MECHANICAL VENTILATION SYSTEM FOR RESPIRATION WITH DECISION SUPPORT

Abstract

The invention relates to a mechanical ventilation system (10) for respiration of a patient (5), the system being adapted for providing decision support for mechanical ventilation. Control means (12) is adapted for using both first data (D1) and second data (D2), indicative of the respiratory feedback in the blood, in physiological models (MOD) descriptive of, at least, lung mechanics, and/or gas exchange in the lungs of the patient, the physiological models comprising a number of model parameters (MOD_P). The control means is further arranged for simulating the effect on one, or more, model parameters (MOD_P) of the physiological models for a suggested value of the positive end expiratory pressure (PEEP) setting for the ventilation means, and thereby provide decision support in relation to said suggested PEEP value. The invention is advantageous for providing mathematical based models of changes in physiology in response to changes in ventilator settings of the PEEP thereby allowing mathematical physiological models to predict changes in clinical variables for a given value of PEEP.

Claims

1. A mechanical ventilation system for respiration of an associated patient, the system being adapted for providing decision support for mechanical ventilation, the system comprising: ventilation means capable of mechanical ventilating said patient with air and/or one or more medical gases, the ventilation means having a plurality of settings (V_SET) comprising a positive end expiratory pressure (PEEP) setting, control means, the ventilator means being controllable by said control means by operational connection thereto, and measurement means arranged for measuring the inspired gas and/or the respiratory feedback of said patient in the expired gas in response to the mechanical ventilation, the measurement means being capable of delivering first data (D1) to said control means, wherein the control means is adapted for using both the first data (D1) and second data (D2) indicative of the respiratory feedback in the blood in physiological models (MOD) descriptive of, at least, lung mechanics, and/or gas exchange in the lungs of the patient, the physiological models comprising a number of model parameters (MOD_P), and wherein the control means is further arranged for simulating the effect on one, or more, model parameters (MOD_P) of the physiological models for a suggested value of the positive end expiratory pressure (PEEP) setting for the ventilation means, and thereby provide decision support in relation to said suggested PEEP value.

2. The mechanical ventilation system according to claim 1, wherein the control means is further arranged for simulating the effect on one, or more, model parameters (MOD_P) of the physiological models for a plurality of values (PEEP; 1, . . . ,n) of the positive end expiratory pressure setting for the ventilation means, and thereby provide decision support in relation to said plurality of PEEP values.

3. The mechanical ventilation system according to claim 2, wherein the control means is further arranged for suggesting an optimum value between the plurality of values of the positive end expiratory pressure setting for the ventilation means (PEEP; 1, . . . ,n).

4. The mechanical ventilation system according to claim 1, wherein the control means is further arranged for simulating the effect on one, or more, parameters (MOD_P) of the physiological models for one, or more, values in the positive end expiratory pressure setting for the ventilation means (PEEP) performed by a simulation based on at least two previous values of the PEEP setting for the ventilation means.

5. The mechanical ventilation system according to claim 1, wherein the control means is further arranged for simulating the effect on one, or more, parameters (MOD_P) of the physiological models for one, or more, values in the positive end expiratory pressure setting for the ventilation means (PEEP) performed by a simulation based on at least two simulated values of the PEEP setting for the ventilation means.

6. The system according to claim 1, wherein the control means comprises one, or more, positive end expiratory pressure (PEEP) models, each PEEP model comprising a model parameter (MOD_P) of a physiological model as a function of the PEEP settings for the ventilation means.

7. The system according to claim 6, wherein one, or more, of the PEEP models are adapted to the patient, and/or the clinical condition of the patient, before initiating changes of the PEEP settings, and/or while changing the PEEP settings.

8. The system according to claim 7, wherein the one, or more, PEEP models are adapted to the patient by a learning module, the learning module having a set of a priori settings of PEEP model parameters (MOD_P_PEEP) based on the currently measured model parameter (MOD_P) value of the corresponding physiological models (MOD), preferably based on a Bayesian distribution based on patient type and/or clinical condition.

9. The system according to claim 1, wherein one, or more, of the physiological models (MOD) are adapted to the patient, and/or the clinical condition of the patient, before initiating changes of the PEEP settings, and/or while changing the PEEP settings.

