Current methods for monitoring hospitalized patients mostly use bedside fixed monitors, which have limitations in monitoring range and cannot achieve mobile monitoring. Monitoring is easily interrupted when patients turn over or get out of bed, making it difficult to obtain continuous physiological parameter data. At the same time, traditional monitoring equipment mostly uses a single threshold alarm mode, without personalized threshold settings based on individual patient differences, which easily generates invalid alarms, increases the workload of medical staff, and the alarm information is not accurate enough, resulting in a lag in medical staff response.

[0004] Therefore, to meet existing needs, a wearable monitoring device for hospitalized patients, integrating comprehensive monitoring, automatic reminders, and dialogue, is proposed. Summary of the Invention

[0005] The purpose of this invention is to provide a wearable monitoring device for hospitalized patients that integrates comprehensive monitoring, automatic reminders, and dialogue. By collecting patients' vital signs and body position data, and through data processing and intelligent analysis, based on preset multi-parameter association rules and weighted coefficient calculations, the device comprehensively judges the level of abnormality. According to the level of abnormality, a graded response is triggered, thereby realizing the function of patient self-intervention and rapid medical intervention, effectively improving monitoring efficiency and patient safety, and solving the problems mentioned in the background art.

[0006] To achieve the above objectives, the present invention provides the following technical solution: A comprehensive monitoring device for hospitalized patients, featuring automatic alerts and dialogue, includes: a wearable main body, a multi-sensor acquisition module and a graded early warning module integrated within the wearable main body, and a patient interaction terminal integrated on the outside of the wearable main body; The multi-sensor acquisition module consists of a heart rate sensor, a blood oxygen saturation sensor, a respiratory rate sensor, a core body temperature sensor, and a body position sensor. It is configured to continuously and dynamically acquire the patient's key vital signs and body position, and output the raw acquisition data. The graded early warning module is configured to compare and analyze the preprocessed data with a preset normal parameter threshold range, and combine multi-parameter association rules to determine whether the data is abnormal and the level of abnormality. The level of abnormality includes normal, slight abnormality, moderate abnormality and severe abnormality, and triggers corresponding response actions according to the level of abnormality. The patient interaction terminal is configured to receive and send tiered early warning reminders to the patient, and supports dialogue and interaction between the patient and the device, as well as status feedback.

[0007] Furthermore, the tiered early warning module includes: The rule-building unit is configured to be based on standard physiological and pathophysiological knowledge and historical clinical diagnosis and treatment data, and to sort out the clinical pathological mechanisms of abnormal linkage of various physiological parameters and establish a clinical data sample library. Based on clinical pathological mechanisms and sample database data, we analyzed common abnormal physiological parameter association scenarios in patients, identified the core associated parameter combinations in various scenarios, and divided them into five categories of associated scenarios, including: cardiopulmonary parameter linkage scenario, body temperature and vital signs coordination scenario, body position and vital signs matching scenario, multi-parameter coordinated deterioration scenario, and parameter fluctuation trend association scenario. Statistical analysis methods were used to calculate the incidence of abnormal linkages of different parameters and the correlation of disease deterioration under various scenarios; combined with clinical diagnosis and treatment experience, association rules and their initial weight coefficients for various scenarios were defined.

[0008] Furthermore, the association rules for various scenarios and their initial weight coefficients are defined, including: Define the judgment conditions and corresponding weighting coefficients for abnormal linkage of heart rate, respiratory rate and blood oxygen saturation in the cardiopulmonary parameter linkage scenario; Define the linkage mode of body temperature, heart rate, and respiratory rate in the scenario of coordinated body temperature and vital signs, and set the weight enhancement coefficient of the grade and the corresponding abnormality level mapping according to the magnitude and duration of body temperature rise and the severity of accompanying vital signs. Establish a baseline for the safe duration of different body positions in a body position and vital sign matching scenario, and define the judgment logic and weight enhancement mechanism for abnormal body positions linked to heart rate fluctuations and blood oxygen decline. Define a judgment logic based on the combination of the number and severity of parameter anomalies in a multi-parameter collaborative deterioration scenario, and set a high-weight boosting coefficient for it; Define a quantitative standard for a continuously deteriorating trend in the context of parameter fluctuation trend correlation, and set a weighting factor for forward-looking early warning.

[0009] Furthermore, defining the association rules for various scenarios and their initial weight coefficients also includes: The established association rules and weighting coefficients are applied to clinical sample data and pilot patient monitoring. Data on the accuracy of anomaly detection, false alarm rate, and false negative rate are collected after application. The results are compared with the actual clinical treatment outcomes to analyze the rule suitability. For false alarm and false negative scenarios, optimize the accuracy of scenario decomposition and weight coefficient assignment, and iteratively update the content of association rules until the accuracy of rule anomaly judgment meets the requirements of clinical monitoring. The iteratively optimized association rules and weight parameters are embedded into the control program of the hierarchical early warning module to establish a real-time rule invocation mechanism; When receiving processed data, quickly match the corresponding related scenarios and call the corresponding weight coefficients to complete the anomaly level determination; Establish personalized adjustment rules and preset weight coefficient adjustment ranges for hospitalized patients of different ages, diseases, and severity of illness; Based on the individual treatment plans of the monitored patients, the rules and their weights are adjusted in a personalized manner, and the adjusted association rules and weight parameters are saved.

[0010] Furthermore, corresponding response actions are triggered based on the anomaly level, including: When the anomaly level is normal, no alert action will be initiated, and the processed data will be uploaded to the central monitoring system for storage and archiving. When the abnormality level is minor, a reminder instruction is sent to the patient's interactive terminal to guide the patient to adjust their position or state on their own through vibration and voice prompts, while recording the reminder time and data information. When the abnormality level is moderate, instructions are sent to the patient interaction terminal and the medical staff mobile terminal simultaneously. The patient interaction terminal triggers vibration and voice reminder, and the medical staff mobile terminal receives text warning information, including abnormal parameters, abnormality level and patient identity information. When the abnormality level is severe, the wearable device will immediately trigger an audible and visual alarm, send an emergency pop-up command to the medical staff mobile terminal and an alarm command to the central monitoring system, and simultaneously record the time of the abnormality, the value of the abnormal parameters, the level of the abnormality and the patient's position, forming an abnormal event report and uploading it for archiving.

[0011] Furthermore, the tiered early warning module also includes: The data processing module is configured to use an adaptive Kalman filter algorithm to filter out motion interference noise, electromagnetic interference noise and baseline drift interference in the raw acquired data, and retain the valid data. A multi-sensor fusion calibration algorithm is used to correct the deviation of the filtered effective data by combining the patient's basic physiological information and the historical benchmark data of the sensors.

[0012] Furthermore, the wearable device also integrates a power management module that supplies power to each module, as well as a communication module that enables data interaction and command transmission.

[0013] Furthermore, the wearable main body is a vest-style design, made of flexible and breathable material. Three Velcro straps are provided around the front of one side and the inside of the other side of the wearable main body, which can adjust the tightness of the fit according to the patient's body shape.

