Design of a Fuzzy-LABVIEW-Based Mechanical Ventilator Hasan GULER1, Fikret ATA1 1
Firat University, Electrical-Electronics Engineering Department, 23119, Elazig/TURKEY e-mails:
[email protected] ,
[email protected]
Abstract The purpose of the present study is to design a fuzzy-based mechanical ventilator and obtain respiration waveforms. An intelligent model for the mechanical ventilator was firstly developed in Matlab/Simulink after which a fuzzy-LabVIEW based mechanical ventilator prototype was practically designed. A fuzzy logic controller was used independently for both the simulation and experimental setup. In the simulation, three linear lung models namely the simple one-compartment, series two-compartment and parallel two-compartment models have been used to represent the lung. Inspiration and expiration times were estimated using the implemented Simulink model. Secondly, a fuzzy- LabVIEW based mechanical ventilator prototype was designed and estimation of inspiration/expiration time was made using 8 female Wistar Albino rats. It is seen that when lung resistance, R, increased in the simulation and the experiments, the inspiration time and pressure in the lung increased just like in practice. In addition to this, it can be said that irregular respiration rates do not create any problems in the ventilation of rats. Keywords: Fuzzy Logic, LabVIEW, Mechanical Ventilator, Respiration, Lung Mechanics.
1.
Introduction
Mechanical ventilators are widely used in intensive care units (ICU) to provide adequate oxygenation for blood and to stabilize the metabolism of patients. Oxygen is important for human beings since it provides energy. Respiration can be defined as the transportation of oxygen and carbon dioxide. It occurs in two phases: inspiration and expiration [1]. The process of taking air into the lungs is called inhalation or inspiration, and the process of breathing it out is called exhalation or expiration. The respiration process is artificially carried out via mechanical ventilators in patients. A mechanical ventilator is a machine designed to move breathable air into and out of the lung
1
mechanically. Ventilation can be done three in different ways [2-3]. These are negative pressure ventilation, positive pressure ventilation and high frequency ventilation. There are several papers about the control of mechanical ventilators in the literature. In the beginning, researchers developed an open loop control technique for ventilators. However, because of the fact that this control increases the workload of clinicians, researchers focused on closed loop control to reduce the workload of clinicians. This technique offers many advantages to both patients and clinicians. There have been many attempts to control respiratory parameters in order to support clinicians. Laubscher T.P et al. implemented the adaptive lung ventilation process with closed loop control [4]. By doing this, they developed three different lung models to calculate respiration rate, tidal volume and inspiration pressure. Stegmaier and Zollinger A. developed a special open loop control ventilator and monitored real time airway signals and lung function data [5]. The experiments measured the flow, pressure and CO2 concentration of airway signals, as well as the total capacitance and resistance and instant PEEP (Positive Expiration End Pressure) of lung function in eight patients. Laubscher T.P et.al monitored respiration rate and tidal volume using PSIMV (Pressure Synchronized Intermittent Mandatory Ventilation) mode [6]. Cappa and Scuito developed an automatic measurement system for newborn babies [7]. Ventilation parameters and pressure variety in patients were measured with this system. Chatburn revealed an electrical circuit for the relationship between the ventilator and patient and exemplified open and closed loop control [8]. While Rios and Tafur managed to mix oxygen and air and to send it to the ventilator using a lung-ventilator prototype [9], Luepschen et al. designed the PID (Proportional-Integral-Derivative) controller to control SaO2 [10]. Hooven et al. made a simulation for a lung and system model and monitored inspired O2 and N2O from the computer [11]. Ahmadi and Bates implemented the inspiration and expiration process via open/closed vane using developed software [12]. After artificial intelligence techniques such as fuzzy logic and neural networks were successfully applied to many systems, these techniques have begun to be widely used in biomedical applications in recent years. Rees et al. developed a decision support system for the parameters of ventilators used in intensive care units. They merged fuzzy and
2
