Breathing system air leakage detection method for mechanical ventilation

Abstract

The invention relates to the technical field of medical instruments, and discloses a respiratory system air leakage detection method for mechanical ventilation, which comprises the following steps: acquiring airway pressure, tidal volume and flow velocity waveforms in real time, performing multi-parameter fusion analysis, combining with an air leakage characteristic model constructed by machine learning to judge waveform abnormity, and dynamically adjusting a model judgment threshold according to the type of a patient; when the waveform is abnormal, three-stage progressive troubleshooting is started, the sealing performance of nine preset connecting points of a breathing machine pipeline is checked firstly, then the pressure of an air bag is monitored and adjusted to 30-50 cm H2O, dynamic minimum closing pressure monitoring is conducted, and finally the airway matching performance is evaluated; the method further comprises the steps of equipment fault verification and imaging examination prompt generation during suspected pneumothorax. The method improves the detection sensitivity and specificity, shortens the detection time, adapts to different patients, and reduces the risk of mechanical ventilation complications.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of medical devices, in particular to a method for detecting air leakage of a breathing system for mechanical ventilation. BACKGROUND

[0002] Invasive mechanical ventilation is a treatment method for delivering gas to the lungs of a patient through a tracheal tube, which is a commonly used life support means in intensive care units and plays an irreplaceable role in the treatment of critically ill patients. However, in clinical practice, air leakage problems of the respiratory system of patients receiving mechanical ventilation are common, which seriously threatens the safety of treatment and the prognosis of patients.

[0003] Air leakage of the respiratory system includes the following three cases:

[0004] Leakage of the breathing machine pipeline: such as the gap leakage of the breathing machine gas return connector, gas inlet connector, humidification bottle connection part, water collecting cup and other pipeline components.

[0005] Tracheal tube related air leakage: such as insufficient airbag pressure of the tracheal tube, airbag damage, or airway sealing failure caused by airbag mismatch with the patient's airway.

[0006] Pneumothorax related air leakage: pneumothorax caused by barotrauma, etc., gas leakage from the lung damage to the pleural cavity, which is a life-threatening serious air leakage type.

[0007] In clinical practice, nursing staff often face the difficulty of quickly identifying the air leakage position. After air leakage occurs, on the one hand, it will lead to insufficient ventilation of the patient, and the alveoli cannot be fully inflated, thereby affecting oxygen exchange and carbon dioxide emission, and causing respiratory dysfunction such as hypoxemia and hypercapnia; on the other hand, airway leakage will destroy the airway sealing environment, making it easy for oral bacteria to invade the airway and lungs, significantly increasing the risk of ventilator-associated pneumonia (VAP), and aggravating the patient's condition and prolonging the treatment period. The existing clinical identification and positioning means for air leakage of the respiratory system lack precision and efficiency, and cannot meet the needs of critically ill patients for treatment safety and timeliness. SUMMARY

[0008] Therefore, the present application provides a method for detecting air leakage of a breathing system for mechanical ventilation to solve the problem of how to quickly and accurately detect air leakage of the respiratory system.

[0009] The present application provides a method for detecting air leakage of a breathing system for mechanical ventilation, comprising the following steps:

[0010] In response to monitoring the abnormality of the breathing machine waveform, a three-level progressive air leakage investigation path is performed;

[0011] The three-level progressive air leakage investigation path comprises:

[0012] First-stage tubing sealing check: Check the sealing of the nine preset connection points in the ventilator tubing system;

[0013] Second-stage cuff dynamic pressure monitoring: If no leaks are found in the first-stage inspection and the waveform does not return to normal, monitor and adjust the endotracheal intubation cuff pressure to the range of 30-50 cmH2O, and observe whether the waveform returns to normal.

[0014] Level 3 airway compatibility assessment: If the waveform is still abnormal after Level 2 adjustment, an airway compatibility assessment is performed, including using a stethoscope to determine if there is a whistling sound in the larynx.

[0015] Among them, the judgment of waveform anomalies is based on the multi-parameter fusion analysis of airway pressure, tidal volume and flow velocity waveforms, and is realized by a leakage feature model constructed through machine learning algorithm. The leakage feature model is identified based at least on waveform distortion rate and pressure attenuation slope.

