Intelligent monitoring device for anesthesia recovery period

Abstract

The invention discloses an intelligent monitoring device for an anesthesia recovery phase in the technical field of medical instruments, and the intelligent monitoring device comprises a sensor patch, a repeater, a ward monitoring assembly, a cloud storage platform and a physical sign monitoring platform. The fitting state and the deviation amount of the sensor paste and the body surface of a patient are monitored, so that the deformation of the electric response piece is controlled to drive the sensor paste to compensate the deviation amount, the vital sign data sensor on the sensor paste is automatically moved to a standard position, the monitoring precision of vital sign data is improved, the situation that medical staff repeatedly adjust the position of the sensor is reduced, and the medical staff experience is improved. And more real data is provided for subsequent risk assessment.

Description

Technical Field

[0001] The present invention belongs to the technical field of medical devices, and in particular is an intelligent monitoring device for anesthesia recovery period. Background Art

[0002] Post-anesthesia monitoring and treatment (PATS) refers to the medical care provided to patients recovering from surgery during anesthesia. Critically ill patients are admitted directly to the intensive care unit (ICU) for recovery. PAT focuses on restoring the patient's protective reflexes, monitoring and treating any physiological disturbances, ensuring stable vital signs, identifying and promptly addressing anesthetic and post-operative complications, and reducing morbidity and mortality. PAT generally does not exceed 24 hours.

[0003] Post-anesthesia monitoring and treatment requires a bedside monitor that can monitor pulse oximetry, ECG, non-invasive blood pressure, end-tidal carbon dioxide, muscle relaxation, body temperature, and other functions. Special monitoring equipment such as invasive pressure monitoring (direct arterial pressure measurement, central venous pressure measurement), intracranial pressure monitoring, and cardiac output measurement can be equipped as needed. Monitoring equipment must be in standby mode, and sufficient portable monitors must be available for patient transport. Key monitoring parameters such as ECG (electrocardiogram), SPO2 (blood oxygen saturation), BP (blood pressure), Temp (body temperature), and RR (respiration) are crucial for accurately assessing a patient's anesthesia recovery risk and awakening status.

[0004] Existing intelligent monitoring devices for the anesthesia recovery period often focus on integrating sensors for monitoring patient vital signs data to improve the portability of monitoring instruments, or utilize deep learning algorithm models to accurately assess a patient's recovery status and anesthesia recovery risk based on vital sign data detected by sensors, as shown in patent document publication number US7920914B2. However, adaptive optimization of sensor detection accuracy during anesthesia recovery monitoring is neglected. This can easily lead to unreliable results in subsequent monitoring of patient vital signs data and calculation and assessment of patient recovery status and recovery risk using patient vital sign data. Missing or discontinuous data may delay the optimal rescue opportunity. Even if the monitoring sensors have the function of alerting medical staff to abnormal monitoring positions, medical staff are required to accurately reinstall these sensor devices and ensure their data collection status is stable. The installation and adjustment process is complicated and inconvenient, increasing the workload of medical staff.

[0005] To this end, it is necessary to propose an intelligent anesthesia recovery period monitoring device that can automatically monitor the position of anesthesia recovery period monitoring sensor on the patient's body surface, calculate the offset and automatically compensate for the offset based on the preset marks on the anatomical model, thereby improving the monitoring accuracy of the anesthesia recovery period monitoring sensor on the patient's vital signs data, providing real data for the subsequent assessment of the patient's anesthesia recovery risk and awakening status, and reducing the workload of medical staff. Summary of the Invention

[0006] In order to solve the above problems, the purpose of the present invention is to provide an intelligent monitoring device for the anesthesia recovery period. Through the infrared ranging sensor set on the sensor patch and the optical sensor of the ward monitoring component, the fit state and deviation between the sensor patch and the patient's body surface are monitored, thereby controlling the deformation of the electric response part to drive the sensor patch to compensate for the offset, automatically moving the vital signs data sensor on the sensor patch to the standard position, improving the monitoring accuracy of vital signs data, reducing the need for medical staff to repeatedly adjust the sensor position, and providing more realistic data for subsequent risk assessment.

[0007] In order to achieve the above objectives, the technical solution of the present invention is as follows: an intelligent monitoring device for anesthesia recovery period, comprising a sensor patch, a repeater, a ward monitoring component, a cloud storage platform and a vital sign monitoring platform,

[0008] The sensor patch is used to monitor vital signs of patients recovering from anesthesia, including electrocardiogram (ECG), blood oxygen saturation, blood pressure, body temperature, and respiration. The patch is magnetically positioned to mark the location on the patient's body surface. The patch detects the gap between the patch and the patient's skin using infrared ranging and changes the contact between the patch and the patient's skin using electro-deformation response.

[0009] A repeater, used to relay the detection data signal and control signal of the sensor patch, and transmit the above detection data signal and control signal between the sensor patch and the ward monitoring component through wireless transmission;

[0010] The ward monitoring component is used to display and upload detected patient vital sign data to a cloud storage platform. It uses optical sensors to collect patient surface contour image data, matches it to a preset anatomical model library, uses magnetic field sensing positioning sensors to detect the position of sensor patches, calculates the offset between the sensor patches and surface anatomical landmarks, and controls the deformation compensation offset of the sensor patches based on a preset offset threshold. When a time series prediction model trained with complete vital sign time series data detects signal loss or noise interference, it uses historical data to predict missing segments and cross-validates the rationality of the predicted data. Sensor patch failures are handled in a hierarchical manner based on a multi-channel fault classification response algorithm.

[0011] a. When local signal interference occurs, the time series prediction model automatically predicts data and cross-validates;

[0012] b. When the sensor patch has poor contact, control the deformation of the sensor patch to make it close or offset, and change its relative position to the patient's skin to optimize the monitoring site;

[0013] c. If the hardware is damaged or there is no signal for 3 minutes, notify medical staff for maintenance.