conventional controllers and PI in their system [13]. Stegmaier et al. developed a system that tracks patients’ coughs. Measured pressure is converted to digital form via ADC and is sent to fuzzy logic [14]. Nemoto et al. implemented patients’ weaning process from a ventilator using fuzzy logic according to parameters such as heart rate, tidal volume, breathing rate and SaO2 [15]. They established ninety-six rules to determine the healing/deteriorating trend of the patients. Nelson et al. performed a study to control respiration rate and SaO2 by fuzzy logic [16]. They developed seven different programs in Matlab/Simulink and investigated which one gave good results. H.F.Kwok et al. developed a hybrid algorithm for mechanical ventilators and simulated it into Matlab/Simulink [17]. There are two parts in their system. In the first part the system changes the parameters by itself while in the second part, the parameters to be changed are recommended to clinicians. Zhu and Moller controlled the nonlinear respiration system via a neural-fuzzy system [18]. Whereas flow is controlled using fuzzy logic, the respiration system is controlled via neural networks. Tzavaras et al. performed a study to model respiration rate and tidal volume by ANFIS for COPD (Chronic Obstructive Pulmonary Disease) patients [19]. They used real patient data to train and test the artificial neural network. Jong–Chen et al. developed an intelligent system to assist clinicians in making decisions about the weaning of patients from a ventilator [20]. They evaluated 27 parameters to wean a patient off a ventilator. Mogensen et al. developed a lung model to investigate the link between airway pressure, lung volume and perfusion [21]. Inspiration and expiration time are important during ventilation because those times give a clue about the patient’s condition. Up to now, there have been studies about determining the respiration rate as a whole (total inspiration and expiration time) but there are no studies about determining the inspiration and expiration time separately. These parameters help to achieve an adequate minute tidal volume with the lowest possible airway pressure. Firstly, in this study a lung and mechanical ventilator was modeled to determine inspiration and expiration times separately. Three basic lung models were used to represent the lung. These models are one-compartment, series two-compartment and parallel two-compartment models [22-24]. The lung models and mechanical ventilator
3
models were combined into a simulation. The developed system was controlled via fuzzy logic in the simulation. The inspiration and expiration times were estimated according to the pressure differences in the rat lung model. Finally, a fuzzy-based mechanical ventilator was designed. The designed ventilator can be run in different ventilation modes, including Pressure Controlled Ventilation (PCV), Pressure Support Ventilation (PSV), Intermittent Mandatory Ventilation (IMV) and Synchronized Intermittent Mandatory Ventilation (SIMV) with the help of the implemented software. 8 female Wistar albino rats were used to test the fuzzy-LabVIEW based ventilator and to determine inspiration/expiration time. In the rat experiments the pressure controlled ventilation mode was used. The used rats were healthy animals and so the respiration system of the rats was first disturbed using chemical agents to induce neuromuscular diseases in the rats. The experiments were carried out with the approval of Firat University Animal Ethics Committee.
2.
Materials and Methods A. Lung Mechanics
Many researchers define the lung as a mechanical system and many papers have already been published about it. In these studies the lung mechanics are represented by electrical circuits [22-24]. Many variables like pressure, volume and flow are represented by voltage, charge and current respectively. In reviewing the literature, it was observed that R-C circuit equipment is commonly used to model the lung. Therefore, in this study, three different lung models were used: one compartment model, series twocompartment model and parallel two-compartment model. The single compartment model is shown in Figure 1 [22]. An electrical analogy schematic diagram of the one compartment model of the lung is given in Figure 2. Since this model is anatomically analogous to a real lung, it is preferred.
4
V(t)
E
R
Figure 1.One-compartment model of the lung
The general pressure equation for this figure is P = E.V + R.
(1)
Where E is elastance (1/C), R is flow resistance; V is volume and
is the derivative of
V with respect to time. R Pao C
Patm
Figure 2. Electrical circuit for the one-compartment model
Pao and Patm represent pressure at the airway opening and atmosphere pressure, respectively. While R is the lung resistance, C is the lung compliance. The transfer function of this model is obtained from equation (2) and equation (3) represents the final transfer function of the one-compartment lung model. (2)
(3)
5
R=0.2 cmH2O.S.L-1, C=0.01 l.cmH2O-1 were chosen to determine the transfer function. These R and C values belong to the rat lung [26]. A rat lung is about 10 times smaller than a human lung. Thus, its resistance and elastance values are about 10 times smaller. Another electrical analogy diagram of the lung is the series two-compartment model. A figure of this diagram is given in Figure 3. R1
R2
Pao
C1
C2
Patm
Figure 3. Electrical circuit for the series two-compartment model
R1, R2, C1, C2 are lung resistance and lung compliance respectively. The transfer function is
(4)
The parallel two-compartment lung model is shown in Figure 4 and its transfer function is given in equation (5). Rc
R2
Pao
R1
C2
C1 Patm
Figure 4. Electrical circuit for the parallel two-compartment model
In this model, there is only one difference from the series lung model. Rc is central airway resistance and the other parameters are the same as in the series lung model.