[0016] This method for detecting air leaks in mechanically ventilated respiratory systems involves a three-stage progressive leak detection process. When the ventilator monitoring system detects an abnormal ventilator waveform by performing multi-parameter fusion analysis of airway pressure, tidal volume, and flow rate waveforms, and utilizing a leak characteristic model built using machine learning algorithms (based on waveform distortion rate and pressure attenuation slope), a three-stage progressive leak detection path is initiated to precisely locate the leak. First, a first-stage tubing seal check is performed, meticulously verifying the seals of each of the nine pre-set connection points in the ventilator tubing system. If no leak is found during this check and the ventilator waveform still does not return to normal, the second-stage dynamic cuff pressure monitoring stage is initiated. This stage monitors the endotracheal intubation cuff pressure in real time and adjusts it to a range of 30-50 cmH2O, while continuously observing whether the waveform returns to normal with pressure adjustment. If the waveform remains abnormal after the second-stage cuff pressure monitoring and adjustment, a third-stage airway compatibility assessment is conducted, using a stethoscope to determine if there is a leaking sound in the patient's throat, further pinpointing the source of the leak. This method effectively improves the accuracy of ventilator waveform anomaly identification by leveraging multi-parameter fusion analysis and precise interpretation by machine learning models, avoiding the missed detection problems caused by single-parameter judgment in traditional screening. The three-level progressive screening path significantly shortens the leak detection time by focusing on each level and gradually narrowing the screening scope, while improving the sensitivity and specificity of leak detection. It can quickly locate the leak point to reduce the duration of ineffective ventilation and reduce the risk of complications caused by respiratory system leaks. In addition, the clear screening logic simplifies the clinical operation process and improves the efficiency and accuracy of medical staff in screening for leaks.

[0017] In one alternative implementation, between the first-stage tubing sealing check and the second-stage cuff dynamic pressure monitoring, a device fault verification step is also included: disconnecting the ventilator tubing from the patient's artificial airway and connecting it to a mannequin or simple respirator; if the waveform returns to normal at this time, the ventilator itself is determined to be fault-free.

[0018] This process, by adding a device fault verification step between the first and second levels of investigation, can eliminate interference from ventilator malfunctions in leak detection in advance, avoiding detours in subsequent investigations. Combined with multi-parameter fusion analysis and precise interpretation by machine learning models, it not only improves the accuracy of waveform anomaly identification, but also shortens leak detection time through a progressive investigation logic, effectively improving the sensitivity and specificity of leak detection, reducing the duration of ineffective ventilation due to leaks, lowering the risk of related complications, and optimizing clinical operation procedures to improve the efficiency of medical staff in investigations.

[0019] In one optional implementation, the second-stage airbag dynamic pressure monitoring includes a dynamic minimum closure pressure monitoring step:

[0020] Gradually reduce the airbag pressure while simultaneously observing waveform changes until a leaking sound or abnormal waveform is detected, in order to determine the critical pressure value at which the airway closes.

[0021] This process, by adding a dynamic minimum closure pressure monitoring step to the second-stage cuff dynamic pressure monitoring, can more accurately match the individual airway closure needs of patients, avoiding misjudgments of leaks caused by cuff pressure settings that are too high or too low. Combined with multi-parameter fusion analysis and preliminary interpretation by machine learning models, it not only improves the accuracy of waveform anomaly identification and leak location, but also further optimizes the rationality of cuff pressure adjustment, effectively shortening the time for leak detection and problem resolution, reducing the duration of ineffective ventilation, and lowering the risk of complications caused by improper cuff pressure or leaks. At the same time, it provides a more accurate pressure reference for subsequent airway compatibility assessment, improving overall screening efficiency.

[0022] In one optional implementation, the preset connection points are respectively the air delivery interface between the ventilator and the tubing, the air return interface between the ventilator and the tubing, the airway humidification fluid inlet to the humidification bottle interface, the air delivery interface between the ventilator tubing and the humidification bottle, the air outlet interface between the ventilator tubing and the humidification bottle, the water collection cup of the air return section of the ventilator tubing, the water collection cup of the air delivery section of the ventilator tubing, the Y-shaped connector of the ventilator tubing, and the interface between the Y-shaped connector and the patient's endotracheal tube.

[0023] In one optional implementation, the third-level airway compatibility assessment includes:

[0024] If the larynx leaking sound disappears after increasing the cuff pressure to more than 30 cmH2O, it indicates a mismatch between the endotracheal intubation cuff and the patient's airway, and the endotracheal tube should be replaced.

[0025] This process clarifies the correlation between "cuff pressure > 30 cmH2O and disappearance of laryngeal air leakage sound" in the third-level airway compatibility assessment. This allows for precise identification of air leakage caused by incompatibility between the endotracheal intubation cuff and the airway, avoiding missed or misdiagnosed cases in traditional screening methods. Furthermore, by combining precise waveform anomaly identification from prior multi-parameter fusion analysis, comprehensive screening of the first-level tubing seal, and dynamic adjustment of the second-level cuff pressure, a progressive and targeted air leakage screening system is formed. This effectively shortens the time required to resolve air leakage problems, reduces continuous ineffective ventilation due to airway mismatch, lowers the risk of related complications, and provides clear operational guidelines and decision-making basis for medical staff, improving the efficiency and accuracy of clinical air leakage screening.

[0026] In one alternative implementation, during the third-level assessment, if the airbag pressure decreases within 1 minute after inflation, the airbag is considered ruptured, and the endotracheal tube is replaced.