[0014] The cloud storage platform is used to store long-term patient vital sign data and transmit the data to the vital sign monitoring platform. A multi-source data fusion algorithm is used to extract heart rate variability, blood oxygen fluctuation, and respiratory rate variability characteristics based on the patient's vital sign data to generate a recovery period risk assessment to assist medical staff in decision-making.

[0015] The vital signs monitoring platform is used to display the patient's vital signs data and curves, and to alert the medical staff on duty when the patient's vital signs data deviates from the normal value range.

[0016] The basic solution works as follows: The intelligent monitoring device for anesthesia recovery uses the magnetic positioning on the sensor patch to mark the location of the sensor patch on the patient's body surface, while also using an infrared ranging sensor to detect the gap between the sensor patch and the patient's skin. The ward monitoring component uses an optical sensor to collect image data of the patient's body contour and matches it with a preset anatomical model library to determine the exact location of the sensor patch on the patient's body surface. A magnetic field sensing positioning sensor detects the position of the sensor patch and calculates the offset between the sensor patch and the anatomical landmarks on the body surface. When the offset exceeds the preset offset threshold, the system controls the deformation of the sensor patch, changing the contact between the sensor patch and the patient's skin through electro-deformation response, automatically compensating for the offset and moving the vital sign data sensor on the sensor patch to the standard position to improve monitoring accuracy.

[0017] In terms of data processing, the ward monitoring component relies on a time-series prediction model trained with complete vital sign time-series data. When detecting signal loss or noise interference, it uses historical data to predict missing segments and cross-validate the rationality of the predicted data. Furthermore, a multi-channel fault-grading response algorithm is used to handle sensor patch failures: when localized signal interference occurs, the time-series prediction model automatically predicts data and cross-validates it. When sensor contact is poor, the sensor patch is deformed, either clinging to the patch or shifting its relative position to the patient's skin to optimize the monitoring site. When hardware is damaged or there is persistent no signal, medical staff are notified for maintenance.

[0018] The beneficial effects of the basic solution are: 1. By automatically monitoring the position of the sensor sticker and the patient's body surface and compensating for the offset, the vital signs data sensor is ensured to be in the standard position, thereby improving the monitoring accuracy of the vital signs data and providing more realistic data for the subsequent assessment of the patient's anesthesia recovery risk and awakening status.

[0019] 2. It avoids medical staff from frequently adjusting the position of the sensor, reduces the tedious steps of installation and adjustment, reduces the workload of medical staff, and improves the convenience of operation.

[0020] 3. When detecting signal loss or noise interference, the time series prediction model is used to predict the missing segment data and cross-validate it, ensuring the integrity and reliability of the data and avoiding delays in the best rescue time due to data omissions or discontinuities.

[0021] 4. Adopting multi-channel fault classification response algorithm, taking corresponding treatment measures according to different types of faults, effectively improving the stability and reliability of the system and ensuring the continuity of the monitoring process.

[0022] 5. The cloud storage platform stores long-term monitored patient vital signs data, extracts key features through multi-source data fusion algorithms, generates recovery period risk assessments, assists medical staff in decision-making, helps better manage the medical activities of patients recovering from anesthesia, and reduces the morbidity and mortality of patients.

[0023] Furthermore, the sensor stickers include electrocardiogram electrode sensor stickers, blood oxygen saturation optical sensor stickers, blood pressure impedance sensor stickers, body temperature sensor stickers and respiratory impedance sensor stickers, which respectively detect the above-mentioned corresponding vital signs data.

[0024] The basic solution offers the following benefits: the sensor patches include electrocardiogram (ECG) electrode sensor patches, blood oxygen saturation optical sensor patches, blood pressure impedance sensor patches, body temperature sensor patches, and respiratory impedance sensor patches. Each sensor patch detects specific vital sign data. This targeted design ensures accurate monitoring of the patient's electrocardiogram (ECG), blood oxygen saturation (SPO2), blood pressure (BP), body temperature (Temp), and respiration (RR). The collaborative operation of multiple sensor patches comprehensively covers the monitoring needs of key vital signs of patients during the anesthesia recovery period, providing medical staff with complete information on the patient's physiological status, facilitating the timely detection and treatment of possible complications after anesthesia and surgery, ensuring the stability of the patient's vital signs, reducing the morbidity and mortality of patients during the anesthesia recovery period, and providing highly accurate basic data for subsequent risk assessment and recovery analysis.

[0025] Furthermore, the sensor patch includes a base plate, which is circumferentially fixedly connected to a plurality of base strips, a deformable plate fixedly connected inside the base plate, and deformable strips fixedly connected inside the base strips. The deformable plate and the deformable strips are both made of electrically responsive materials.

[0026] The basic solution offers the following benefits: 1. When the sensor patch deviates from the preset anatomical landmark, the system controls the deformation of the deformable disk and strip, driving the entire sensor patch to move, thereby repositioning the vital sign data sensor attached to the patch to the standard position. This automatic position adjustment effectively reduces monitoring errors caused by sensor patch misalignment, improves the accuracy of vital sign data monitoring, and provides a more reliable data foundation for subsequent assessment of patient recovery status and recovery risk.

[0027] During the recovery period from anesthesia, the patient's body position may change, or the sensor patch may shift due to physical activity. The electrical response characteristics of the deformable disk and strip enable the sensor patch to respond to these changes in real time and dynamically compensate for their position. This prevents data loss or inaccuracy caused by sensor position shift, ensuring that medical staff can obtain complete vital sign information during the patient's recovery period and promptly identify potential complications.

[0028] 3. Because the sensor patch has an automatic position adjustment function, medical staff do not need to meticulously adjust its position during installation and use, reducing the number of steps and operational difficulty. The sensor patch in this solution can automatically compensate for offset, allowing medical staff to complete sensor patch installation and maintenance tasks more efficiently, freeing up more time and energy to focus on clinical care and treatment of patients, significantly reducing their workload and improving overall care efficiency.