[0014] Furthermore, the rule-building unit, based on standard physiological and pathophysiological knowledge and historical clinical diagnosis and treatment data, and by sorting out the clinical pathological mechanisms of abnormal linkages of various physiological parameters, establishes a standardized clinical data sample library, including: Based on standard physiological and pathophysiological knowledge and clinical guidelines, four core dimensions were identified: circulatory system, respiratory system, metabolic system, and neuro-motor regulation system. Each core dimension was bound to the corresponding sensor's acquisition parameters, units, accuracy, and initial abnormality judgment criteria, resulting in a sensor clinical data dictionary. Historical clinical diagnosis and treatment data are acquired, and parameter mapping and preliminary screening are performed on the historical clinical diagnosis and treatment data based on the sensor clinical data dictionary to identify abnormal parameters collected by the sensors. An expert working group was formed to analyze the abnormal linkage logic of the abnormal parameters based on the disease pathology mechanism, and to determine the abnormal linkage combinations and types. The expert working group, based on the GRADE standard, labeled each abnormal linkage combination with a supporting evidence level and determined the evidence level. An initial table of abnormal linkage relationships of physiological parameters was constructed based on abnormal linkage parameter combinations, abnormal linkage types, associated diseases, pathological mechanisms, and levels of evidence. Based on preset screening criteria, the historical clinical diagnosis and treatment data mapped by parameters are screened a second time to obtain the original diagnosis and treatment dataset; the original diagnosis and treatment dataset includes abnormal linkage parameter combination codes, intervention methods, parameter change rates, prognostic grading, and follow-up periods. The original clinical dataset is validated based on a preset validation rule base. After successful validation, a valid clinical dataset is output. The valid clinical dataset includes raw sensor data, clinical symptom recording time, diagnostic and treatment intervention information, and prognostic data. Calculate the treatment effectiveness weights based on the aforementioned effective clinical dataset; A set of clinical expert consensus nodes was constructed. Based on the scores of multidisciplinary experts on each combination in the initial table of abnormal linkages of the physiological parameters, different dimensions were scored, and the average score of different dimensions was taken as the expert consensus weight. Define the weight coefficient for each level of evidence to obtain the weight of each level of evidence; The comprehensive pathological weight is determined based on the weight of treatment effectiveness, expert consensus, and evidence level. The comprehensive pathological weight is then associated with the corresponding combination in the initial table of physiological parameter abnormality linkage, generating a physiological parameter abnormality linkage-pathological weight library. Based on the clinical symptom recording time and treatment intervention time in the effective clinical dataset, the disease course is divided into stages according to disease progression. The effective clinical dataset is grouped based on the disease course stage and abnormal linkage combination. The 95% confidence interval of the abnormal parameters in each group is calculated as the initial threshold of the group. The pre-built and continuously maintained calibration event library is called, and the initial threshold is calibrated based on the threshold adjustment cases in the calibration event library to obtain the abnormal threshold table of parameters by disease course. Based on a single record in a valid clinical dataset, and combined with a table of abnormal thresholds for disease course parameters and a preliminary table of abnormal linkages of physiological parameters, sample units are constructed; a clinical data sample library is then built based on these sample units.

[0015] Furthermore, the calculation of the incidence rate of abnormal linkages of different parameters and the correlation degree of disease deterioration under various scenarios includes: According to the preset monitoring time step, obtain the patient identifier set in the target scene where the corresponding parameter combination has abnormal linkage within each preset monitoring time step; based on the preset total duration of the monitoring time window, perform a union operation on the patient identifier sets within all preset monitoring time steps to obtain the deduplicated patient set where the parameter combination has abnormal linkage throughout the entire monitoring cycle. Obtain the total number of effective monitored patients in various scenarios from the clinical data sample library, and calculate the incidence of abnormal linkage of different parameters in various scenarios based on the ratio of the number of elements in the deduplicated patient set to the total number of effective monitored patients.

[0016] ; in, Let be the occurrence rate of abnormal linkage of the p-th parameter combination in the s-th scenario; The number of deduplicated patients who experience abnormal linkage of the p-th parameter combination under the s-th scenario within the monitoring period; The total number of patients in the s-th scenario of the clinical data sample library for effective monitoring; Based on the incidence of abnormal linkages of different parameters in various scenarios, the correlation degree of disease deterioration corresponding to different parameter combinations in various scenarios is calculated. ; in, Let be the correlation between the p-th parameter combination and the q-th abnormal event in the s-th scenario; The number of parameters in the p-th parameter combination; The abnormal severity weight of the i-th parameter in the p-th parameter combination is determined based on the threshold of the current disease stage; This is the duration coefficient of the abnormal linkage in the q-th abnormal event; Let be the clinicopathological correlation coefficient for the s-th scenario; Let be the initial weight coefficients for the p-th parameter combination in the s-th scenario.

[0017] Compared with the prior art, the beneficial effects of the present invention are: 1. By breaking down five common physiological parameter association scenarios and quantifying the association rules and initial weight coefficients for each scenario, the clinical adaptability and scientific validity of the association rules can be achieved. Through rule embedding and real-time invocation mechanisms, rapid matching of monitoring data with associated scenarios and accurate determination of abnormality levels can be realized, improving the efficiency of early warning response. By adjusting the rules and optimizing the weight coefficients in a personalized manner, the problems of misjudgment and missed judgment caused by single parameter early warning or uniform rules can be further avoided, enhancing the accuracy and foresight of abnormal early warning, providing medical staff with reliable basis for judging the condition, and helping to quickly manage acute deterioration of the condition.

[0018] 2. Establish a four-level tiered early warning mechanism, combined with personalized threshold settings for patients, to achieve more accurate and timely abnormal early warnings, effectively reduce invalid alarms, and lower the workload of medical staff; at the same time, through multi-terminal collaborative response, ensure that serious abnormalities are dealt with quickly, and improve nursing efficiency and treatment safety.

[0019] 3. It integrates voice dialogue and patient interaction functions, which can guide patients to adjust their status independently, realize the combination of active care and passive monitoring, improve patient participation and user experience, and at the same time facilitate patients to query their own monitoring data in real time, thereby improving the efficiency of doctor-patient communication. Attached Figure Description

[0020] Figure 1 This is a schematic diagram of the overall monitoring device of the present invention; Figure 2 This is a schematic diagram of the internal structure of the monitoring device of the present invention; Figure 3 This is a schematic diagram of another state of the monitoring device of the present invention; Figure 4 This is a flowchart of the monitoring process of the present invention.

[0021] In the diagram: 1. Wearable main unit; 2. Heart rate sensor; 3. Blood oxygen saturation sensor; 4. Respiratory rate sensor; 5. Core body temperature sensor; 6. Body position sensor; 7. Patient interaction terminal; 8. Power management module; 9. Communication module; 10. Velcro. Detailed Implementation

[0022] 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.