[0027] This process, by simultaneously covering the judgment logic of two leakage causes, "airway mismatch" and "balloon rupture," in the third-level airway compatibility assessment, combined with the accurate waveform anomaly identification from the previous multi-parameter fusion analysis, the comprehensive investigation of the first-level tubing seal, and the dynamic control of the second-level balloon pressure, forms a more complete targeted investigation system. This not only avoids the omission of problems such as balloon rupture, but also further clarifies the treatment direction in different leakage scenarios, effectively shortens the time for solving leakage problems, reduces the duration of ineffective ventilation caused by continuous leakage, and lowers the risk of related complications. At the same time, it provides medical staff with clearer operational and decision-making basis, improving the efficiency and accuracy of clinical leakage investigation.

[0028] In one alternative implementation, if the air leak and suspected pneumothorax are still not resolved after completing the three-level progressive path, then imaging examinations are performed.

[0029] After completing the entire process of the above three-level progressive pathway investigation, if the air leakage problem in the respiratory system is still not resolved, and the previous investigation has ruled out related issues such as tubing sealing, cuff pressure and integrity, and airway compatibility, and pneumothorax is suspected, imaging examinations (such as chest X-rays) will be performed so that medical staff can further confirm the pneumothorax condition and take targeted treatment measures based on the examination results to effectively resolve the air leakage problem and ensure the effectiveness of mechanical ventilation treatment.

[0030] In one optional implementation, the multi-parameter fusion analysis includes:

[0031] The waveform was analyzed using a leakage feature model established using machine learning algorithms;

[0032] The leakage characteristic model identifies leakage based at least on the waveform distortion rate and pressure attenuation slope.

[0033] In an optional implementation, the method further includes:

[0034] The judgment threshold used for the air leakage feature model is dynamically adjusted according to the patient type. Attached Figure Description

[0035] To more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the specific embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0036] Figure 1 This is a structural diagram of nine potential leak points in the breathing tubing;

[0037] Figure 2 This is a schematic diagram of a normal waveform;

[0038] Figure 3 This is a schematic diagram of an air leak waveform;

[0039] Figure 4 This is a comparison chart of the air leak detection speeds of the experimental group and the control group.

[0040] Figure 5 This is a flowchart illustrating a method for detecting leaks in a respiratory system used in mechanical ventilation.

[0041] Explanation of reference numerals in the attached figures:

[0042] 1. The air delivery interface between the ventilator and the tubing;

[0043] 2. The return air interface between the ventilator and the tubing;

[0044] 3. The airway humidification fluid enters the humidification bottle inlet;

[0045] 4. Ventilator tubing and humidifier bottle air delivery interface;

[0046] 5. Ventilator tubing and humidifier bottle outlet;

[0047] 6. Water collection cup for the return section of the ventilator tubing;

[0048] 7. Water collection cup for the air delivery section of the ventilator tubing;

[0049] 8. Y-shaped connector for ventilator tubing;

[0050] 9. Y-shaped connector and patient endotracheal tube interface. Detailed Implementation

[0051] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, 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, 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.

[0052] Ventilators are essential medical devices for rescuing patients with respiratory failure and treating respiratory diseases. They provide invasive ventilation via endotracheal intubation or non-invasive ventilation via face masks, nasal masks, or other ventilation interfaces. The main ventilation modes include volume-controlled ventilation and pressure-controlled ventilation. Although non-invasive ventilation is increasingly used in home and clinical settings due to its advantages such as not requiring endotracheal intubation, good patient tolerance, and lower risk of secondary lung injury, airway leakage is an unavoidable problem in non-invasive ventilation. Leakage can lead to insufficient ventilation, reduced ventilation tolerance, severely impacting ventilation effectiveness, and even threatening life. Therefore, there is an urgent need in this field for a method that can rapidly and accurately detect respiratory system leaks.

[0053] The following is combined with Figures 1 to 5 The following describes embodiments of the present invention.

[0054] According to an embodiment of the present invention, a method for detecting air leaks in a mechanically ventilated respiratory system is provided, comprising the following steps:

[0055] In response to the detection of abnormal ventilator waveforms, a three-level progressive leak investigation path is executed;

[0056] The three-tiered, progressive leak detection process includes:

[0057] First-stage tubing sealing check: Check the sealing of the nine preset connection points in the ventilator tubing system;

[0058] Second-stage cuff dynamic pressure monitoring: If no leaks are found in the first-stage inspection and the waveform does not return to normal, monitor and adjust the endotracheal intubation cuff pressure to the range of 30-50 cmH2O, and observe whether the waveform returns to normal.

[0059] Level 3 airway compatibility assessment: If the waveform is still abnormal after Level 2 adjustment, an airway compatibility assessment is performed, including using a stethoscope to determine if there is a whistling sound in the larynx.

[0060] Among them, the judgment of waveform anomalies is based on the multi-parameter fusion analysis of airway pressure, tidal volume and flow velocity waveforms, and is realized by a leakage feature model constructed through machine learning algorithms. The leakage feature model is based on at least waveform distortion rate and pressure attenuation slope for identification.