6
(5) While obtaining the transfer function of these two models, the parameters were chosen for the mean values as below: Rc=R1=R2=0.1 cmH2O.S.L-1, C1=0.001 l.cmH2O-1, C2=0.01 l.cmH2O-1. C in the one-compartment model corresponds to C2 in the two-compartment model. B. Fuzzy Logic Artificial intelligence (AI) control techniques have been successfully applied to various control applications in recent years. As one of the AI techniques, the fuzzy system has been widely used in biomedical applications for decades. Prof. Lotfi Zadeh first introduced fuzzy sets in 1965. After the success of the fuzzy system in the studies of many scientists, it became popular in the literature and has been applied in different fields such as control, communication, integrated circuit production and medicine. Fuzzy logic controllers have been used in several applications due to both their theoretical and practical success [13-21, 26-27]. Because of the fact that respiratory system of living-being is so complex and does not be modeled properly, classical control techniques such as PI and PID do not give necessary response for respiratory system. For this reason, instead of classical control, fuzzy control, which is based on expert opinion, was used to control living-being’s respiratory system. Clinician evaluated all possibility when determining the rule table of fuzzy system. Expert opinion plays an important role in the treatment of patients in ICUs due to the complexity and vagueness of the respiratory system. Clinicians determine the method of treatment according to the needs of the patient. Since the fuzzy system includes expert opinions, fuzzy control was chosen to control system variables. In this paper fuzzy was used to determine the inspiration and expiration time of the rat model in the simulations and rats in the experiments. Fuzzy has two inputs which are the error; the difference between the constant reference pressure and gas pressure arriving at the patient, and the change in the error. The block diagram of the system is shown in Figure 5. The membership functions of fuzzy inputs and outputs are shown in Figure 6 and 7.
7
Pe(k) = Pref – Pill
(6)
Pe(k) = Pe(k) - Pe(k-1)
(7)
+
Pref
Pe(k)
Fuzzy Logic Controller
Pill
Mechanical Ventilator
Pe(k)
+
Pe(k-1) -1
Z
Figure 5. System block diagram µpe(k) NB
-40
NS
-20
Z
0
µδpe(k) PS
PB
20
40
NS
NB
-10
cmH2O
Z
-5
PS
0
PB
5
10
cmH2O
Figure 6. Membership functions of inputs µins
µexp
INS1
0
INS2
INS3
1
2
INS4
3
INS5
4
INS6
5
INS7
6
EXP1 EXP2
sec
0
1
EXP3
EXP4
2
3
EXP5
4
EXP6
EXP7
5
6
sec
Figure 7. Membership functions of the outputs
Rule tables for inspiration and expiration times are shown in Table 1 and 2. While these rules were being created, it was considered that lungs have different specifications for each rat. Twenty-five rules were separately created for inspiration and expiration according to the requirement of each rat in order to obtain the inspiration and expiration times. These rules were created by taking the opinions of experts.
8
Table 1. Rule tables for inspiration time e(k)/δe(k)
NB
NS
Z
PS
PB
NB
INS1
INS1
INS2
INS3
INS4
NS
INS1
INS2
INS3
INS4
INS5
Z
INS2
INS3
INS4
INS5
INS6
PS
INS3
INS4
INS5
INS6
INS7
PB
INS4
INS5
INS6
INS7
INS7
Table 2. Rule tables for expiration time e(k)/δe(k)
NB
NS
Z
PS
PB
NB
EXP7
EXP7
EXP6
EXP5
EXP4
NS
EXP7
EXP6
EXP5
EXP4
EXP3
Z
EXP6
EXP5
EXP4
EXP3
EXP2
PS
EXP5
EXP4
EXP3
EXP2
EXP1
PB
EXP4
EXP3
EXP2
EXP1
EXP1
C. Experimental Setup The fuzzy-based mechanical ventilator was designed as shown in Figure 8. The controller of this system was implemented with NI cRIO-9073. In the software LabVIEW and fuzzy logic controlled the valves and the regulator. Thus, the device is called an intelligent mechanical ventilator. The prototype mechanical ventilator is shown in Figure 9.
