Research On Mechanical Ventilation Assisted Decision Based On Fuzzy Logic

Objective: Mechanical ventilation is a clinical widely used life support method for correcting hypoxemia and hypercapnia disease. 80% of patients who stay for more than 24 hours in the intensive care unit(ICU) require ventilator to assist respiration, most of which have respiratory failure and 13% of which suffer from acute lung injury(ALI) or acute respiratory distress syndrome(ARDS). As an important means of first aid, the mechanical ventilation plays an important role in special occasions such as battlefield wounded treatment and disaster relief. However, there are still many problems in the clinical application of mechanical ventilation: ①The wide application of mechanical ventilation is limited by many parameters, complex operation, and higher requirements for clinical users. ②The inappropriate mechanical ventilation parameter setting may cause or aggravate the acute lung injury in patients to form mechanical ventilation lung injury(VALI), even can increase the risk of dying of a patient. ③The mechanical ventilation parameters need to be adjusted timely according to the patient’s condition. The busy clinical work can lead to delaying the opportunity to adjust the mechanical ventilation parameters easily by users, which cause the patient’s condition is delayed.Clinically, a number of mature experience and clinical guidelines for mechanical ventilation have been summarized. However, these experiences and guidelines have not got extensive promotion and effective use, and the incorrect operation still occurs at times. In addition to the human and organizational factors, the individual differences of patients and state instability is also the important reason for the use chaos of mechanical ventilation. With the development of intelligent control technology, the application of advanced control theory to transfer the clinical guide into computer language to realize the intelligent control of ventilator becomes possible, and the control algorithm based on this purpose is the key to realize the intelligent control. This study is doing the relevant control algorithm research to realize the intelligent control ventilator to form aid decision making system for mechanical ventilation.Methods and contents: The aim of this study is mechanical ventilation intelligent control algorithm. By investigation of research hotspot and trend of intelligent control of mechanical ventilation, the computer model of human respiratory system is proposed and considered as the basis for realizing mechanical ventilation aided decision algorithm combined fuzzy control. Then, the research of mechanical ventilation aid decision making algorithm is divided into three main research contents:(4) Study on computer model of human respiratory system. The computer model of human respiratory system is divided into the mechanical model of the lung and the gas exchange model. The mechanical model of the lung mainly studies the characteristics of the pressure, capacity, lung compliance and airway resistance of the lung. Considering the complexity of the model, the simplified model of the lung electricity is adopted, which can truly reflect the physiological state of the patient and reduce the amount of calculation. Gas exchange model mainly simulates the exchange, transportation and metabolism of oxygen and carbon dioxide in human body. The gas exchange model of double cavity lung is used in the system, and the model includes five processes: alveolar ventilation, alveolar gas exchange, pulmonary shunt, human blood transportation, and tissue gas exchange, which can truly simulate the gas exchange in the patient and reflect the patient’s status. The computer model of human respiratory system is established to provide a reliable simulation platform for the assessment of the physiological state of lung of patients and influence of mechanical ventilation on patients.(5) Study on mechanical ventilation aided decision making algorithm based on fuzzy logic. The aided decision algorithm of mechanical ventilation is based on expert knowledge base and provides decision support for mechanical ventilation by using fuzzy logic to transform expert knowledge base into computer language. The research of the algorithm is divided into two parts: the establishment of expert knowledge base and the design of fuzzy logic controller. Expert knowledge base is a rule for the use of mechanical ventilation selected and summarized based on the existing mechanical ventilation clinical guidelines and the use of rules. The establishment of expert knowledge base is the clinical basis for the design of fuzzycontroller. According to the clinical practical needs, the fuzzy controller is designed for the multivariable fuzzy controller with 6 inputs and 4 outputs. In the light of problems such as complex structure of the controller, coupling between parameters, complex fuzzy control rules and large calculation amount in the design of multivariable fuzzy controller, a hierarchical multi rule structure is adopted in the design of the fuzzy controller to realize fuzzy controller. The design of the forthcoming fuzzy controller is divided into two parts: positive end expiratory pressure(PEEP)- inspiration oxygen concentration(Fi O2) and inspiration gas supported pressure(Pinsp)- respiratory frequency(F). The parameters setting of PEEP and Fi O2 are guided by the comprehensive judgment of PEEP and Fi O2. Such structure simplifies the complexity of the design of fuzzy control rules, reduces the amount of calculation and improves the accuracy of the fuzzy controller.