[0061] This embodiment provides a method for detecting air leaks in a mechanically ventilated respiratory system. When the ventilator monitoring system determines that the ventilator waveform is abnormal by performing multi-parameter fusion analysis of airway pressure, tidal volume, and flow rate waveforms, and by relying on a leak characteristic model constructed based on machine learning algorithms, which is at least based on waveform distortion rate and pressure attenuation slope identification, a three-level progressive leak investigation path is immediately initiated to accurately locate the leak problem: First, a first-level tubing sealing check is performed, meticulously verifying the sealing of each of the nine preset connection points in the ventilator tubing system. If no leak is found in this check and the ventilator waveform still does not return to normal, the second-level cuff dynamic pressure monitoring stage is entered, which monitors the endotracheal intubation cuff pressure in real time and adjusts it to the range of 30-50 cmH2O, while continuously observing whether the waveform returns to normal with pressure adjustment; if the waveform is still abnormal after the second-level cuff pressure monitoring and adjustment, the third-level airway compatibility assessment is carried out, using a stethoscope to determine whether there is a leaking sound in the patient's throat, further pinpointing the source of the leak.

[0062] In terms of effectiveness, this method, through multi-parameter fusion analysis and precise interpretation by machine learning models, effectively improves the accuracy of ventilator waveform anomaly identification, avoiding the missed detection problems caused by single-parameter judgment in traditional screening. The three-level progressive screening path significantly shortens the leak detection time by focusing on each level and gradually narrowing the screening scope, while improving the sensitivity and specificity of leak detection. It can quickly locate the leak point to reduce the duration of ineffective ventilation and reduce the risk of complications caused by respiratory system leaks. In addition, the clear screening logic simplifies the clinical operation process and improves the efficiency and accuracy of medical staff in screening for leaks.

[0063] In one embodiment, between the first-stage tubing sealing check and the second-stage cuff dynamic pressure monitoring, a device fault verification step is also included: disconnecting the ventilator tubing from the patient's artificial airway and connecting it to a mannequin or simple respirator; if the waveform returns to normal at this time, it is determined that the ventilator itself is not faulty.

[0064] In the air leakage detection method for a respiratory system used in mechanical ventilation in this embodiment, firstly, by multi-parameter fusion analysis of airway pressure, tidal volume and flow rate waveforms, combined with an air leakage feature model constructed by machine learning algorithm, which identifies at least based on waveform distortion rate and pressure attenuation slope, the system monitors and judges in real time whether there is an abnormality in the ventilator waveform; when an abnormality in the ventilator waveform is detected, a three-level progressive air leakage investigation path is immediately initiated. First, the first-level tubing sealing check is performed, and the sealing of the nine preset connection points in the ventilator tubing system is carefully checked one by one.

[0065] If the first-stage tubing seal check does not find any leaks and the ventilator waveform still does not return to normal, proceed to the equipment fault verification step: disconnect the ventilator tubing from the patient's artificial airway, and then connect the disconnected ventilator tubing to the mannequin or simple respirator; if the ventilator waveform returns to normal after connection, it is determined that the ventilator itself is not faulty, and then proceed to the second-stage cuff dynamic pressure monitoring stage; if the waveform still does not return to normal after connection, it indicates that the ventilator itself is faulty, and the equipment should be repaired first, and subsequent leak investigation should continue after the equipment fault is eliminated.

[0066] After confirming that the ventilator is not faulty, the second stage of dynamic cuff pressure monitoring is carried out to monitor the pressure of the endotracheal intubation cuff in real time and adjust it precisely to the range of 30-50 cmH2O. At the same time, the ventilator waveform is continuously observed to see if it returns to normal with the pressure adjustment. If the waveform is still abnormal after the second stage of cuff pressure monitoring and adjustment, the third stage of airway compatibility assessment is initiated. A stethoscope is used to determine whether there is a leak sound in the patient's throat, so as to gradually narrow down the scope of the leak investigation and accurately locate the source of the leak.

[0067] This process, by adding a device fault verification step between the first and second levels of investigation, can eliminate interference from ventilator malfunctions in leak detection in advance, avoiding detours in subsequent investigations. Combined with multi-parameter fusion analysis and precise interpretation by machine learning models, it not only improves the accuracy of waveform anomaly identification, but also shortens leak detection time through a progressive investigation logic, effectively improving the sensitivity and specificity of leak detection, reducing the duration of ineffective ventilation due to leaks, lowering the risk of related complications, and optimizing clinical operation procedures to improve the efficiency of medical staff in investigations.

[0068] In one embodiment, the second-stage airbag dynamic pressure monitoring includes a dynamic minimum closure pressure monitoring step:

[0069] Gradually reduce the airbag pressure while simultaneously observing waveform changes until a leaking sound or abnormal waveform is detected, in order to determine the critical pressure value at which the airway closes.