Figure 8. Block diagram of the mechanical ventilator 9
Figure 9. The designed intelligent mechanical ventilator and rat
The mean weights of the rats used in the experiment were 225 ± 25 gr. The respiratory variables of the rats were chosen in accordance with those stated in the relevant literature [28- 29]. Therefore, FiO2: 1(Fraction of Inspired Oxygen), PEEP: 0 cmH2O (Positive End Expiratory Pressure), tidal volume: 6 ml/kg were used in the experiments. The pressure regulator was set to 30 cmH2O. A software-based interface was developed in LabVIEW in order to monitor the respiratory variables of the rats, as shown in Figure 10.
Figure 10. The developed software-based interface to monitor respiratory variables
The healthy rats were firstly anesthetized with 0.3 ml ketamine and a tracheotomy operation was carried out to enable artificial respiration. Afterwards, 0.3 ml Blok-L was
10
injected to paralyze the respiratory system of the rats. Thus, the device carried out respiration for the rats.
3.
Results
Simulations were carried out using Matlab/Simulink. In this paper, three different lung models were used to represent the rat lung. These are the one-compartment model, the series two-compartment model and the parallel two-compartment model. A block diagram of the simulation for the one-compartment model is shown in Figure 11.
Figure 11. Simulink model of the system for the one-compartment model
The total respiration time was chosen as 5 seconds. In the simulations lung compliance was fixed for 0.01 L.cmH2O-1 but lung resistance varied between 0.2 and 0.4 cmH2O.L-1. Respiration waveforms for the different coefficients of all lung models can be seen in Figures 12, 13, 14.
11
35 R=0.2 cmH2O.S.L-1, C=0.01 l.cmH2O-1 R=0.3 cmH2O.S.L-1, C=0.01 l.cmH2O-1 R=0.4 cmH2O.S.L-1, C=0.01 l.cmH2O-1
30
Lung Pressure-cmH2O
25
20
15
10
5
0
0
0.5
1
1.5
2
2.5
3
3.5
4
4.5
5
time-sec
Figure 12. The obtained respiration waveform for the one-compartment model
40 R=0.2 cmH2O.S.L-1, C=0.001 l.cmH2O-1 R=0.3 cmH2O.S.L-1, C=0.001 l.cmH2O-1
35
R=0.4 cmH2O.S.L-1, C=0.001 l.cmH2O-1
Lung Pressure-cmH2O
30
25
20
15
10
5
0
0
0.5
1
1.5
2
2.5 time-sec
3
3.5
4
4.5
5
Figure 13. The obtained respiration waveform for the series two-compartment model
12
40 R=0.2 cmH2O.S.L-1, C=0.001 l.cmH2O-1 R=0.3 cmH2O.S.L-1, C=0.001 l.cmH2O-1
35
R=0.4 cmH2O.S.L-1, C=0.001 l.cmH2O-1
Lung Pressure-cmH2O
30
25
20
15
10
5
0
0
0.5
1
1.5
2
2.5 time-sec
3
3.5
4
4.5
5
Figure 14. The obtained respiration waveform for the parallel two-compartment model
Inspiration and expiration times were calculated for three different lung resistances which are R=0.2, 0.3, 0.4 cmH2O.S.L-1. For the simulations when lung resistance, R, increased in the simulation, the inspiration time and pressure in the lung increased just like in practice. The shortest inspiration time among the simulations was obtained for the one-compartment model for R=0.2 cmH2O.S.L-1 and the longest time was for the parallel two-compartment model for R= 0.4 cmH2O.S.L-1. The change in inspiration time due to the change in lung resistance-R is shown in Figure 15. As can be seen in this figure, when lung resistance increased, the inspiration time started to increase for all lung models. The shortest inspiration time was obtained in the one-compartment model, while the longest was obtained in the parallel twocompartment model. Mean inspiration time value was 2.112 sec. for the onecompartment model, 2.254 sec. for the series two-compartment model and 2.289 sec. for the parallel two-compartment model.