[0029] 4. The rapid response of the electroresponsive material enables the sensor patch to quickly adjust its position and promptly restore accurate monitoring status. In the face of sudden monitoring signal anomalies or positional shifts, the system can quickly adjust by controlling the deformable disk and strip, avoiding monitoring interruptions or data anomalies caused by sensor position issues. This rapid self-adjustment and recovery capability enhances the stability and reliability of the entire monitoring system, ensuring uninterrupted vital sign monitoring for patients recovering from anesthesia and providing stronger assurance for postoperative patient safety.

[0030] Furthermore, the bottom walls of the deformable disks are fixedly connected with corresponding types of vital sign sensors, the bottom walls of the deformable disks are also fixedly connected with infrared ranging sensors, and the top walls of the deformable disks are fixedly connected with batteries and magnetic positioning tags.

[0031] The beneficial effects of the basic solution are: 1. The bottom wall of the deformation disk is fixedly connected with corresponding types of sensors and various electrical devices, which makes it convenient for the ward monitoring components to track and adjust the position of the sensor stickers in real time, realize automatic compensation of offset, and improve the integration level and functional diversity of the entire monitoring system.

[0032] 2. When the sensor patch is detected to be offset or in poor contact with the skin, the system controls the electrical response of the deformation disk and strip based on distance information fed back by the infrared ranging sensor, automatically adjusting the position of the sensor patch to reattach the sensor to the standard position, ensuring the continuity and reliability of the monitoring data. This synergistic effect of precise monitoring and automatic compensation effectively reduces data errors caused by sensor position offset or poor contact, providing high-quality data support for subsequent patient recovery risk assessment.

[0033] 3. The sensor, infrared ranging sensor, battery and magnetic positioning tag are centrally arranged on the bottom and top walls of the deformation disk, making the structure of the sensor sticker more compact and simplified, reducing the connection complexity between components, and improving the reliability and stability of the device.

[0034] Furthermore, the bottom wall of the base plate and the base strip are both covered with a bonding layer, which is made of biocompatible material. Several three-dimensional liquid diodes are opened in the base plate, base strip and bonding layer, and the bonding layer is used to be adhered to the patient's skin.

[0035] The basic solution offers the following benefits: 1. Both the base layer and the conforming layer are made of flexible, biocompatible materials. The conforming layer is applied to the patient's skin, ensuring a comfortable fit and reducing the risk of skin irritation and damage. The flexible material adapts to the contours and curvatures of individual patients' skin, ensuring a good fit even during recovery from anesthesia, when patients may be in various positions or undergo minor limb movements. This ensures greater comfort and wear stability.

[0036] 2. Several three-dimensional liquid diodes are embedded in both the base layer and the contact layer. This structural design helps optimize heat transfer and sweat removal between the sensor patch and the skin. This maintains close contact between the sensor and the skin, reduces signal interference and noise caused by poor contact, improves patient comfort when wearing the sensor patch for extended periods, and reduces the likelihood of skin allergic reactions.

[0037] 3. The design of the conforming layer simplifies the installation process of the sensor patch. Medical staff only need to adhere the conforming layer to the patient's skin, without the need for complex fixing devices or tedious operation steps. At the same time, the adaptability of the flexible base layer and the conforming layer makes the sensor patch suitable for patients of different ages and body types, including those with sensitive or fragile skin, such as the elderly and children, improving the versatility and applicability of the monitoring device.

[0038] Furthermore, the ward monitoring component includes a monitoring display, an optical sensor, an offset positioning unit, a compensation control unit and a data prediction unit.

[0039] A monitoring display for displaying real-time vital sign data transmitted by the repeater;

[0040] Optical sensors, including several ToF ranging sensors and cameras, are used to detect images and distance data of the patient's body contours;

[0041] An offset positioning unit is used to match the body surface contour image data of the patient detected by the optical sensor and the position of the sensor patch detected by the magnetic field induction positioning sensor, match the body surface anatomical landmarks on the preset anatomical model, and calculate the offset;

[0042] A compensation control unit, configured to control the deformation of the deformable disk on the sensor patch to reduce the distance between the sensor patch and the patient's skin when the offset reaches an offset threshold, and to control the deformable strip to curl and pull the sensor patch to change its relative position with the patient's skin;

[0043] The data prediction unit is used to predict data missing segments caused by sensor movement or interference using a time series prediction model with a self-attention mechanism, and to cross-validate the rationality of the predicted data.

[0044] The basic solution offers the following benefits: 1. The monitor in the ward monitoring unit displays real-time vital sign data transmitted by the repeater. Medical staff can access the patient's latest status through the monitor, including key indicators such as electrocardiogram (ECG), blood oxygen saturation, blood pressure, temperature, and respiration. This real-time data display helps medical staff detect abnormalities in the ward and respond quickly, ensuring patient safety during the anesthesia recovery period.

[0045] 2. The optical sensor integrates a Time of Flight (ToF) ranging sensor and camera to accurately detect images and distance data of the patient's body contours. The ToF ranging sensor provides highly accurate distance information, while the camera captures visual images of the body surface. The combination of the two provides rich data support for the offset positioning unit.

[0046] 3. Based on the offset calculated by the offset positioning unit, the compensation control unit precisely controls the deformation of the sensor patch's deformable disk when the offset reaches a preset threshold, reducing the distance between the sensor and the patient's skin. Simultaneously, the deformable strip is controlled to curl, pulling the sensor patch to change its relative position relative to the patient's skin. This precise control mechanism ensures the sensor patch automatically adjusts to the optimal position, restoring accurate monitoring status, reducing monitoring errors caused by positional offset, and improving the reliability of monitoring data.