[0070] In the air leak detection method for mechanical ventilation in this embodiment, firstly, multi-parameter fusion analysis of airway pressure, tidal volume, and flow rate waveforms is performed, combined with an air leak feature model constructed by machine learning algorithms, which identifies leaks based at least on waveform distortion rate and pressure decay slope, to monitor and determine in real time whether there are abnormalities in the ventilator waveform. When an abnormality in the ventilator waveform is detected, a three-level progressive leak investigation path is immediately initiated. The first level is a tubing sealing check, which checks the sealing of nine preset connection points in the ventilator tubing system one by one. If no leak is found in the first level check and the waveform has not returned to normal, the second level, dynamic pressure monitoring of the cuff, is initiated.

[0071] The second-stage dynamic cuff pressure monitoring, in addition to real-time monitoring and adjusting the endotracheal intubation cuff pressure to the range of 30-50 cmH2O and observing whether the waveform returns to normal, also includes a dynamic minimum closure pressure monitoring step: Based on the cuff pressure being 30-50 cmH2O, the cuff pressure is gradually reduced while continuously and synchronously observing changes in the ventilator waveform. During this process, close attention is paid to any leak-related signals until a leaking sound is detected in the patient's throat or the ventilator waveform becomes abnormal again. The pressure value recorded at this point is the critical pressure value for airway closure. If the critical pressure value is determined to be reasonable through dynamic minimum closure pressure monitoring, and the waveform returns to normal after adjusting the cuff pressure to near this critical pressure, then the leak problem is determined to be due to improper cuff pressure setting. If the waveform remains abnormal after the above dynamic minimum closure pressure monitoring and pressure adjustment, the third-stage airway compatibility assessment is initiated. A stethoscope is used to determine whether there is a persistent leaking sound in the patient's throat, further pinpointing the source of the leak.

[0072] This process, by adding a dynamic minimum closure pressure monitoring step to the second-stage cuff dynamic pressure monitoring, can more accurately match the individual airway closure needs of patients, avoiding misjudgments of leaks caused by cuff pressure settings that are too high or too low. Combined with multi-parameter fusion analysis and preliminary interpretation by machine learning models, it not only improves the accuracy of waveform anomaly identification and leak location, but also further optimizes the rationality of cuff pressure adjustment, effectively shortening the time for leak detection and problem resolution, reducing the duration of ineffective ventilation, and lowering the risk of complications caused by improper cuff pressure or leaks. At the same time, it provides a more accurate pressure reference for subsequent airway compatibility assessment, improving overall screening efficiency.

[0073] In one embodiment, the preset connection points are as follows: 1. Air delivery interface between the ventilator and the tubing; 2. Air return interface between the ventilator and the tubing; 3. Airway humidification fluid inlet to humidification bottle interface; 4. Air delivery interface between the ventilator tubing and the humidification bottle; 5. Air outlet interface between the ventilator tubing and the humidification bottle; 6. Water collection cup of the air return section of the ventilator tubing; 7. Water collection cup of the air delivery section of the ventilator tubing; 8. Y-shaped connector of the ventilator tubing; and 9. Interface between the Y-shaped connector and the patient's endotracheal tube.

[0074] In one embodiment, the third-level airway compatibility assessment includes:

[0075] If the larynx leaking sound disappears after increasing the cuff pressure to more than 30 cmH2O, it indicates a mismatch between the endotracheal intubation cuff and the patient's airway, and the endotracheal tube should be replaced.

[0076] In the respiratory system leak detection method for mechanical ventilation in this embodiment, firstly, multi-parameter fusion analysis of airway pressure, tidal volume, and flow rate waveforms is performed. Combined with a leak characteristic model constructed by machine learning algorithms, which identifies leaks based at least on waveform distortion rate and pressure attenuation slope, the system monitors and judges in real time whether there are abnormalities in the ventilator waveform. When an abnormality in the ventilator waveform is detected, a three-level progressive leak investigation path is immediately initiated. The first level is a tubing sealing check, which checks the sealing of nine preset connection points in the ventilator tubing system one by one. If no leak is found in the first level check and the waveform still does not return to normal, the second level is the cuff dynamic pressure monitoring stage. In addition to real-time monitoring and adjusting the endotracheal intubation cuff pressure to the range of 30-50 cmH2O and observing whether the waveform returns to normal, this stage also includes a dynamic minimum closure pressure monitoring step (gradually reducing the cuff pressure and simultaneously observing waveform changes until a leak sound or waveform abnormality is detected to determine the critical pressure value for airway closure). If the waveform is still abnormal after the second level of monitoring and adjustment, the third level is the airway compatibility assessment stage.

[0077] The third-level airway compatibility assessment uses stethoscope judgment as the core operation: First, place the stethoscope close to the patient's larynx to listen for any air leakage sounds; if air leakage sounds are detected in the larynx, gradually increase the pressure from the original cuff pressure. When the cuff pressure increases to more than 30 cmH2O, if the air leakage sounds in the larynx disappear, then there is a mismatch between the endotracheal tube cuff and the patient's airway. In this case, replace the endotracheal tube with one that matches the patient's airway; if the air leakage sounds in the larynx do not disappear after increasing the cuff pressure to more than 30 cmH2O, further investigation is needed by combining previous waveform monitoring data and dynamic minimum closure pressure monitoring results to reconfirm the direction of air leakage investigation.