13
2.4
2.35
one compartment model series two compartment model parallel two compartment model
inspiration time-sec
2.3
2.25
2.2
2.15
2.1
2.05
2 0.2
0.22
0.24
0.26 0.28 0.3 0.32 0.34 R-Lung Resistance-cmH2O.S.L-1
0.36
0.38
0.4
Figure 15. Inspiration time vs R-lung resistances
In the intelligent mechanical ventilator, fuzzy logic membership functions were chosen to be the same as those for the simulations with the only difference being the respiration time. The respiration time was 5 seconds in the simulation, while it was 0.75 seconds in the experimental setup. The obtained respiration waveforms from the rats are shown in Figures 16, 17.
14
30 Rat1 Rat2 Rat3 Rat4
Lung Pressure-cmH2O
25
20
15
10
5
0
0
0.1
0.2
0.3
0.4 time-sec
0.5
0.6
0.7
0.8
Figure 16. Obtained respiration waveform from Rat 1 to Rat 4
30 Rat Rat Rat Rat
Lung Pressure-cmH2O
25
5 6 7 8
20
15
10
5
0
0
0.1
0.2
0.3
0.4 time-sec
0.5
0.6
0.7
0.8
Figure 17. The obtained respiration waveform from Rat 5 to Rat 8
Lung resistances of the rats were calculated using equation (8) [30].
(8)
15
Where Ppik is peak pressure, Pplato is the pressure applied to the small airways and alveoli during positive pressure mechanical ventilation and
is flow. The calculated
lung resistance of the rats and the estimated inspiration and expiration time are shown in Table 3 and Table 4, respectively. Table 3. The calculated lung resistances of rats
The calculated lung resistances (cmH2O.S.L-1)
Rat1
Rat2
Rat3
Rat4
Rat5
Rat6
Rat7
Rat8
0.873
0.910
0.892
0.783
0.981
0.952
0.967
0.924
Table 4. Estimated inspiration and expiration times
Rat1 Rat2 Rat3 Rat4 Rat5 Rat6 Rat7 Rat8
Inspiration time (sec)
Expiration time (sec)
0.268 0.285 0.280 0.255 0.307 0.281 0.283 0.279
0.482 0.465 0.470 0.495 0.443 0.469 0.467 0.471
It can be seen from Figures 16-17 that if the lung resistance of a rat is higher, then the inspiration time is longer and the peak pressure is higher. The shortest inspiration time and the smallest peak pressure were obtained in rat 4, whose lung resistance was the lowest, while the longest inspiration time and the highest peak pressure were obtained in rat 5, whose lung resistance was the highest. It was observed from the simulation and experiments that the results were consistent and that the inspiration time and peak pressure in the lung increased with increasing lung resistance.
4.
Discussion and Conclusion
Mechanical ventilation is often used to treat patients who require breathing assistance in ICUs. In hospitals the settings of the ventilator are adjusted manually by clinicians according to patients’ physiological status. To decrease the workload of clinicians, many researchers have focused on automatic ventilator control. They have developed or designed many systems for ventilation. This paper is one of them. The study consists of two phases. The first phase is simulation and the second phase is experiment. Initially 16
the lung mechanics and mechanical ventilator were modeled in Matlab/Simulink. Lung mechanics are generally represented by electrical circuits. There is no general model to represent the lung in the literature. Thus, three broadly accepted lung models were used to represent the lung in this simulation. These are the one-compartment lung model, series two-compartment model and parallel two-compartment model. In the mechanical ventilator model, the respiration rate was selected for 12 breaths per minute. Therefore, total respiration time was 5 seconds. The objective of this simulation was to show the interaction between the ventilator and lung model. While doing this, fuzzy logic was used. The reason why fuzzy was chosen is that respiratory system is so complex and modeling of it is not easy. Thus, fuzzy was used to control the respiratory system. Fuzzy logic has two inputs and two outputs. The fuzzy inputs