[0047] 4. The data prediction unit utilizes a time series prediction model with a self-attention mechanism. This allows it to predict missing data segments when sensor movement or interference causes data loss. The self-attention mechanism better captures long-term dependencies in time series data, improving prediction accuracy. Furthermore, cross-validation of the predicted data ensures consistency with actual data, thus guaranteeing the integrity and continuity of vital sign data.

[0048] 5. The ward monitoring component reduces the frequency of medical staff manually adjusting the position of sensor stickers through automated offset positioning and compensation control functions, reduces operational complexity, and enables medical staff to devote more time and energy to patient care and treatment.

[0049] Furthermore, the offset positioning unit uses the measured landmark points to match the anatomical model landmark points, converts the measured body surface contour image data coordinate system to the anatomical model coordinate system, and calculates the offset between the magnetic positioning position and the ideal position based on the converted measured body surface contour image data coordinate system.

[0050] The basic solution offers the following benefits: 1. The offset positioning unit converts the coordinate system of the measured body contour image data to the anatomical model coordinate system. This coordinate conversion technology ensures that the measured landmarks accurately match the landmarks on the anatomical model, improving positioning accuracy and reliability. Through this precise coordinate conversion and matching, the offset positioning unit can more accurately calculate the offset between the actual and ideal positions of the magnetic positioning tags, providing a solid foundation for subsequent compensation control.

[0051] 2. By precisely calculating the offset, the compensation control unit can more accurately adjust the position of the sensor patch to ensure that the sensor patch is aligned with the anatomical landmarks on the patient's body surface, thereby reducing monitoring errors caused by sensor position deviation and improving the monitoring accuracy of vital sign data.

[0052] 3. Accurate offset calculation and automatic compensation ensure that the sensor patch is always in the optimal monitoring position, reducing the frequency of manual adjustment of the sensor patch by medical staff. This allows medical staff to devote more time and energy to other important patient care activities, optimize the overall care process, and improve care efficiency.

[0053] Furthermore, the cross-validation of the data prediction unit introduces other vital signs data, constrains the rationality of the predicted data through joint distribution, and introduces a discriminator to evaluate the realism of the predicted data. When the predicted data conflicts with the measured data and the physiological joint distribution constraints at the same time, the dynamic threshold verification triggers re-prediction.

[0054] The basic solution has the following benefits: 1. The data prediction unit innovatively incorporates data from other sensors during cross-validation, assessing the rationality of the predicted data through joint distribution constraints. This approach differs from traditional single-source predictions in that it fully leverages the data correlations between multiple sensors and the constraints between physiological parameters, making the predictions more accurate and reliable.

[0055] 2. A discriminator is introduced to assess the fidelity of predicted data, determining whether the predictions align with physiological characteristics based on both the details and overall trends of the data. This intelligent evaluation mechanism can identify potential abnormal patterns in the predicted data, further ensuring the quality of the predicted data, giving medical staff greater confidence in the predictions and reducing misjudgments due to data quality issues.

[0056] 3. When predicted data conflict with measured data and physiological joint distribution constraints, the system triggers a re-prediction through dynamic threshold verification. This dynamic adjustment mechanism optimizes the prediction process in real time based on actual monitoring conditions, enhancing the system's adaptability to complex and changing clinical environments. It promptly corrects prediction deviations, ensuring that predicted data remains within a reasonable range and providing a stable basis for subsequent medical decision-making.

[0057] Furthermore, the cloud storage platform includes a database and a multi-source data fusion unit.

[0058] A database for continuously storing real-time monitored patient vital signs data;

[0059] The multi-source data fusion unit is used to align and fuse vital sign data, extract heart rate variability, blood oxygen fluctuation and respiratory rate variability features, and generate a recovery period risk assessment.

[0060] The basic solution offers the following benefits: 1. The database can store long-term patient vital sign data, providing medical staff with a complete record of the patient's recovery period. This long-term storage not only facilitates real-time monitoring but also provides detailed historical data for subsequent medical analysis and review, enabling medical staff to trace the patient's condition over time and provide a basis for adjusting diagnosis and treatment plans.

[0061] 2. The multi-source data fusion unit aligns and fuses vital sign data from different sensors, comprehensively considering characteristics such as heart rate variability (HRV), blood oxygen fluctuations, and respiratory rate variability (RFV). This fusion technology integrates information from multiple physiological parameters, avoiding the potential bias of a single data source, generating a more comprehensive and accurate recovery risk assessment and providing more reliable decision support for medical staff.

[0062] Furthermore, the multi-source data fusion unit unifies the time base of the vital sign data by interpolating and aligning the timestamps between the vital sign data and aligning the periodic features. The XGBoost classifier is used to evaluate the key feature vectors after standardized fusion and output the risk level.

[0063] The beneficial effects of the basic solution are: 1. The multi-source data fusion unit aligns the timestamps of vital signs data through interpolation and aligns periodic features, effectively unifying the data time base, enabling the accurate fusion of vital signs data from different sources, avoiding data misalignment and error accumulation, and providing a more accurate basis for subsequent analysis.

[0064] 2. Use the XGBoost classifier to evaluate the standardized fused key feature vectors and output the risk level. This machine learning-based method can comprehensively consider the impact of multiple physiological characteristics on recovery risk, provide a more reliable quantitative risk assessment, and assist medical staff in promptly identifying high-risk patients and taking targeted intervention measures to reduce the incidence of adverse events in patients. BRIEF DESCRIPTION OF THE DRAWINGS

[0065] Figure 1 FIG. 4 is an axonometric diagram of the sensor patch in the standby state according to an embodiment of the present invention.

[0066] Figure 2 FIG. 4 is an axonometric diagram of the sensor patch in the deformed state according to an embodiment of the present invention.

[0067] Figure 3 FIG. 4 is a side cross-sectional view of the sensor patch in the standby state according to an embodiment of the present invention.

[0068] Figure 4 FIG. 4 is a cross-sectional view of the base layer of the sensor patch in an embodiment of the present invention.