[0078] This process clarifies the correlation between "cuff pressure > 30 cmH2O and disappearance of laryngeal air leakage sound" in the third-level airway compatibility assessment. This allows for precise identification of air leakage caused by incompatibility between the endotracheal intubation cuff and the airway, avoiding missed or misdiagnosed cases in traditional screening methods. Furthermore, by combining precise waveform anomaly identification from prior multi-parameter fusion analysis, comprehensive screening of the first-level tubing seal, and dynamic adjustment of the second-level cuff pressure, a progressive and targeted air leakage screening system is formed. This effectively shortens the time required to resolve air leakage problems, reduces continuous ineffective ventilation due to airway mismatch, lowers the risk of related complications, and provides clear operational guidelines and decision-making basis for medical staff, improving the efficiency and accuracy of clinical air leakage screening.

[0079] Furthermore, in the third-level assessment, if the airbag pressure shows a downward trend within 1 minute after inflation, the airbag is considered ruptured, and the endotracheal tube should be replaced.

[0080] During the assessment process, the airbag pressure changes are continuously monitored. If the airbag pressure shows a downward trend within 1 minute after inflation, the airbag is likely ruptured, and the endotracheal tube should be replaced.

[0081] This process, by simultaneously covering the judgment logic of two leakage causes, "airway mismatch" and "balloon rupture," in the third-level airway compatibility assessment, combined with the accurate waveform anomaly identification from the previous multi-parameter fusion analysis, the comprehensive investigation of the first-level tubing seal, and the dynamic control of the second-level balloon pressure, forms a more complete targeted investigation system. This not only avoids the omission of problems such as balloon rupture, but also further clarifies the treatment direction in different leakage scenarios, effectively shortens the time for solving leakage problems, reduces the duration of ineffective ventilation caused by continuous leakage, and lowers the risk of related complications. At the same time, it provides medical staff with clearer operational and decision-making basis, improving the efficiency and accuracy of clinical leakage investigation.

[0082] In one embodiment, if the air leak and suspected pneumothorax are still not resolved after completing the three-level progressive approach, imaging examinations are performed.

[0083] In the air leak detection method for mechanical ventilation in this embodiment, firstly, multi-parameter fusion analysis of airway pressure, tidal volume, and flow rate waveforms is performed, combined with an air leak characteristic model constructed by a machine learning algorithm that identifies leaks based at least on waveform distortion rate and pressure decay slope, to monitor and determine in real time whether there are abnormalities in the ventilator waveform; when an abnormality in the ventilator waveform is detected, a three-level progressive leak investigation path is immediately initiated:

[0084] The first step is to perform a first-stage tubing leak check, verifying the seal of each of the nine pre-set connection points in the ventilator tubing system. If no leaks are found and the waveform still does not return to normal, the second stage, dynamic cuff pressure monitoring, is initiated. During the second stage, the endotracheal intubation cuff pressure is monitored in real time and adjusted to the range of 30-50 cmH2O. Simultaneously, dynamic minimum closure pressure monitoring is performed—gradually reducing the cuff pressure while observing waveform changes until a leak is detected or an abnormal waveform is observed. This determines the critical pressure value for airway closure. If the waveform remains abnormal after pressure adjustment and monitoring, the third stage, airway compatibility assessment, is then initiated.

[0085] The third-level assessment is conducted using a dual-judgment logic: on the one hand, a stethoscope is placed close to the patient's throat to listen for air leakage sounds. If the air leakage sounds disappear after the cuff pressure is increased to more than 30 cmH2O, it indicates that the endotracheal intubation cuff is not compatible with the patient's airway, and a recommendation to replace the endotracheal tube is generated. On the other hand, the cuff pressure changes are tracked simultaneously. If the pressure decreases within 1 minute after the cuff is inflated, it indicates that the cuff is damaged, and a recommendation to replace the endotracheal tube is also generated.

[0086] After completing the entire process of the above three-level progressive pathway investigation, if the air leakage problem in the respiratory system is still not resolved, and the previous investigation has ruled out related issues such as tubing sealing, cuff pressure and integrity, and airway compatibility, and pneumothorax is suspected, imaging examinations (such as chest X-rays) will be performed so that medical staff can further confirm the pneumothorax condition and take targeted treatment measures based on the examination results to effectively resolve the air leakage problem and ensure the effectiveness of mechanical ventilation treatment.

[0087] In one embodiment, multi-parameter fusion analysis includes:

[0088] The waveform was analyzed using a leakage characteristic model built with machine learning algorithms;

[0089] Leakage feature models identify leaks based at least on waveform distortion rate and pressure decay slope.