[0023] To address the limitations of existing bedside monitors, which are mostly fixed and lack mobile monitoring capabilities, making it difficult to obtain continuous physiological parameter data due to patient movement such as turning over or getting out of bed, and the fact that traditional monitoring equipment often uses a single threshold alarm mode without individualized threshold settings based on patient differences, leading to invalid alarms, increased workload for medical staff, and inaccurate alarm information transmission resulting in delayed response from medical staff, please refer to [link to relevant documentation]. Figures 1-4 This embodiment provides the following technical solution: This inpatient monitoring device integrates comprehensive monitoring, automatic reminders, and dialogue. It includes: a wearable main unit 1, a multi-sensor acquisition module and a graded early warning module integrated within the main unit 1, and a patient interaction terminal 7 integrated on the outside of the main unit 1. The main unit 1 also integrates a power management module 8, which powers all modules. This module uses a lithium battery, supports wireless charging, and integrates a low-battery monitoring function. When the battery level falls below a preset threshold, it triggers voice and vibration reminders and simultaneously sends a low-battery alert to the medical staff's mobile terminal. A communication module 9 enables data interaction and command transmission. This module uses a Bluetooth BLE+5G dual-mode communication design, enabling data transmission between the wearable main unit 1 and the patient interaction terminal 7 at close range, and high-speed data and command transmission with the medical staff's mobile terminal and the central monitoring system at long distances, ensuring real-time delivery of emergency alarm commands. Furthermore, the communication module 9 uses AES encryption to encrypt patient physiological data and identity information, protecting patient privacy and complying with medical data security standards. The wearable main body 1 is designed as a vest and made of flexible and breathable material. Three Velcro straps 10 are arranged around the front of one side and the inside of the other side to adjust the tightness of the wear according to the patient's body shape, so that the wearable main body 1 fits the human body and can be adapted to patients of different body types. It is also waterproof to prevent the device from being damaged during the patient's daily activities and cleaning.

[0024] The multi-sensor acquisition module consists of a heart rate sensor 2, a blood oxygen saturation sensor 3, a respiratory rate sensor 4, a core body temperature sensor 5, and a body position sensor 6. It is configured to continuously and dynamically acquire the patient's key vital signs and body position, and output the raw acquisition data. The body position sensor 6 adopts a three-axis accelerometer and gyroscope fusion design to identify the patient's supine, lateral, prone, sitting, and standing postures. When the body position is continuously in an abnormal posture and accompanied by abnormal vital signs, the weight of the abnormality level judgment is increased.

[0025] The beneficial effects achieved by the above are: enabling continuous dynamic monitoring of multiple parameters, covering five key indicators: heart rate, blood oxygen saturation, respiratory rate, core body temperature, and body position; employing a multi-sensor fusion design, combined with adaptive filtering and calibration algorithms, significantly improving data acquisition accuracy and reducing errors caused by interference, thus providing a reliable basis for disease diagnosis; and simultaneously featuring encrypted data transmission, dual backup, and backtracking functions, protecting patient privacy while facilitating subsequent disease analysis and treatment reference for medical staff, while also meeting the requirements for medical data security and archiving.

[0026] The graded early warning module is configured to compare and analyze preprocessed data with preset normal parameter threshold ranges, and combine multi-parameter association rules, such as: a higher association weight for increased heart rate accompanied by shortness of breath and decreased blood oxygen, to determine whether the data is abnormal and its abnormality level. The abnormality level includes normal, slight abnormality, moderate abnormality, and severe abnormality, and triggers corresponding response actions according to the abnormality level; specifically including: When the abnormality level is normal, no alert action is initiated, and the processed data is uploaded to the central monitoring system for storage and archiving. When the abnormality level is minor, an alert instruction is sent to the patient interaction terminal 7, guiding the patient to adjust their position or status independently through vibration and voice prompts, while recording the alert time and data information. When the abnormality level is moderate, instructions are simultaneously sent to the patient interaction terminal 7 and the medical staff mobile terminal. The patient interaction terminal 7 triggers vibration and voice alerts, and the medical staff mobile terminal receives text warning information, including abnormal parameters, abnormality level, and patient identity information. When the abnormality level is severe, the wearable main unit 1 immediately triggers an audible and visual alarm, simultaneously sending an emergency pop-up instruction to the medical staff mobile terminal and an alarm instruction to the central monitoring system, while simultaneously recording the time of the abnormality, abnormal parameter values, abnormality level, and patient position status, forming an abnormal event report and uploading it for archiving.

[0027] The beneficial effects achieved by the above are as follows: By establishing a four-level graded early warning mechanism and combining it with patient-specific threshold settings, the accuracy and timeliness of abnormal early warnings can be achieved, effectively reducing invalid alarms and lowering the workload of medical staff; at the same time, through multi-terminal collaborative response, serious abnormalities can be dealt with quickly, improving nursing efficiency and the safety of diagnosis and treatment.

[0028] The tiered early warning module includes: The data processing module is configured to use an adaptive Kalman filter algorithm to filter out motion interference noise, electromagnetic interference noise, and baseline drift interference in the raw acquired data, retaining valid data; it also employs a multi-sensor fusion calibration algorithm, combining the patient's basic physiological information and historical sensor benchmark data to correct deviations in the filtered valid data, thereby improving data accuracy; and it also integrates an abnormal data backtracking function, which can store at least 30 days of continuous monitoring data and abnormal event records, supporting medical staff to query and export data.

[0029] The rule-building unit is configured to be based on standard physiological and pathophysiological knowledge and historical clinical diagnosis and treatment data. It sorts out the clinical pathological mechanisms of various physiological parameter abnormalities, filters effective data and removes extreme outliers and invalid data to establish a clinical data sample library. Based on the clinical pathological mechanisms and sample library data, it analyzes common abnormal physiological parameter association scenarios in patients, clarifies the core associated parameter combinations in various scenarios, and divides them into five types of association scenarios, including: cardiopulmonary parameter linkage scenario, body temperature and vital signs coordination scenario, body position and vital signs matching scenario, multi-parameter coordinated deterioration scenario, and parameter fluctuation trend association scenario. Statistical analysis methods were used to calculate the incidence of abnormal linkages of different parameters and the correlation of disease deterioration under various scenarios; combined with clinical experience, association rules and their initial weight coefficients for various scenarios were defined; the specific steps included: Define the judgment conditions (e.g., amplitude range and time synchronization) and the corresponding weight enhancement coefficients (e.g., 60%-80%) for abnormal linkage of heart rate, respiratory rate and blood oxygen saturation in the cardiopulmonary parameter linkage scenario.

[0030] Define the linkage mode of body temperature, heart rate, and respiratory rate in the scenario of coordinated body temperature and vital signs. Based on the magnitude and duration of body temperature rise and the severity of accompanying vital signs, set the weight enhancement coefficient of the classification (e.g., 50%-70% and higher) and the corresponding abnormality level mapping (moderate or severe abnormality).

[0031] Establish a baseline for the safe duration of different body positions (e.g., prone and sitting) in a scenario where body position is matched with vital signs, and define the judgment logic and weight enhancement mechanism (e.g., 40%-50%) that links abnormal body position with heart rate fluctuations and blood oxygen decline, and clarify the upgrade conditions from reminder to warning.

[0032] Define a judgment logic based on the combination of the number and severity of abnormal parameters in a multi-parameter collaborative deterioration scenario (e.g., three mild or two moderate abnormalities equivalent to severe abnormalities), and set a high weighting factor (e.g., 70%-90%) to cover acute deterioration scenarios.

[0033] Define quantitative standards for a continuously deteriorating trend in the context of parameter fluctuation trend correlation (e.g., time window, fluctuation amplitude), and set a weighting factor for forward-looking early warning (e.g., 30%-40%).