are the error and change in error. The fuzzy outputs are the inspiration and expiration times. Fuzzy logic controlled the inspiration and expiration valves in the ventilator block. According to the simulation results, it was observed that the inspiration time was longer than that of expiration in the case of a hard and resistive lung. In the event that the lung was flexible and could expand easily, it was observed that the expiration time was longer than the inspiration time. The shortest inspiration time was observed in the one-compartment model, while the longest was obtained for the parallel two-compartment model. Secondly, the mechanical ventilator prototype was designed using NI c-RIO 9073 chassis, analog/digital input and output units. The whole system was controlled via fuzzy logic. While creating fuzzy membership function and rules, LabVIEW 2010 academic software was used. Membership functions and rule tables were the same as in the simulation excluding those for respiration time. Respiratory speed was selected as 80 breaths per minute. Thus, the respiration period was 0.75 seconds. According to the experiment results, the shortest inspiration time was obtained in rat 4, whose lung resistance was the lowest, while the longest inspiration time was obtained in rat 5, whose lung resistance was the highest. It can be said that the inspiration time starts to increase when lung resistance increases. In this paper, it was attempted to estimate inspiration and expiration times separately. The conducted studies in the literature have usually focused on estimating or
17
calculating respiration rate as a whole. In these studies, inspiration/expiration rate (I/E) has been generally accepted to be two. However, it is known that the I/E rate is not two even for healthy people with no respiratory problems. In this study, I/E rate was not accepted to be constant. Therefore, inspiration and expiration times were separately estimated in the simulations and experiments. The simulation results are in agreement with the experimental results. The experiments on the rats showed that inconstant respiration rates do not create any problems in the ventilation and this revealed that treatment can be done in this way. This will create future research-driven protocols.
Acknowledgements This study is part of a project funded by FUBAP grant no.1911. The authors would like to thank Prof. Jason H.T. BATES and the doctors at USA-Vermont Lung Center for providing assistance in order to understand lung mechanics and also Prof. Servet KILINC and Asst. Prof. Kadri KULUALP for the tracheotomy operation.
References 1. A. Perel, Stock Mc (1992): Handbook of Mechanical Ventilatory Support. 1st Ed. Williams and Wilkins, Philadelphia. 2. R.A Smith (1986), Respiratory Care Mechanical Anesthesia 2nd Ed. Churchill Livingstone, New York. 3. Pillbeam SP (1992), Mechanical ventilation: Physiological and Clinical Application, 2nd Ed. St Louis, Mosby-Year. 4. T.P. Laubscher, W. Heinrichs, N. Weiler, G. Hartmann, J.X. Brunner (1994), An Adaptive Lung Ventilation Controller, IEEE Transactions on Biomedical Engineering Vol.41,51-58. 5. P.A Stegmaier, A. Zollinger (1995), A Ventilator Workstation for Simultaneous Recording of Lung Function Indices and Airway Signals, IEEE-EMBC and CMBEC Clinical Engineering/Medical Informatics Conference, 731-732. 6. T.P. Laubscher, W. Heinrichs, N. Weiler, G. Hartmann, J.X. Brunner (1992), The Minimal Alveolar Ventilation Controller, Engineering in Medicine and Biology Society, Proceedings of the Annual International Conference of the IEEE,Vol.6, 2711-2712. 7. P. Cappa, S.A. Scuito (2000), Experimental Analysis of the Airway Circuit Effects on Breathing Pattern Generates by Neonatal Pulmonary Ventilators, Proceedings-22nd International Conference-IEEE/EMBS-Vol.4, 3132-3135.