[0090] In the respiratory system leak detection method for mechanical ventilation in this embodiment, multi-parameter fusion analysis is the core step in judging abnormal ventilator waveforms. The specific operation process is as follows: First, the ventilator monitoring system collects three key parameters—airway pressure, tidal volume, and flow rate waveforms—in real time and synchronously during the patient's mechanical ventilation process to ensure the continuity and timeliness of data collection and provide a basis for subsequent analysis. Then, the collected three types of parameter data are input into a leak feature model pre-constructed by a machine learning algorithm, and the model performs a comprehensive analysis on the parameter data and corresponding waveforms. The core identification logic of this air leakage feature model revolves around at least two key indicators: waveform distortion rate and pressure decay slope. The waveform distortion rate is determined by calculating the deviation between the actual monitored waveform (such as tidal volume waveform and airway pressure waveform) and the preset normal waveform. For example, when the tidal volume waveform cannot recover to zero or the airway pressure waveform shows unexpected fluctuations, the distortion rate will increase accordingly. The pressure decay slope is used to quantify the rate at which the monitored pressure (such as endotracheal intubation cuff pressure and airway pressure) decreases over time. For example, when the pressure drops rapidly in a short period of time after the cuff is inflated, the pressure decay slope will show an abnormally increasing trend. By synergistically analyzing these two indicators, the leak feature model can accurately distinguish between normal ventilation waveforms and abnormal waveforms caused by leaks, effectively avoiding the misjudgment and missed judgment problems that are prone to occur in traditional single-parameter judgments. This provides a reliable basis for whether to initiate a three-level progressive leak investigation path. At the same time, relying on the adaptability of machine learning algorithms, the model can also dynamically optimize the analysis logic according to the ventilation parameter characteristics of different patients (such as children and COPD patients), further improving the adaptability and accuracy of waveform abnormality identification, and laying the foundation for the efficient implementation of the overall leak detection process.

[0091] In one embodiment, the method further includes:

[0092] The judgment threshold used for the air leakage feature model is dynamically adjusted according to the patient type.

[0093] By dynamically adjusting the judgment threshold of the leak feature model according to patient type, the problem of poor adaptability of fixed thresholds to different patient groups is effectively solved. This significantly improves the accuracy of leak identification for special groups such as children and COPD patients, reduces false positives and false negatives caused by threshold mismatch, and further optimizes the accuracy of multi-parameter fusion analysis. This provides a more patient-specific judgment basis for the efficient implementation of the subsequent three-level progressive screening path, ultimately reducing ineffective ventilation time and the risk of complications, and meeting the clinical needs for leak detection in different types of mechanically ventilated patients.

[0094] Table 1. Baseline characteristics of mechanically ventilated patients in the control and experimental groups

[0095] Parameters Control group Experimental group x 2 / t]]> P Gender (male / female) 88 / 39 115 / 60 0.427 0.513 Age 60.69±12.62 62.00±12.73 0.884 0.377 Weight (Kg) 64.74±10.09 65.29±11.66 0.428 0.669 Height (cm) 165.2±7.16 165.8±8.26 0.623 0.534 BMI 23.68±3.14 23.69±3.48 0.022 0.983 Peak inspiratory pressure (cmH20) 19.90±3.78 20.31±3.87 0.932 0.352 Tidal volume volume (ml) 586.9±79.22 588.1±98.9 0.114 0.910

[0096] Here, P is the P-value (probability value), ranging from 0 to 1. In the table, the P-values ​​for all parameters (such as gender, age, BMI, etc.) are greater than 0.05 (common significance level), indicating that there are no statistically significant differences between the control group and the experimental group in these baseline characteristics, and the two groups are comparable. Chi-square (χ²) represents the chi-square test, a statistical method used to analyze categorical variables (such as gender, surgical method, satisfaction level, etc.). It is used to determine whether there are statistically significant differences in the distribution of categorical variables between two or more samples. For example, comparing whether there is a significant difference in the gender ratio (male / female) between the control group and the experimental group, and analyzing whether two categorical variables are independent (such as whether "treatment method" is related to "efficacy"). The t-test represents a statistical method used to analyze continuous numerical variables (such as age, height, weight, etc.), used to determine whether there is a statistically significant difference in the population means of two independent samples. For example, comparing whether there is a significant difference in the mean age between the control group and the experimental group.

[0097] Appendix Figure 2 This is a schematic diagram of a normal waveform. Figure 3 This is a schematic diagram of an air leak waveform. Figure 4 This is a comparison chart of the air leak detection speeds of the experimental group and the control group. From... Figure 4 It can be seen that the air leak detection speed of the experimental group is much faster than that of the control group.