[0034] The established association rules and weight coefficients are applied to clinical sample data and pilot patient monitoring. Data on the accuracy of anomaly detection, false alarm rate, and missed alarm rate are collected after application. The results are compared with the actual clinical treatment outcomes to analyze the rule adaptability. For false alarm and missed alarm scenarios, the accuracy of association scenario decomposition and weight coefficient assignment are optimized, and the content of association rules is iteratively updated until the anomaly detection accuracy of the rules meets the requirements of clinical monitoring. The iteratively optimized association rules and weight parameters are embedded into the control program of the hierarchical early warning module to establish a real-time rule invocation mechanism.

[0035] Upon receiving processed data, the system quickly matches the corresponding related scenarios and calls the corresponding weight coefficients to determine the anomaly level. It establishes personalized adjustment rules and presets the weight coefficient adjustment range based on hospitalized patients of different ages, diseases, and disease severity. Based on the individual treatment plans of the monitored patients, the system makes personalized adjustments to the rules and their weights and saves the adjusted related rules and weight parameters to ensure that the rules are adapted to personalized medical monitoring needs.

[0036] In one embodiment, for example, regarding the association rule for cardiopulmonary parameters: when heart rate increases by 10%-20% above the upper limit of the normal threshold, accompanied by shortness of breath (10%-20% above the upper limit of the normal threshold) and decreased blood oxygen saturation (5%-10% below the lower limit of the normal threshold), the abnormal weight is increased by 60%-80% compared to a single parameter abnormality, and it is preferentially judged as a moderate or higher abnormality, corresponding to the early warning needs of coordinated abnormalities in clinical center pulmonary function, especially suitable for cardiopulmonary function monitoring of postoperative and critically ill patients; if the increase in heart rate exceeds the upper limit of the normal threshold by 20%, the increase in shortness of breath exceeds the upper limit of the normal threshold by 20%, and the blood oxygen saturation is more than 10% below the lower limit of the normal threshold, when the three factors are linked to trigger the superposition of severe abnormal weights, the abnormality level judgment weight is directly increased to the highest level, quickly triggering an emergency alarm to investigate the risk of acute diseases such as respiratory failure and heart failure.

[0037] Regarding the rules for linking body temperature and vital signs: When core body temperature rises by 0.5-1°C above the upper limit of the normal threshold, accompanied by increased heart rate and respiratory rate, the abnormal weight increases by 50%-70%, and it is preferentially judged as a moderate abnormality. This aligns with the clinical characteristics of the physiological stress response of febrile patients and avoids misjudging simple fever as a severe abnormality. If the core body temperature rises by more than 1°C above the upper limit of the normal threshold, accompanied by significant abnormalities in heart rate and respiratory rate, and lasting for more than 5 minutes, the abnormal weight increases further, and it is judged as a severe abnormality, alerting to critical situations such as worsening infection and sepsis.

[0038] Regarding the rules for linking body position with vital signs: The body position sensor 6 detects that the patient is in a prone or sitting position for too long, such as exceeding the preset safe duration, and is accompanied by heart rate fluctuations exceeding the normal range and a slight decrease in blood oxygen saturation. The abnormal weight increases by 40%-50%, and it is judged as a mild to moderate abnormality, triggering a body position adjustment reminder. This is especially suitable for bedridden patients and patients after cervical / spinal surgery to prevent complications such as pressure sores and postural hypoxia. If the duration of the abnormal body position exceeds twice the safe duration and is accompanied by a continuous deterioration in heart rate and blood oxygen saturation, the abnormal weight is added up, upgrading to a moderate or higher abnormality, and a medical staff warning is triggered simultaneously to prevent further development of the condition.

[0039] Regarding the multi-parameter coordinated deterioration rule: When any three or more of the four parameters—heart rate, blood oxygen saturation, respiratory rate, and core body temperature—show mild abnormalities simultaneously, or any two show moderate abnormalities simultaneously, the abnormality weight increases by 70%-90%, directly determining it as a severe abnormality and triggering multi-terminal emergency alarms. This is suitable for monitoring scenarios involving multiple organ dysfunction and acute deterioration of the condition. If a single parameter shows a severe abnormality, accompanied by mild or above abnormalities in any of the other parameters, the abnormality weights are superimposed, strengthening the alarm priority and ensuring rapid intervention by medical staff.

[0040] Regarding the correlation rules for parameter fluctuation trends: If the monitoring data shows a continuous deterioration trend, such as a continuous increase in heart rate and a continuous decrease in blood oxygen saturation over 10 minutes, and the fluctuation range exceeds the normal range every 2 minutes, even if the threshold for severe abnormality is not reached at a single point in time, the abnormal weight is increased by 30%-40% based on the trend change and the fluctuation of the associated parameters, and the abnormality level is upgraded in advance to achieve a proactive warning and buy time for medical staff to deal with the situation.

[0041] The beneficial effects achieved by the above content are as follows: By decomposing five common physiological parameters associated with abnormal scenarios and quantifying the association rules and initial weight coefficients for each scenario, the clinical adaptability and scientific validity of the association rules can be achieved; with the help of rule embedding and real-time invocation mechanisms, rapid matching of monitoring data with associated scenarios and accurate determination of abnormality levels can be achieved, improving the efficiency of early warning response; by adjusting the rules in a personalized manner and optimizing the weight coefficients, the problems of misjudgment and missed judgment caused by single parameter early warning or uniform rules can be further avoided, enhancing the accuracy and foresight of abnormal early warning, providing medical staff with reliable basis for judging the condition, and helping to quickly deal with acute deterioration of the condition.

[0042] The patient interaction terminal 7 is configured to receive and provide tiered warning and reminder instructions to patients, and supports dialogue and status feedback between patients and the device. Patients can query current vital sign data, reasons for abnormal reminders, and self-adjustment suggestions through voice. Based on the built-in voice recognition and response database, it can realize real-time recognition and accurate response of dialogue content.

[0043] The beneficial effects achieved by the above content are as follows: By integrating voice dialogue and patient interaction functions, patients can be guided to adjust their status independently, realizing a combination of proactive care and passive monitoring, improving patient participation and user experience, while also facilitating patients to query their own monitoring data in real time, and improving the efficiency of doctor-patient communication.

[0044] Working principle: By collecting patients' vital signs and position data, and through data processing and intelligent analysis, based on preset multi-parameter association rules and weight coefficient calculations, the abnormality level is comprehensively judged. According to the abnormality level, a graded response is triggered, ranging from gentle reminders and guidance to patients to adjust themselves through the patient interaction terminal 7, to simultaneous early warning to medical staff terminals, and finally to immediate triggering of audible and visual alarms on the device, emergency pop-ups on the medical staff terminals, and alarms in the central monitoring system in the event of serious abnormalities. This realizes the function of enabling patients to intervene themselves and medical staff to intervene quickly, effectively improving monitoring efficiency and patient safety.