[0069] Figure 5 2 is a cross-sectional view of a sensor bonding layer in an embodiment of the present invention.

[0070] Figure 6 Schematic diagram of the system of the intelligent monitoring device for anesthesia recovery period in an embodiment of the present invention.

[0071] Figure 7 Schematic diagram of the system operation of the intelligent monitoring device for anesthesia recovery period in an embodiment of the present invention.

[0072] The figure marks in the drawings of the specification include: 1. base plate; 2. bonding layer; 3. deformation strip; 4. deformation disk; 5. vital sign sensor; 6. magnetic positioning tag; 7. battery; 8. infrared ranging sensor; 9. base strip. DETAILED DESCRIPTION

[0073] The following is further described in detail through specific implementation methods:

[0074] Example 1

[0075] Basically as attached Figure 1 、 Figure 2 、 Figure 6 and Figure 7 Shown: An intelligent monitoring device for anesthesia recovery period, including sensor stickers, repeaters, ward monitoring components, cloud storage platform and vital signs monitoring platform,

[0076] The sensor patch is used to monitor vital signs of patients recovering from anesthesia, including electrocardiogram (ECG), blood oxygen saturation, blood pressure, body temperature, and respiration. The patch is magnetically positioned to mark the location on the patient's body surface. The patch detects the gap between the patch and the patient's skin using infrared ranging and changes the contact between the patch and the patient's skin using electro-deformation response.

[0077] A repeater, used to relay the detection data signal and control signal of the sensor patch, and transmit the above detection data signal and control signal between the sensor patch and the ward monitoring component through wireless transmission;

[0078] The ward monitoring component is used to display and upload detected patient vital sign data to a cloud storage platform. It uses optical sensors to collect patient surface contour image data, matches it to a preset anatomical model library, uses magnetic field sensing positioning sensors to detect the position of sensor patches, calculates the offset between the sensor patches and surface anatomical landmarks, and controls the deformation compensation offset of the sensor patches based on a preset offset threshold. When a time series prediction model trained with complete vital sign time series data detects signal loss or noise interference, it uses historical data to predict missing segments and cross-validates the rationality of the predicted data. Sensor patch failures are handled in a hierarchical manner based on a multi-channel fault classification response algorithm.

[0079] a. When local signal interference occurs, that is, the instantaneous amplitude of the vital sign signal output by the sensor, such as ECG and SpO2, exceeds the preset physiological range, such as ECG R wave amplitude >5mV or <0.1mV, or the signal-to-noise ratio falls below the set threshold, such as SNR <10dB, high-frequency noise accounts for more than 15% of the total signal, or consecutive data points are missing for more than 5 sampling cycles, the time series prediction model automatically predicts the data and cross-validates it;

[0080] b. When the sensor patch is in poor contact, that is, when the infrared distance measurement detects a gap between the sensor patch and the skin greater than 2mm, or when the fluctuation amplitude of the vital sign signal exceeds the physiological threshold, such as blood pressure fluctuation greater than 20mmHg / min or respiratory rate change greater than 5 times / minute, the sensor patch is deformed to move closer to or offset the patient's skin, changing its relative position to optimize the monitoring site;

[0081] c. When hardware is damaged, such as when an internal circuit detects a break, short circuit, or voltage abnormality, or when there is no signal for three minutes, or when the magnetic field sensing positioning tag signal is lost or cannot be matched with the preset anatomical model, medical staff will be notified for maintenance;

[0082] The ward monitoring component includes a monitoring display, an optical sensor, an offset positioning unit, a compensation control unit, and a data prediction unit;

[0083] A monitoring display is used to display real-time vital sign data transmitted by the repeater; an optical sensor, including several ToF ranging sensors and cameras, is used to detect images and distance data of the patient's body surface contour; an offset positioning unit is used to detect the position of the sensor patch based on the patient's body surface contour image data detected by the optical sensor and the magnetic field induction positioning sensor, match the body surface anatomical landmarks on the preset anatomical model, and calculate the offset; a compensation control unit is used to control the deformation of the sensor patch to reduce the distance between the sensor patch and the patient's skin when the offset reaches the offset threshold, or to curl and pull the sensor patch to change its relative position with the patient's skin; a data prediction unit is used to use a time series prediction model with a self-attention mechanism to predict data missing segments caused by sensor patch movement or interference, and cross-validate the rationality of the predicted data.

[0084] The offset positioning unit uses the measured landmark points to match the anatomical model landmark points, converts the measured body surface contour image data coordinate system to the anatomical model coordinate system, and calculates the offset between the magnetic positioning position and the ideal position based on the converted measured body surface contour image data coordinate system.

[0085] The cross-validation of the data prediction unit is to introduce other vital signs data, predict the rationality of the data through joint distribution constraints, and introduce a discriminator to evaluate the realism of the predicted data. When the predicted data conflicts with the measured data and the physiological joint distribution constraints at the same time, the dynamic threshold verification triggers re-prediction.

[0086] The cloud storage platform is used to continuously store real-time patient vital sign data and transmit the data to the vital sign monitoring platform. A multi-source data fusion algorithm is used to extract heart rate variability, blood oxygen fluctuation, and respiratory rate variability characteristics from the patient's vital sign data to generate a recovery period risk assessment to assist medical staff in decision-making.

[0087] The cloud storage platform consists of a database and a multi-source data fusion unit. The database stores long-term patient vital sign data. The multi-source data fusion unit aligns and fuses the vital sign data, extracting features such as heart rate variability, blood oxygen fluctuation, and respiratory rate variability to generate a recovery risk assessment. The multi-source data fusion unit unifies the vital sign data time base by interpolating and aligning the timestamps between the vital sign data and periodic features. The XGBoost classifier is then used to evaluate the standardized fused key feature vectors and output a risk level.

[0088] The vital signs monitoring platform is used to display the patient's vital signs data and curves, and to alert the medical staff on duty when the patient's vital signs data deviates from the normal value range.