[0098] Figure 5In a specific implementation of a method for detecting air leaks in a mechanically ventilated respiratory system, when a leak waveform appears, first check points 1 to 9 of the ventilator tubing. If the leak can be resolved and the ventilator waveform returns to normal, the cause is likely a leak at points 1 to 9 of the ventilator tubing. If the ventilator waveform does not return to normal, pressurize the endotracheal tube cuff to 30 cmH2O. If the leak is resolved and the ventilator waveform returns to normal, the cause is insufficient cuff pressure. If the ventilator waveform still does not return to normal, disconnect the ventilator tubing at point 9 from the endotracheal tube, and disconnect the tubing from the mannequin lung. Connect the simple respirator to the endotracheal tube to maintain the patient's ventilation. If the ventilator waveform fails to return to normal, the ventilator is faulty, or the leak between 1 and 9 o'clock cannot be found. If the ventilator waveform returns to normal, listen to the throat with a stethoscope. If there is no leak sound in the throat, the patient may have pneumothorax, and a chest X-ray is needed for diagnosis. If there is a leak sound in the throat, and the pressure drops quickly after the cuff is inflated, the cuff may be leaking, and it is recommended to replace the endotracheal tube. If the leak sound in the throat decreases or disappears after the cuff is inflated, the cuff of the endotracheal tube may not be compatible with the patient's airway, and it is recommended to replace it with a compatible endotracheal tube.

[0099] Although embodiments of the invention have been described in conjunction with the accompanying drawings, those skilled in the art can make various modifications and variations without departing from the spirit and scope of the invention, and such modifications and variations all fall within the scope defined by the appended claims.

Claims

1. A method for detecting air leaks in a mechanically ventilated respiratory system, characterized in that, Includes the following steps: In response to the detection of abnormal ventilator waveforms, a three-level progressive leak investigation path is executed; The three-level progressive leak detection path includes: First-stage tubing sealing check: Check the sealing of the nine preset connection points in the ventilator tubing system; Second-stage cuff dynamic pressure monitoring: If no leaks are found in the first-stage inspection and the waveform does not return to normal, monitor and adjust the endotracheal intubation cuff pressure to the range of 30-50 cmH2O, and observe whether the waveform returns to normal. Level 3 airway compatibility assessment: If the waveform is still abnormal after Level 2 adjustment, an airway compatibility assessment is performed, including using a stethoscope to determine if there is a whistling sound in the larynx. Among them, the judgment of waveform anomalies is based on the multi-parameter fusion analysis of airway pressure, tidal volume and flow velocity waveforms, and is realized by a leakage feature model constructed through machine learning algorithm. The leakage feature model is identified based at least on waveform distortion rate and pressure attenuation slope.

2. The method for detecting air leakage in a mechanically ventilated respiratory system according to claim 1, characterized in that, Between the first-stage tubing sealing check and the second-stage cuff dynamic pressure monitoring, there is also a device failure verification step: disconnect the ventilator tubing from the patient's artificial airway and connect it to a mannequin lung or a simple respirator. If the waveform returns to normal at this time, it is determined that the ventilator itself is not faulty.

3. The method for detecting air leakage in a mechanically ventilated respiratory system according to claim 1 or 2, characterized in that, The second-stage airbag dynamic pressure monitoring includes the dynamic minimum closure pressure monitoring step: Gradually reduce the airbag pressure while simultaneously observing waveform changes until a leaking sound or abnormal waveform is detected, in order to determine the critical pressure value at which the airway closes.

4. The method for detecting air leakage in a mechanically ventilated respiratory system according to claim 1, characterized in that, The preset connection points are respectively the air delivery interface (1) of the ventilator and the tubing, the air return interface (2) of the ventilator and the tubing, the airway humidification fluid entering the humidification bottle interface (3), the air delivery interface of the ventilator tubing and the humidification bottle (4), the air outlet interface of the ventilator tubing and the humidification bottle (5), the water collection cup of the air return section of the ventilator tubing (6), the water collection cup of the air delivery section of the ventilator tubing (7), the Y-shaped connector of the ventilator tubing (8), and the interface between the Y-shaped connector and the patient's endotracheal tube (9).

5. The method for detecting air leakage in a mechanically ventilated respiratory system according to claim 1, characterized in that, The third-level airway compatibility assessment includes: If the larynx leaking sound disappears after increasing the cuff pressure to more than 30 cmH2O, it indicates a mismatch between the endotracheal intubation cuff and the patient's airway, and the endotracheal tube should be replaced.

6. The method for detecting air leakage in a mechanically ventilated respiratory system according to claim 5, characterized in that, In the third-level assessment, if the airbag pressure decreases within 1 minute after inflation, the airbag is considered ruptured and the endotracheal tube should be replaced.

7. The method for detecting air leakage in a mechanically ventilated respiratory system according to claim 1, characterized in that, If the air leak and suspected pneumothorax are still not resolved after completing the three-level progressive path, then imaging examinations should be performed.

8. The method for detecting air leakage in a mechanically ventilated respiratory system according to claim 1, characterized in that, The multi-parameter fusion analysis includes: The waveform was analyzed using a leakage feature model established using machine learning algorithms; The leakage characteristic model identifies leakage based at least on the waveform distortion rate and pressure attenuation slope.

9. The method for detecting air leakage in a mechanically ventilated respiratory system according to claim 7, characterized in that, The method further includes: The judgment threshold used for the air leakage feature model is dynamically adjusted according to the patient type.

Images