[0045] The rule-building unit is based on standard physiological and pathophysiological knowledge and historical clinical diagnosis and treatment data, and sorts out the clinical pathological mechanisms of abnormal linkages of various physiological parameters to establish a standardized clinical data sample library, including: Based on standard physiological and pathophysiological knowledge and clinical guidelines, four core dimensions were identified: circulatory system, respiratory system, metabolic system, and neuro-motor regulation system. Each core dimension was bound to the corresponding sensor's acquisition parameters, units, accuracy, and initial abnormality judgment criteria, resulting in a sensor clinical data dictionary. Historical clinical diagnosis and treatment data are acquired, and parameter mapping and preliminary screening are performed on the historical clinical diagnosis and treatment data based on the sensor clinical data dictionary to identify abnormal parameters collected by the sensors. An expert working group was formed to analyze the abnormal linkage logic of the abnormal parameters based on the disease pathology mechanism, and to determine the abnormal linkage combinations and types. The expert working group, based on the GRADE standard, labeled each abnormal linkage combination with a supporting evidence level and determined the evidence level. An initial table of abnormal linkage relationships of physiological parameters was constructed based on abnormal linkage parameter combinations, abnormal linkage types, associated diseases, pathological mechanisms, and levels of evidence. Based on preset screening criteria, the historical clinical diagnosis and treatment data mapped by parameters are screened a second time to obtain the original diagnosis and treatment dataset; the original diagnosis and treatment dataset includes abnormal linkage parameter combination codes, intervention methods, parameter change rates, prognostic grading, and follow-up periods. The original clinical dataset is validated based on a preset validation rule base. After successful validation, a valid clinical dataset is output. The valid clinical dataset includes raw sensor data, clinical symptom recording time, diagnostic and treatment intervention information, and prognostic data. Calculate the treatment effectiveness weights based on the aforementioned effective clinical dataset; A set of clinical expert consensus nodes was constructed. Based on the scores of multidisciplinary experts on each combination in the initial table of abnormal linkages of the physiological parameters, different dimensions were scored, and the average score of different dimensions was taken as the expert consensus weight. Define the weight coefficient for each level of evidence to obtain the weight of each level of evidence; The comprehensive pathological weight is determined based on the weight of treatment effectiveness, expert consensus, and evidence level. The comprehensive pathological weight is then associated with the corresponding combination in the initial table of physiological parameter abnormality linkage, generating a physiological parameter abnormality linkage-pathological weight library. Based on the clinical symptom recording time and treatment intervention time in the effective clinical dataset, the disease course is divided into stages according to disease progression. The effective clinical dataset is grouped based on the disease course stage and abnormal linkage combination. The 95% confidence interval of the abnormal parameters in each group is calculated as the initial threshold of the group. The pre-built and continuously maintained calibration event library is called, and the initial threshold is calibrated based on the threshold adjustment cases in the calibration event library to obtain the abnormal threshold table of parameters by disease course. Based on a single record in a valid clinical dataset, and combined with a table of abnormal thresholds for disease course parameters and a preliminary table of abnormal linkages of physiological parameters, sample units are constructed; a clinical data sample library is then built based on these sample units.

[0046] In this embodiment, the abnormal linkage logic of abnormal parameters of the sensor is sorted out based on the pathological mechanism of the disease to determine the abnormal linkage combination and abnormal linkage type. Among them, the abnormal linkage type includes direct causal linkage, synergistic abnormal linkage and compensatory linkage. Direct causal linkage, for example: acute myocardial infarction → heart rate > 120 beats / min + blood oxygen saturation < 90% + supine position → insufficient tissue perfusion; synergistic abnormal linkage, for example: severe infection → core body temperature > 38.5℃ + respiratory rate > 25 breaths / min + heart rate > 110 beats / min → systemic inflammatory response; compensatory linkage, for example: dehydration → heart rate > 105 beats / min + core body temperature > 37.8℃ + decreased urine output (indirect correlation) → insufficient blood volume.

[0047] In this embodiment, the historical clinical diagnosis and treatment data mapped by parameters are screened a second time based on preset screening conditions to obtain the original diagnosis and treatment dataset. The preset screening conditions are: to screen complete clinical records that include sensor parameter linkage combinations, diagnosis and treatment interventions and prognostic results to obtain the original diagnosis and treatment dataset.

[0048] In this embodiment, the treatment effectiveness weight is calculated based on the effective clinical dataset. For the same abnormal linkage combination, the weight is calculated as "number of effective intervention cases / total number of cases × 10". For example, if the effective cases of a certain combination account for 82%, the treatment effectiveness weight is 8.2.

[0049] In this embodiment, a set of clinical expert consensus nodes is constructed. Based on the scores of multidisciplinary experts on each combination in the initial table of abnormal linkages of physiological parameters, different dimensions are scored. The average score of different dimensions is taken as the expert consensus weight. The different dimensions include the clarity of pathological mechanism, clinical early warning value, and urgency of diagnosis and treatment intervention. The average score of the three items, namely the clarity of pathological mechanism, the clinical early warning value, and the urgency of diagnosis and treatment intervention, is taken as the expert consensus weight. For example, if the scores of the three items for a certain combination are 9, 8, and 10, then the expert consensus weight is 9.0.

[0050] In this embodiment, the comprehensive pathological weight is determined based on the weight of treatment effectiveness, the weight of expert consensus, and the weight of evidence level; wherein, the comprehensive pathological weight = treatment effectiveness weight × 0.5 + expert consensus weight × 0.3 + evidence level weight × 0.2.

[0051] In this embodiment, the original diagnosis and treatment dataset is validated based on a preset validation rule base. The preset validation rule base includes integrity validation, time consistency validation, and instrument credibility validation. Integrity validation requires that the linkage combination contains ≥2 sensor parameter data and the single parameter missing rate is ≤10%. Time consistency validation requires that the difference between the sensor data acquisition time and the clinical symptom recording time is ≤2 hours. Instrument credibility validation calculates a risk value, and data with a risk value ≥0.8 is removed, while data with a risk value of 0.5-0.7 is marked as low credibility.

[0052] In this embodiment, based on the clinical symptom recording time and treatment intervention time in the effective clinical dataset, the disease course is divided into stages according to the disease progression. Specifically, the disease progression is divided into acute onset period (0-72 hours), stable period (72 hours-7 days), and recovery period (>7 days), and the disease course stage is marked for each record in the effective clinical dataset.

[0053] In this embodiment, the effective clinical dataset is grouped based on the disease stage and abnormal linkage combination, and the 95% confidence interval of the abnormal parameters in each group is used as the initial threshold for that group. For example, in the acute phase, the combination of "heart rate ↑ + arterial blood oxygen saturation ↓" has a heart rate abnormality threshold of >115 beats / minute, and in the stable phase, it is >105 beats / minute.

[0054] In this embodiment, a pre-built and continuously maintained calibration event library is invoked. The initial threshold is calibrated based on the threshold adjustment cases in the calibration event library to obtain a table of abnormal thresholds for parameters at different disease stages. The pre-built and continuously maintained calibration event library is pre-built based on the collection of "threshold adjustment cases" in clinical practice (such as the heart rate threshold of elderly patients in the acute phase of an acute attack needs to be lowered to >110 beats / minute after clinical verification), and records "case details + adjustment basis + expert review opinions". The thresholds are updated quarterly based on the calibration event library. The update rule is: if there are ≥5 cases of the same threshold adjustment and the expert review pass rate is ≥80%, then the abnormal threshold of the corresponding disease stage is adjusted, and the initial table of abnormal physiological parameter linkage is updated simultaneously.