[0089] The specific implementation process is as follows: Existing anesthesia recovery monitoring devices generally focus on integrating sensors and predicting patient vital signs data, but ignore the cumbersome installation and potential instability of the sensors themselves, resulting in inaccurate patient vital signs detection data and the need for medical staff to repeatedly adjust the sensor installation position, which is an adverse consequence.

[0090] Therefore, the present invention mainly combines the patient's vital sign sensor 5 with magnetic positioning technology. When the patient enters the anesthesia recovery ward, the medical staff installs the sensor sticker on the patient. The optical sensor in the ward monitoring component and the ToF ranging sensor obtain the three-dimensional point cloud data P of the patient's body surface contour. i ,y i , z i The camera detects anatomical landmarks (such as the suprasternal notch) and outputs bounding box coordinates. The sensor sticker coordinates are then compared with the ideal position in the pre-set anatomical model to calculate the offset.

[0091] The offset positioning unit transforms the measured point cloud P into the preset standard anatomical model coordinate system M and solves the rotation matrix R and translation vector T through singular value decomposition (SVD). The calculation formula is as follows:

[0092]

[0093] Where p i is the measured point, q i It is the anatomical landmark of the body surface model.

[0094] If the offset Δd is greater than 5 mm, the compensation control unit controls the electrical response of the sensor patch to deform, curl, and pull the position of the sensor patch on the skin so that it is close to the standard detection position, without the need for manual adjustment by medical staff.

[0095] Since the sensor patch needs to be fixed on the patient's body surface for a long time, it may not be firmly attached, resulting in inaccurate detection data. Therefore, the compensation control unit can also detect the distance between the vital sign sensor 5 and the patient's skin through the sensor patch. When the distance reaches the threshold, the vital sign sensor 5 is moved closer to the patient's skin through electrical response deformation, thereby enhancing the continuity and stability of the detection data.

[0096] The stability of the patient's vital signs data will inevitably be affected during the process of sensor patch detection interference and position adjustment. In order to avoid missing important vital signs data in this process, the data prediction unit inputs historical vital signs data X t-γ:t , predict missing segment X t+1:t+Δ ,

[0097]

[0098] The loss function is:

[0099]

[0100] In addition, other vital sign data are introduced to predict the rationality of the signal through joint distribution constraints, and the formula for defining the physiological consistency loss is:

[0101]

[0102] Where f(·) is the physiological feature (such as heart rate variability) extracted from the predicted signal, and g(·) is the measured feature from other sensors. If the predicted signal conflicts with the measured parameters, a re-prediction is triggered.

[0103] The above processing is based on a multi-channel fault classification response algorithm to address various possible sensor patch faults. The vital sign data detected by these adjusted sensor patches is displayed on a monitoring monitor for doctors in the ward to observe, making it easier for medical staff to monitor the patient's physical condition.

[0104] In addition, these data are uploaded to the database. The multi-source data fusion unit unifies the time base of these vital sign data, eliminates the sampling rate and phase differences, and the interpolation formula (taking PPG as an example) is:

[0105]

[0106] In the formula is the basis function, and h is the interpolation step size.

[0107] Key features of the different data sets were then extracted. For heart rate variability (HRV), the standard deviation and root mean square (RMS) of the differences between adjacent RR intervals were calculated from the RR interval sequence of the ECG signal. The power ratio of low-frequency (0.04-0.15 Hz) to high-frequency (0.15-0.4 Hz) signals was calculated using FFT. For blood oxygen fluctuation (SpO2), the amplitude and low-frequency oscillation frequency were calculated from the SpO2 sequence of the PPG signal. Respiratory rate variability and the inspiration / expiration time ratio were calculated based on the respiratory signal period.

[0108] After feature vector construction and standardized fusion, the XGBoost classifier is used to construct the objective function:

[0109]

[0110] Where, is the risk probability, Ω(f i ) is the tree complexity regularization term.

[0111] Finally, the original vital signs data and risk levels of multiple patients are output on the vital signs monitoring platform, which is convenient for on-duty medical staff to monitor, reduce manpower requirements, and reduce risks: p i <0.3, medium risk: 0.3≤p i <0.7, high risk: p i ≥0.7.

[0112] Example 2

[0113] The difference from the above embodiment is that, as shown in the attached Figure 1 、 Figure 2 、 Figure 3 、 Figure 4 and Figure 5 As shown: the sensor stickers include electrocardiogram electrode sensor stickers, blood oxygen saturation optical sensor stickers, blood pressure impedance sensor stickers, body temperature sensor stickers and respiratory impedance sensor stickers, which respectively detect the above corresponding vital signs data.

[0114] The sensor patch includes a base plate 1, which has a plurality of base strips 9 integrally formed around the base plate 1. A deformation disk 4 is embedded inside the base plate 1, and deformation strips 3 are embedded in the base strips 9. Both the deformation disk 4 and the deformation strips 3 are made of electrically responsive materials.

[0115] The bottom wall of the deformation disk 4 is bonded with a corresponding type of vital sign sensor 5 , the bottom wall of the deformation disk 4 is also bonded with an infrared distance sensor 8 , and the top wall of the deformation disk 4 is bonded with a battery 7 and a magnetic positioning tag 6 .

[0116] The bottom walls of the base plate 1 and the base strip 9 are both covered with a bonding layer 2, which is made of biocompatible material. Several three-dimensional liquid diodes are opened in the base plate 1, the base strip 9 and the bonding layer 2, and the bonding layer 2 is used to be adhered to the patient's skin.

[0117] The specific implementation process is as follows: The sensor patch's base plate 1, base strip 9, and conforming layer 2 are all made of flexible, biocompatible materials (such as medical silicone or polyurethane) to ensure comfort and adaptability when applied to the patient's skin. The bottom surface of the conforming layer 2 is coated with hypoallergenic medical adhesive for stable adhesion to the skin surface. Several three-dimensional liquid diode structures are evenly distributed within the base plate 1, base strip 9, and conforming layer 2. Their microchannel design accelerates sweat drainage and prevents slippage caused by sweat accumulation.