18
8. R.L. Chatburn (2003), Engineering Principles Applied to Mechanical Ventilator, Proceedings- 25th International Conference-IEEE/EMBS,Vol.1, 406-410. 9. C.A. Rios, J.C. Tafur (2003), Mathematical Model and Control of the Pneumatic System of a Lung Ventilator Prototype, Proceedings-25th International Conference-IEEE/EMBS, Vol.3, 2776-2779. 10. H. Luepschen, L. Zhu, S. Leonhardt (2007), Robust Closed Loop Control of the Inspired Fraction of Oxygen for the Online Assessment of Recruitment Maneuver, Proceedings-29th International Conference-IEEE/EMBS,495-498. 11. S.V.D. Hoeven, S. Duncan, A. Farmery, C. Hann (2007), Internal Model Control of Inhaled Anesthesia, American Control Conference ACC-07, 582-587. 12. M. Ahmadi, J.H.T. Bates (1999), A Computer-Controlled Mechancal Ventilator Based on A Rotating Vane, Engineering in Medicine and Biology, 21st Annual Conf. and the Annual Fall Meeting of the Biomedical Engineering Soc. BMES/EMBS Conference, Vol.1, 336-337. 13. S.E. Rees, S. Kjaergaard, P. Thorgaard, T. Egon, S. Andreassen (2003), A Physiological Model Based Approach to Medical Decision Support in the Intensive Care Unit, Engineering in Medicine and Biology Society, Proceedings of the 25th Annual International Conference of the IEEE,Vol.1, 432-433. 14. P.A. Stegmaier, J.X. Brunner, N.N. Tschichold, T.P. Laubscher, W. Liebert (1994), Fuzzy Logic Cough Detection: A First Step Towards Clinical Application, Fuzzy Systems, IEEE World Congress on Computational Intelligence., Proceedings of the Third IEEE Conference, 1000-1005. 15. T. Nemoto, G.E. Hatzakis, C.W. Thorpe, R. Olivenstein, S. Dial, J.H.T. Bates (1999), Automatic Control of Pressure Support Mechanical Ventilation Using Fuzzy Logic, American Journal Respiratory and Critical Care Medicine,Vol.160, 550-556. 16. D.S. Nelson, J.H. Strickland, T.C. Jannet (1997), Simulation of Fuzzy Control for Management of Respiratory Rate in Assist Control Mechanical Ventilation, Proceedings-19th International Conference-IEEE/EMBS-Vol.3, 1104-1107. 17. H.F. Kwok, D.A. Linkens, M. Mahfouf, G.H. Mills (2004), SIVA: Hybrid knowledge and model based advisory system for intensive care ventilators, IEEE Transactions on Information Technology in Biomedicine,Vol.8, 161-171. 18. H. Zhu, K. Möller (2008), Ventilator Control Based on Fuzzy-Neural Network Approach, Bioinformatics and Biomedical Engineering, ICBBE 2008, 747-750. 19. A. Tzavaras, P.R. Weller, B. Spyropoulos (2007), A Neuro-Fuzzy Controller for the Estimation of Tidal Volume and Respiration Frequency Ventilator Settings for COPD Patients Ventilated in Control Mode, Proceedings of the Annual International Conference of the IEEE-EMBS, 3765-3768.
19
20. J.C. Chen, S.W. Chien, J. Hsu (2009), A Study On Weaning Results Of VentilatorDependent Patients With An Artificial Neuromolecular System , Biomedical Engineering: Applications, Basis and Communications, Vol. 21, 253-263. 21. M.L Mogensen., K.S.Steimle, D.S. Karbing, S. Andreassen (2011), A model of Perfusion of the Healthy Human Lung, Computer Methods and Programs in Biomedicine, 101,156–165. 22. J.H.T. Bates (2009), Lung Mechanics-An Inverse Modeling Approach, 1st Ed., Cambridge University Press, New York. 23. E.Saatcı, A.Akan (2008), Respiratory Parameter Estimation in Linear Lung Models, 30th Annual International IEEE EMBS Conference. 24. E.Saatcı, A.Akan (2007), Lung Model Parameter Estimation by Unscented Klman Filter, Proceedings of the 29th Annual International Conference of IEEE EMBS. 25. C.Thamrin, P.D.Sly, Z.Hantos (2005), Broadband Frequency Depence of Respiratory Impedance, Journal of Applied Physiology, 99(4), 1364-1371. 26. Z.Tang, M.J. Er and F.Qi (2011), Dynamic Fuzzy-Neural Network for the Intelligent Control Humanoid Robot, Computer Systems Science & Engineering, Vol.26:3. 27. P. Khatri, S. Tapaswi, U.P. Verma (2014), Trust evaluation in wireless ad-hoc networks using fuzzy system, Computer Systems Science & Engineering, Vol.29:1. 28. F.Eroğlu, H.E.Eroğlu (2011), Ratlarda Aneljezi ve Anestezi, Journal of Clinical and Analytical Medicine. 29. S.Tanju (2010), Solunum Sistemi Hastalıklarında Deneysel Hayvan ModelleriDeney Hayvanı Kullanım Teknikleri, Türk Toraks Derneği 13.yıllık Kongresi. 30. C.S.Wang, D. Shaw, K.S.Jih (1998), An Intelligent Control System for Ventilators, Medical Engineering & Physics, Vol. 20, 534-542.
20