[0055] In this embodiment, a sample unit is constructed based on a single record in the effective clinical dataset, combined with a table of abnormal thresholds for disease course parameters and a preliminary table of abnormal linkages for physiological parameters. Each sample unit has seven core fields: a unique sample code, a core parameter group, an abnormal linkage label, a pathological mechanism label, a treatment intervention label, a treatment effectiveness label, and a data credibility level. The unique sample code adopts the format of "disease type-linkage combination code-disease course stage-serial number". The core parameter group is filled with the original sensor data and the standardized data. The abnormal linkage label is matched from the preliminary table of abnormal linkages for physiological parameters. The pathological mechanism label is associated with the pathological mechanism of the corresponding linkage. The treatment intervention label and the treatment effectiveness label are filled using a combination of automatic filling and manual supplementation. Manually supplemented data must be checked by two people.

[0056] The working principle and beneficial effects of the above technical solution are as follows: By determining the core dimensions and binding corresponding sensor information, a sensor clinical data dictionary is formed, realizing standardized management of various physiological parameter data, laying the foundation for subsequent data processing and analysis, and facilitating unified parameter mapping and screening of historical clinical diagnosis and treatment data; clarifying the abnormal linkage logic of abnormal parameters, identifying abnormal linkage combinations and types, and marking the supporting evidence level helps to deeply understand the relationship between physiological parameters and the underlying pathological mechanisms, providing more comprehensive information for clinical diagnosis and disease monitoring; after secondary screening and verification of the rule base, an effective clinical dataset is obtained, ensuring the accuracy and reliability of the data and improving the scientific nature of subsequent analysis and decision-making; comprehensively considering... The weighting of treatment effectiveness, expert consensus, and evidence level determines the overall pathological weight, making the assessment of the correlation between abnormal physiological parameters and pathological relationships more objective and comprehensive, and better reflecting the actual clinical situation. Effective clinical datasets are grouped according to disease stage and initial thresholds are determined. These are then calibrated using a calibration event database to obtain a disease-stage parameter abnormality threshold table, providing more accurate criteria for judging parameter abnormalities for patients at different disease stages. A clinical data sample library is constructed based on effective clinical data and relevant rules, providing rich and accurate data support for comprehensive monitoring of hospitalized patients' physiological status, automatic alerts for abnormalities, and intelligent dialogue, helping to improve the efficiency and quality of medical monitoring and assisting medical staff in making more timely and effective treatment decisions.

[0057] The calculation of the incidence rate of abnormal linkages of different parameters and the correlation of disease deterioration under various scenarios includes: According to the preset monitoring time step, obtain the patient identifier set in the target scene where the corresponding parameter combination has abnormal linkage within each preset monitoring time step; based on the preset total duration of the monitoring time window, perform a union operation on the patient identifier sets within all preset monitoring time steps to obtain the deduplicated patient set where the parameter combination has abnormal linkage throughout the entire monitoring cycle. Obtain the total number of effective monitored patients in various scenarios from the clinical data sample library, and calculate the incidence of abnormal linkage of different parameters in various scenarios based on the ratio of the number of elements in the deduplicated patient set to the total number of effective monitored patients. ; in, Let be the occurrence rate of abnormal linkage of the p-th parameter combination in the s-th scenario; The number of deduplicated patients who experience abnormal linkage of the p-th parameter combination under the s-th scenario within the monitoring period; The total number of patients in the s-th scenario of the clinical data sample library for effective monitoring; Based on the incidence of abnormal linkages of different parameters in various scenarios, the correlation degree of disease deterioration corresponding to different parameter combinations in various scenarios is calculated. ; in, Let be the correlation between the p-th parameter combination and the q-th abnormal event in the s-th scenario; The number of parameters in the p-th parameter combination; The abnormal severity weight of the i-th parameter in the p-th parameter combination is determined based on the threshold of the current disease stage; This is the duration coefficient of the abnormal linkage in the q-th abnormal event; Let be the clinicopathological correlation coefficient for the s-th scenario; Let be the initial weight coefficients for the p-th parameter combination in the s-th scenario.

[0058] In this embodiment, existing technologies suffer from the problem that when the same patient experiences multiple abnormal linkages with the same parameter combination within a monitoring period, they are easily counted repeatedly, leading to an inflated incidence rate that fails to reflect the true clinical probability. Furthermore, the technology fails to differentiate between sample characteristics of different associated scenarios, including patients from non-target scenarios in the statistics, causing the incidence rate calculation to deviate from clinical reality. To address these issues, this embodiment utilizes the core logic of clinical sample deduplication and scenario-specific screening to quantify the incidence rate through two key steps: First, patient identifiers are collected according to a preset monitoring time step, and duplicate records of multiple abnormal linkages from the same patient are removed using a union operation to obtain a deduplicated patient set. Second, the total number of effective samples matching the target scenario in the clinical data sample library is anchored, and the incidence rate is calculated using the ratio of "number of deduplicated abnormal samples / number of effective samples in the scenario." The core design aims to avoid interference from multiple abnormalities in a single patient on the statistical results, while ensuring a high degree of compatibility between the sample library and the pathological mechanisms of the scenario, thus guaranteeing statistical objectivity. This design effectively avoids numerical distortion caused by repeated statistics, truly reflects the probability of abnormal linkage of a certain parameter combination in the target scenario, and improves the reference value of the data by specifically screening the total number of patients under effective monitoring for the scenario, so that the incidence rate is highly matched with the clinical pathological scenario.

[0059] In this embodiment, for a multi-parameter collaborative deterioration scenario, the following configuration is provided. =1.0, strengthening the inherent association weight between this scenario and disease deterioration; for the position and vital sign matching scenario for elderly patients, since parameter fluctuations are more significant but the risk of deterioration is relatively low when elderly patients change position, it can be configured to... =0.7, reduce the empirical weight; for acutely ill patients, the severity weight of parameter abnormalities is adjusted. The value can be appropriately increased (e.g., severe anomaly = 4) to improve the correlation ratio of severe abnormal events.

[0060] In this embodiment, the abnormal linkage duration coefficient is assigned based on clinical diagnosis and treatment experience. For example, a linkage duration of ≥1 hour is assigned a value of 1, and a duration of <1 hour is assigned a value of 0.5. In clinical practice, the longer the abnormal physiological parameter duration, the higher the risk of disease deterioration, which is an important dimension for modifying the correlation strength.

[0061] In this embodiment, the clinical pathological correlation coefficient is assigned based on the pathophysiological mechanism, ranging from 0.5 to 1.0: the correlation is highest in the multi-parameter synergistic deterioration scenario, assigned a value of 1.0; the cardiopulmonary linkage scenario is next, assigned a value of 0.8 to 0.9; the inherent pathological correlation strength between different scenarios and disease deterioration is corrected to avoid bias in the correlation comparison between different scenarios.

[0062] In this embodiment, the initial weighting coefficient Values ​​are assigned based on the clinical experience of experts, ranging from 0.6 to 1.5, and are fine-tuned for specific clinical scenarios / parameter combinations to adapt to special situations in actual clinical diagnosis and treatment (such as the correlation between abnormal body temperature and vital signs in elderly patients).