[0118] The deformable disk 4 is made of an ion exchange polymer (such as Nafion) or an electroactive hydrogel, and its deformation can be adjusted by voltage. Together, the deformable disk 4 and the deformable strip 3 form a dynamic adjustment mechanism that compensates for contact deviation between the sensor patch and the skin. The magnetic positioning tag 6 is a miniature magnetic disk (made of neodymium iron boron). Its magnetic field characteristics are detected by the magnetic field sensing positioning sensor of the patient monitoring component and used to calibrate the patch's position.

[0119] When the offset positioning unit calculates the offset Δd>5mm based on the coordinates of the sensor patch and the coordinates of the standard anatomical model, the compensation control unit controls the deformation strip 3 in the base strip 9 in the corresponding direction of the sensor patch to be energized and produce a downward bending deformation. The bending degree and strength of the deformation strip 3 can be adjusted according to the magnitude of the voltage. Since the sensor patch is attached to the skin surface as a whole, in order to ensure firm adhesion, the bending of the base strip 9 can only slightly drive the movement of the sensor patch on the patient's skin surface, thereby compensating for the offset of the sensor patch.

[0120] When the infrared ranging sensor 8 detects that the distance between the center of the deformation disk 4 and the patient's skin is greater than 2 mm and abnormal vital sign data appears, the compensation control unit changes the concave and convex of the deformation disk 4 by controlling the voltage change, thereby moving the sensor for detecting vital sign data closer to the patient's skin to improve the accuracy of the data.

[0121] Design experiments to verify the function of the sensor sticker:

[0122] 1. Experimental subjects

[0123] Subjects: 5 healthy volunteers; Control equipment: Standard ECG monitor, oximeter, sphygmomanometer, thermometer, respiratory monitor; Environmental conditions: Temperature 25±2°C, humidity 50±10%; Test duration: 24 hours of continuous monitoring.

[0124] 2. Experimental process

[0125] (1) Initial calibration and baseline data collection: The sensor patch (ECG, SpO2, BP, Temp, RR) was attached to the subject's body surface according to the standard position of the anatomical model. The initial position offset (Δd < 1 mm) was confirmed using magnetic positioning and optical sensors. Baseline vital sign data were collected simultaneously from the traditional device and the sensor patch for 10 minutes.

[0126] (2) Position offset compensation test: Manually offset the sensor to a horizontal position of Δd = 6 mm, triggering the automatic compensation mechanism. Record the compensation completion time and the final offset. Repeat the test five times and take the average value.

[0127] (3) Vital sign data accuracy test: in two states, static and moving (simulating slight limb movement), the monitoring data (ECG, SpO2, BP, Temp, RR) of the traditional device and the sensor patch were recorded synchronously.

[0128] The data error rate was calculated (error rate = |sensor patch data - conventional device data| / conventional device data × 100%).

[0129] (4) Long-term wear stability test: Continuous monitoring for 24 hours, recording the position offset (Δd) of the sensor patch and vital sign data every hour to assess data continuity and stability.

[0130] (5) Fault simulation test, local signal interference: Block part of the sensor patch and observe the prediction data and cross-validation results of the timing prediction model. Poor contact: Artificially create a gap between the sensor patch and the skin (>2mm), record the system response time and the adjustment effect. Hardware damage: Disconnect the power supply of some sensor patches and verify whether the alarm mechanism is triggered.

[0131] 3. Experimental results:

[0132] The automatic compensation function of the sensor patch can compensate the 6mm offset to within 0.7mm within 3.3 seconds, meeting the preset threshold, as shown in Table 1.

[0133] Table 1. Position offset compensation performance test results

[0134]

[0135] The error rate in the static state was less than 1%, which met the clinical monitoring standard. The results are shown in Table 2.

[0136] Table 2. Comparison of error rates of vital signs data

[0137]

[0138] The maximum offset within 24 hours is 2.0 mm, and the data error rate does not exceed 2%, which meets the long-term monitoring requirements. The results are shown in Table 3.

[0139] Table 3. Long-term wearing stability test results (24 hours)

[0140]

[0141] After the sensors are pasted with various interferences or faults, the system can quickly and effectively handle local interference, poor contact and hardware failures to ensure monitoring continuity.

[0142] It should be noted that, in this document, relational terms such as first and second, etc., are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the terms "comprises," "comprising," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that includes a list of elements includes not only those elements but also other elements not explicitly listed, or elements inherent to such process, method, article, or apparatus.

[0143] The above is only an embodiment of the present invention. Common knowledge such as the known specific structures and characteristics in the scheme is not described in detail here. Ordinary technicians in the field are aware of all common technical knowledge in the technical field of the invention before the application date or priority date, can obtain all existing technologies in the field, and have the ability to apply conventional experimental means before that date. Ordinary technicians in the field can improve and implement this scheme in combination with their own abilities under the inspiration given by this application. Some typical known structures or known methods should not become obstacles for ordinary technicians in the field to implement this application. It should be pointed out that for those skilled in the art, without departing from the structure of the present invention, several variations and improvements can be made, which should also be regarded as the scope of protection of the present invention. These will not affect the effect of the implementation of the present invention and the practicality of the patent. The scope of protection required by this application shall be based on the content of its claims, and the specific implementation methods and other records in the specification can be used to interpret the content of the claims.