(6) Study on patient mechanical ventilation status recognition algorithm. Physiological status of patients provides an important basis for guiding the settings of clinical mechanical ventilation parameters, with the recognition of physiological status on patients’ respiratory system as decisive basis for supporting decision-making algorithm. The lung pressure(P)- volume(V) curve of patients is an important clinical tool to study the internal features of patients’ lungs. Recognition algorithm uses the mechanical model of patients’ lungs to fit the P-V curve through Newton iteration method, in order to calculate significant indexes of lungs, such as dynamic lung compliance of patients. Patients’ oxygen saturation(SpO2)-inspiration oxygen concentration(Fi O2) or end-tidal oxygen concentration(FetO2) curve can show the matching conditions between intrapulmonary shunt and ventilation/perfusion during in vivo gas exchange. Intrapulmonary shunt(fs) and ventilation/perfusion ratio(f A2) are important parameters to show in vivo gas exchange of patients. The recognition algorithm can be used for calculating the value of fs and f A2 by fitting SpO2-Fi O2 curve or SpO2-FetO2 curve of patients through Newton iteration method and in combination with the gas exchange model of patients.Results: In order to verify the reliability of mechanical ventilation experts aid decision-making system, we cooperate with the Intensive Care Unit(ICU) of No. 174 People’s Liberation Army Hospital, adopt the method of resolution test before overall test to verify the decision-making system, verifying the system through three experiments on computer simulation platform in combination with actual clinical patient data:(4) Validity experiment of patients’ recognition algorithm and the computer model of respiratory system. Firstly, the state recognition algorithm is adopted to estimate patients’ state parameters based on actual clinical patient data and perfect the computer model of respiratory system. Then, physiological parameters of patients in current mechanical ventilation parameters are simulated by the model, and results are compared with actual patient physiological parameters. Through the comparison, we find that there is significant correlation between patients’ physiological parameters obtained and actual patient physiological parameters, without any significant differences in the difference values(P>0.05). Meanwhile, patients’ physiological parameters obtained through model simulation and actual patient physiological parameters under different strategies of mechanical ventilation are compared, finding that there is no significant difference between them. It is verified that the computer model of patients’ respiratory system established by the experiment can well simulate the condition of patients’ lungs.(5) Validity experiment of the fuzzy logic aid decision-making algorithm. The validity experiment of aid decision-making algorithm is performed by verified computer models. By establishing a closed-loop control system between the algorithm module and computer model, with the model simulating the change in patients’ state based on results of algorithm and the algorithm adjusting the output results based on the state change of the simulation, such system will provide the output of algorithm when reaching stability, as the decision-making suggestion for the final system. Through the experiment results, we can find that aid decision-making algorithm can make effective adjustment based on change in patients’ state, and the final decision-making suggestion completely meets the requirement of established expert knowledge base.(6) Evaluation experiment of algorithm performance. Currently, a complete quality evaluation method of mechanical ventilation has not been established clinically. Therefore, we perform a preliminary discussion for the quality evaluation method of mechanical ventilation in experiment, and finally evaluate the aid decision-making algorithm by clinicians in the form of questionnaires. Through the results of questionnaires, we can find that the results of aid decision-making algorithm meet relevant clinical requirements, so the algorithm can be used as a guide to mechanical ventilation strategy.Conclusion: In this study, expert aid decision-making system for mechanical ventilation is established through comprehensive utilization of fuzzy control and computer model of human respiratory system, acquisition and analysis of patients’ physiological mechanical ventilation parameters are realized, and the patients’ state is recognized by algorithms, thus providing decision-making suggestions for patients’ mechanical ventilation strategies. The system effectively reduces the difficulty in using respirators, facilitates clinical users to master patients’ state and offers timely and valid therapy schemes for mechanical ventilation.