[0063] In this embodiment, The severity weight of the i-th parameter in the p-th parameter combination is determined based on the threshold of the current disease stage. For example, mild = 1, moderate = 2, severe = 3, which aligns with the clinical understanding of the severity of different physiological parameter abnormalities. k is the number of parameters in the parameter combination. The average value is obtained by summing the severity weights of all parameters to eliminate the influence of differences in the number of parameters on the severity assessment. This reflects the pathological logic that the more severe the parameter abnormality, the higher the probability of association with disease deterioration.

[0064] In this embodiment, existing technologies assess risk solely based on incidence rate or single parameter anomalies, neglecting the risk differences between mild and severe anomalies, and between transient fluctuations and persistent anomalies, resulting in low accuracy in early warning. Furthermore, practical applications do not consider the pathological characteristics of different associated scenarios or the differences in parameter fluctuations among different patient groups. To address these issues, this embodiment uses incidence rate as the underlying basis, sequentially superimposing four correction dimensions: average severity of parameter anomalies, duration coefficient of linkage, scenario-pathological correlation coefficient, and clinical experience weight, to achieve a comprehensive quantification of the strength of the correlation between abnormal linkage and disease deterioration. The core design is to balance incidence frequency and clinical risk level, adapting to the clinical heterogeneity of different disease courses, scenarios, and populations. This embodiment achieves a comprehensive assessment of incidence frequency, severity, duration, pathological characteristics, and clinical experience through multi-dimensional weighted fusion, avoiding the one-sidedness of single-dimensional judgment. Through dynamic configuration of the scenario's clinical-pathological correlation coefficient, it can adapt to five types of associated scenarios and special populations (such as elderly patients and critically ill patients), improving the personalization and targeting of early warnings.

[0065] The working principle and beneficial effects of the above technical solution are as follows: By acquiring the set of abnormal linkage patient identifiers according to the monitoring time sequence step, performing a union operation to obtain a deduplicated patient set, and combining it with the total sample size of effectively monitored patients, the incidence rate of abnormal linkage of different parameters under various scenarios can be accurately calculated, providing objective data support for assessing the occurrence and trend of diseases; by acquiring the set of abnormal linkage patient identifiers according to the monitoring time sequence step, performing a union operation to obtain a deduplicated patient set, and combining it with the total sample size of effectively monitored patients, the incidence rate of abnormal linkage of different parameters under various scenarios can be accurately calculated, providing objective data support for assessing the occurrence and trend of diseases; by acquiring the set of abnormal linkage patient identifiers according to the monitoring time sequence step, performing a union operation to obtain a deduplicated patient set, and combining it with the total sample size of effectively monitored patients, the incidence rate of abnormal linkage of different parameters under various scenarios can be accurately calculated, providing objective data support for assessing the occurrence and trend of diseases; by fully utilizing the data in the clinical data sample library, the value behind the data is mined, and the data is transformed into information that is of guiding significance for clinical practice, improving the practicality of monitoring equipment and the utilization efficiency of medical resources.

[0066] 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 high-voltage switchgear 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 high-voltage switchgear.

[0067] 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. Inpatient wearable monitoring devices that integrate comprehensive monitoring, automatic reminders, and dialogue, including: The wearable main body (1), and the multi-sensor acquisition module and the graded early warning module integrated in the wearable main body (1), and the patient interaction terminal (7) integrated on the outside of the wearable main body (1), are characterized in that: The multi-sensor acquisition module consists of a heart rate sensor (2), a blood oxygen saturation sensor (3), a respiratory rate sensor (4), a core body temperature sensor (5), and a body position sensor (6), and is configured to continuously and dynamically acquire the patient's key vital signs parameters and body position status, and output the raw acquisition data. The graded early warning module is configured to compare and analyze the preprocessed data with a preset normal parameter threshold range, and combine multi-parameter association rules to determine whether the data is abnormal and the level of abnormality. The level of abnormality includes normal, slight abnormality, moderate abnormality and severe abnormality, and triggers corresponding response actions according to the level of abnormality. The patient interaction terminal (7) is configured to receive and send tiered early warning reminders to the patient, and supports dialogue and interaction between the patient and the device as well as status feedback.

2. The inpatient wearable monitoring device integrating comprehensive monitoring, automatic reminders, and dialogue as described in claim 1, characterized in that, The tiered early warning module includes: The rule-building unit is configured to be based on standard physiological and pathophysiological knowledge and historical clinical diagnosis and treatment data, and to sort out the clinical pathological mechanisms of abnormal linkage of various physiological parameters and establish a clinical data sample library. Based on clinical pathological mechanisms and sample database data, we analyzed common abnormal physiological parameter association scenarios in patients, identified the core associated parameter combinations in various scenarios, and divided them into five categories of associated scenarios, including: cardiopulmonary parameter linkage scenario, body temperature and vital signs coordination scenario, body position and vital signs matching scenario, multi-parameter coordinated deterioration scenario, and parameter fluctuation trend association scenario. Statistical analysis methods were used to calculate the incidence of abnormal linkages of different parameters and the correlation of disease deterioration under various scenarios; combined with clinical diagnosis and treatment experience, association rules and their initial weight coefficients for various scenarios were defined.

3. The inpatient wearable monitoring device integrating comprehensive monitoring, automatic reminders, and dialogue as described in claim 2, characterized in that, Define association rules and their initial weight coefficients for various scenarios, including: Define the judgment conditions and corresponding weighting coefficients for abnormal linkage of heart rate, respiratory rate and blood oxygen saturation in the cardiopulmonary parameter linkage scenario; Define the linkage mode of body temperature, heart rate, and respiratory rate in the scenario of coordinated body temperature and vital signs, and set the weight enhancement coefficient of the grade and the corresponding abnormality level mapping according to the magnitude and duration of body temperature rise and the severity of accompanying vital signs. Establish a baseline for the safe duration of different body positions in a body position and vital sign matching scenario, and define the judgment logic and weight enhancement mechanism for abnormal body positions linked to heart rate fluctuations and blood oxygen decline. Define a judgment logic based on the combination of the number and severity of parameter anomalies in a multi-parameter collaborative deterioration scenario, and set a high-weight boosting coefficient for it; Define a quantitative standard for a continuously deteriorating trend in the context of parameter fluctuation trend correlation, and set a weighting factor for forward-looking early warning.

4. The inpatient wearable monitoring device integrating comprehensive monitoring, automatic reminders, and dialogue as described in claim 3, characterized in that, The association rules for various scenarios and their initial weight coefficients are defined, and this also includes: The established association rules and weighting coefficients are applied to clinical sample data and pilot patient monitoring. Data on the accuracy of anomaly detection, false alarm rate, and false negative rate are collected after application. The results are compared with the actual clinical treatment outcomes to analyze the rule suitability. For false alarm and false negative scenarios, optimize the accuracy of scenario decomposition and weight coefficient assignment, and iteratively update the content of association rules until the accuracy of rule anomaly judgment meets the requirements of clinical monitoring. The iteratively optimized association rules and weight parameters are embedded into the control program of the hierarchical early warning module to establish a real-time rule invocation mechanism; When receiving processed data, quickly match the corresponding related scenarios and call the corresponding weight coefficients to complete the anomaly level determination; Establish personalized adjustment rules and preset weight coefficient adjustment ranges for hospitalized patients of different ages, diseases, and severity of illness; Based on the individual treatment plans of the monitored patients, the rules and their weights are adjusted in a personalized manner, and the adjusted association rules and weight parameters are saved.