Claims

1. An intelligent monitoring device for anesthesia recovery period, characterized by: Including sensor stickers, repeaters, ward monitoring components, cloud storage platform and vital signs monitoring platform, The sensor patch is used to monitor vital signs of patients recovering from anesthesia, including electrocardiogram (ECG), blood oxygen saturation, blood pressure, body temperature, and respiration. The patch is magnetically positioned to mark the location on the patient's body surface. The patch detects the gap between the patch and the patient's skin using infrared ranging and changes the contact between the patch and the patient's skin using electro-deformation response. A repeater, used to relay the detection data signal and control signal of the sensor patch, and transmit the above detection data signal and control signal between the sensor patch and the ward monitoring component through wireless transmission; The ward monitoring component is used to display and upload the detected patient vital sign data to the cloud storage platform; it uses optical sensors to collect the patient's body surface contour image data, matches it to a preset anatomical model library, uses magnetic field induction positioning sensors to detect the position of the sensor patch, calculates the offset between the sensor patch and the body surface anatomical landmark, and controls the deformation compensation offset of the sensor patch according to a preset offset threshold; When a time series prediction model trained with complete vital sign time series data detects signal loss or noise interference, it uses historical data to predict missing segment data and cross-validates the rationality of the predicted data. Sensor patch failures are handled hierarchically based on a multi-channel fault classification response algorithm: a. When local signal interference occurs, the time series prediction model automatically predicts data and cross-validates; b. When the sensor patch has poor contact, control the deformation of the sensor patch to make it close or offset, and change its relative position to the patient's skin to optimize the monitoring site; c. If the hardware is damaged or there is no signal for 3 minutes, notify medical staff for maintenance; The cloud storage platform is used to store long-term patient vital sign data and transmit the data to the vital sign monitoring platform. A multi-source data fusion algorithm is used to extract heart rate variability, blood oxygen fluctuation, and respiratory rate variability characteristics based on the patient's vital sign data to generate a recovery period risk assessment to assist medical staff in decision-making. The vital signs monitoring platform is used to display the patient's vital signs data and curves, and to alert the medical staff on duty when the patient's vital signs data deviates from the normal value range.

2. The intelligent monitoring device for anesthesia recovery period according to claim 1, characterized in that: The sensor stickers include electrocardiogram electrode sensor stickers, blood oxygen saturation optical sensor stickers, blood pressure impedance sensor stickers, body temperature sensor stickers and respiratory impedance sensor stickers, which respectively detect the corresponding vital signs data mentioned above.

3. The intelligent monitoring device for anesthesia recovery period according to claim 2, characterized in that: The sensor patch comprises a base plate (1), a plurality of base strips (9) are fixedly connected to the base plate in a circumferential direction, a deformable plate (4) is fixedly connected inside the base plate (1), and a deformable strip (3) is fixedly connected inside each of the base strips (9), and both the deformable plate (4) and the deformable strip (3) are made of an electrically responsive material.

4. The intelligent monitoring device for anesthesia recovery period according to claim 3, characterized in that: The bottom wall of the deformation disk (4) is fixedly connected with a corresponding type of vital sign sensor (5), the bottom wall of the deformation disk (4) is also fixedly connected with an infrared distance sensor (8), and the top wall of the deformation disk (4) is fixedly connected with a battery (7) and a magnetic positioning tag (6).

5. The intelligent monitoring device for anesthesia recovery period according to claim 3, characterized in that: The bottom walls of the base plate (1) and the base strip (9) are both paved with a bonding layer (2), which is made of a biocompatible material. A plurality of three-dimensional liquid diodes are provided in the base plate (1), the base strip (9) and the bonding layer (2), and the bonding layer (2) is used for being adhered to the patient's skin.

6. The intelligent monitoring device for anesthesia recovery period according to claim 1, characterized in that: The ward monitoring component includes a monitoring display, an optical sensor, an offset positioning unit, a compensation control unit and a data prediction unit. A monitoring display for displaying real-time vital sign data transmitted by the repeater; Optical sensors, including several ToF ranging sensors and cameras, are used to detect images and distance data of the patient's body contours; An offset positioning unit is used to match the body surface contour image data of the patient detected by the optical sensor and the position of the sensor patch detected by the magnetic field induction positioning sensor, match the body surface anatomical landmarks on the preset anatomical model, and calculate the offset; A compensation control unit is used to control the deformation disk (4) on the sensor patch to deform when the offset reaches an offset threshold value to reduce the distance between the sensor patch and the patient's skin, and to control the deformation strip to curl and pull the sensor patch to change its relative position with the patient's skin; The data prediction unit is used to predict data missing segments caused by sensor movement or interference using a time series prediction model with a self-attention mechanism, and to cross-validate the rationality of the predicted data.

7. The intelligent monitoring device for anesthesia recovery period according to claim 6, characterized in that: The offset positioning unit uses the measured landmark points to match the anatomical model landmark points, converts the measured body surface contour image data coordinate system to the anatomical model coordinate system, and calculates the offset between the magnetic positioning position and the ideal position based on the converted measured body surface contour image data coordinate system.

8. The intelligent monitoring device for anesthesia recovery period according to claim 6, characterized in that: The cross-validation of the data prediction unit is to introduce other vital signs data, predict the rationality of the data through joint distribution constraints, and introduce a discriminator to evaluate the realism of the predicted data. When the predicted data conflicts with the measured data and the physiological joint distribution constraints at the same time, the dynamic threshold verification triggers re-prediction.

9. The intelligent monitoring device for anesthesia recovery period according to claim 1, characterized in that: The cloud storage platform includes a database and a multi-source data fusion unit. A database for continuously storing real-time monitored patient vital signs data; The multi-source data fusion unit is used to align and fuse vital sign data, extract heart rate variability, blood oxygen fluctuation and respiratory rate variability features, and generate a recovery period risk assessment.

10. The intelligent monitoring device for anesthesia recovery period according to claim 9, characterized in that: The multi-source data fusion unit unifies the time base of vital sign data by interpolating and aligning the timestamps between vital sign data and aligning periodic features. It uses the XGBoost classifier to evaluate the key feature vectors after standardized fusion and outputs the risk level.

Images