5. The inpatient wearable monitoring device according to claim 1, which integrates comprehensive monitoring, automatic reminders, and dialogue, is characterized in that... Trigger corresponding response actions based on the anomaly level, including: When the anomaly level is normal, no alert action will be initiated, and the processed data will be uploaded to the central monitoring system for storage and archiving. When the abnormality level is minor, a reminder instruction is sent to the patient interaction terminal (7) to guide the patient to adjust their position or state on their own through vibration and voice prompts, while recording the reminder time and data information. When the abnormality level is moderate, instructions are sent to the patient interaction terminal (7) and the medical staff mobile terminal simultaneously. The patient interaction terminal (7) triggers vibration and voice reminders, and the medical staff mobile terminal receives text warning information, including abnormal parameters, abnormality level and patient identity information. When the abnormality level is severe, the wearable main body (1) will immediately trigger an audible and visual alarm, send an emergency pop-up command to the medical mobile terminal, send an alarm command to the central monitoring system, and simultaneously record the time of the abnormality, the value of the abnormal parameters, the abnormality level and the patient's position, form an abnormal event report and upload it for archiving.

6. The inpatient wearable monitoring device according to claim 1, which integrates comprehensive monitoring, automatic reminders, and dialogue, is characterized in that... The tiered early warning module also includes: The data processing module is configured to use an adaptive Kalman filter algorithm to filter out motion interference noise, electromagnetic interference noise and baseline drift interference in the raw acquired data, and retain the valid data. A multi-sensor fusion calibration algorithm is used to correct the deviation of the filtered effective data by combining the patient's basic physiological information and the historical benchmark data of the sensors.

7. The inpatient wearable monitoring device integrating comprehensive monitoring, automatic reminders, and dialogue as described in claim 1, characterized in that, The wearable main body (1) also integrates a power management module (8) that supplies power to each module, and a communication module (9) that enables data interaction and command transmission.

8. The inpatient wearable monitoring device according to claim 1, which integrates comprehensive monitoring, automatic reminders, and dialogue, is characterized in that... The wearable main body (1) is a vest-style design made of flexible and breathable material. Three Velcro straps (10) are provided around the front of one side and the inside of the other side of the wearable main body (1) so that the tightness of the fit can be adjusted according to the patient's body shape.

9. The inpatient wearable monitoring device integrating comprehensive monitoring, automatic reminders, and dialogue as described in claim 2, characterized in that, The rule-building unit is based on standard physiological and pathophysiological knowledge and historical clinical diagnosis and treatment data, and sorts out the clinical pathological mechanisms of abnormal linkages of various physiological parameters to establish a standardized clinical data sample library, including: Based on standard physiological and pathophysiological knowledge and clinical guidelines, four core dimensions were identified: circulatory system, respiratory system, metabolic system, and neuro-motor regulation system. Each core dimension was bound to the corresponding sensor's acquisition parameters, units, accuracy, and initial abnormality judgment criteria, resulting in a sensor clinical data dictionary. Historical clinical diagnosis and treatment data are acquired, and parameter mapping and preliminary screening are performed on the historical clinical diagnosis and treatment data based on the sensor clinical data dictionary to identify abnormal parameters collected by the sensors. An expert working group was formed to analyze the abnormal linkage logic of the abnormal parameters based on the disease pathology mechanism, and to determine the abnormal linkage combinations and types. The expert working group, based on the GRADE standard, labeled each abnormal linkage combination with a supporting evidence level and determined the evidence level. An initial table of abnormal linkage relationships of physiological parameters was constructed based on abnormal linkage parameter combinations, abnormal linkage types, associated diseases, pathological mechanisms, and levels of evidence. Based on preset screening criteria, the historical clinical diagnosis and treatment data mapped by parameters are screened a second time to obtain the original diagnosis and treatment dataset; the original diagnosis and treatment dataset includes abnormal linkage parameter combination codes, intervention methods, parameter change rates, prognostic grading, and follow-up periods. The original clinical dataset is validated based on a preset validation rule base. After successful validation, a valid clinical dataset is output. The valid clinical dataset includes raw sensor data, clinical symptom recording time, diagnostic and treatment intervention information, and prognostic data. Calculate the treatment effectiveness weights based on the aforementioned effective clinical dataset; A set of clinical expert consensus nodes was constructed. Based on the scores of multidisciplinary experts on each combination in the initial table of abnormal linkages of the physiological parameters, different dimensions were scored, and the average score of different dimensions was taken as the expert consensus weight. Define the weight coefficient for each level of evidence to obtain the weight of each level of evidence; The comprehensive pathological weight is determined based on the weight of treatment effectiveness, expert consensus, and evidence level. The comprehensive pathological weight is then associated with the corresponding combination in the initial table of physiological parameter abnormality linkage, generating a physiological parameter abnormality linkage-pathological weight library. Based on the clinical symptom recording time and treatment intervention time in the effective clinical dataset, the disease course is divided into stages according to disease progression. The effective clinical dataset is grouped based on the disease course stage and abnormal linkage combination. The 95% confidence interval of the abnormal parameters in each group is calculated as the initial threshold of the group. The pre-built and continuously maintained calibration event library is called, and the initial threshold is calibrated based on the threshold adjustment cases in the calibration event library to obtain the abnormal threshold table of parameters by disease course. Based on a single record in a valid clinical dataset, and combined with a table of abnormal thresholds for disease course parameters and a preliminary table of abnormal linkages of physiological parameters, sample units are constructed; a clinical data sample library is then built based on these sample units.

10. The inpatient wearable monitoring device according to claim 2, which integrates comprehensive monitoring, automatic reminders, and dialogue, is characterized in that... The calculation of the incidence rate of abnormal linkages of different parameters and the correlation of disease deterioration under various scenarios includes: According to the preset monitoring time step, obtain the patient identifier set in the target scene where the corresponding parameter combination has abnormal linkage within each preset monitoring time step; based on the preset total duration of the monitoring time window, perform a union operation on the patient identifier sets within all preset monitoring time steps to obtain the deduplicated patient set where the parameter combination has abnormal linkage throughout the entire monitoring cycle. Obtain the total number of effective monitored patients in various scenarios from the clinical data sample library, and calculate the incidence of abnormal linkage of different parameters in various scenarios based on the ratio of the number of elements in the deduplicated patient set to the total number of effective monitored patients. ； in, Let be the occurrence rate of abnormal linkage of the p-th parameter combination in the s-th scenario; The number of deduplicated patients who experience abnormal linkage of the p-th parameter combination under the s-th scenario within the monitoring period; The total number of patients in the s-th scenario of the clinical data sample library for effective monitoring; Based on the incidence of abnormal linkages of different parameters in various scenarios, the correlation degree of disease deterioration corresponding to different parameter combinations in various scenarios is calculated. ； in, Let be the correlation between the p-th parameter combination and the q-th abnormal event in the s-th scenario; The number of parameters in the p-th parameter combination; The abnormal severity weight of the i-th parameter in the p-th parameter combination is determined based on the threshold of the current disease stage; This is the duration coefficient of the abnormal linkage in the q-th abnormal event; Let be the clinicopathological correlation coefficient for the s-th scenario; Let be the initial weight coefficients for the p-th parameter combination in the s-th scenario.

Images