10. The system according to claim 1, wherein one, or more, additional physiological models (MOD) are further descriptive of the metabolism of the patient, the blood circulation of the patient, acid-base status of the patient, oxygen and/or carbon dioxide transport for the patient, and/or the respiratory drive of the patient.

11. The system according to claim 10, wherein at least two physiological models are integrated by having one, or more, variables in common.

12. The system according to claim 1, wherein the control means comprises one, or more, modules for choosing a PEEP changing strategy for the patient.

13. The system according to claim 12, wherein one choice of PEEP changing strategy is based on an assumption of lung recruitment wherein a relatively high airway pressure is applied for a relatively short time followed by stepwise PEEP reduction until the optimal balance is reached.

14. The system according to claim 12, wherein one choice of PEEP changing strategy is based on a stepwise increase of PEEP until an optimal balance is reached.

15. The system according to claim 1, wherein the control means further comprises a plurality of clinical preference functions (CPFs) relating settings of positive end expiratory pressure (PEEP) for the ventilation means to a corresponding set of clinical outcome variables.

16. The system according to claim 15, wherein the plurality of clinical preference functions (CPFs) are chosen from the group consisting of: CPFs inserted by a clinician, CPFs chosen from a database of possible CPFs, CPFs a priori adapted to a specific patient based on general clinical input from a clinician, and CPFs a priori adapted to a specific patient according to patient needs.

17. The system according to claim 15, wherein the plurality of clinical preference functions (CPFs) is applied for providing decision support related to an overall optimisation of the PEEP setting of the mechanical ventilation for the patient.

18. The system according to claim 1, wherein the positive end expiratory pressure (PEEP) setting is further optimized with respect to other mechanical ventilation settings, preferably inspired oxygen (FIO2), tidal volume (VT), pressure above PEEP, and/or respiratory frequency.

19. A decision support system for providing decision support to an associated mechanical ventilation system for respiration aid of a patient, the mechanical ventilation system comprising: ventilation means capable of mechanical ventilating said patient with air and/or one or more medical gases, the ventilation means having a plurality of settings (V_SET) comprising a positive end expiratory pressure (PEEP) setting, and measurement means arranged for measuring the inspired gas and/or the respiratory feedback of said patient in the expired gas in response to the mechanical ventilation, the measurement means being capable of delivering first data (D1) to said decision support system, wherein the decision support system is adapted for using both the first data (D1) and second data (D2) indicative of the respiratory feedback in the blood in physiological models (MOD) descriptive of, at least, lung mechanics, and/or gas exchange in the lungs of the patient, the physiological models comprising a number of model parameters (MOD_P), and wherein the decision support system is further arranged for simulating the effect on one, or more, model parameters (MOD_P) of the physiological models for a suggested value of the positive end expiratory pressure (PEEP) setting for the ventilation means, and thereby provide decision support in relation to said suggested PEEP value.

20. A method for operating an mechanical ventilation system for respiration of an associated patient, the system being adapted for providing decision support for mechanical ventilation, the method comprising: providing ventilation means capable of mechanical ventilating said patient with air and/or one or more medical gases, the ventilation means having a plurality of settings (V_SET) comprising a positive end expiratory pressure (PEEP) setting, providing control means, the ventilator means being controllable by said control means by operational connection thereto, and providing measurement means arranged for measuring the inspired gas and/or the respiratory feedback of said patient in the expired gas in response to the mechanical ventilation, the measurement means being capable of delivering first data (D1) to said control means, wherein the control means is adapted for using both the first data (D1) and second data (D2) indicative of the respiratory feedback in the blood in physiological models (MOD) descriptive of, at least, lung mechanics, and/or gas exchange in the lungs of the patient, the physiological models comprising a number of model parameters (MOD_P), and wherein the control means is further arranged for simulating the effect on one, or more, model parameters (MOD_P) of the physiological models for a suggested value of the positive end expiratory pressure (PEEP) setting for the ventilation means, and thereby provide decision support in relation to said suggested PEEP value.

21. A computer program product being adapted to enable a computer system comprising at least one computer having data storage means in connection therewith to control a mechanical ventilation system according to the method in claim 20.

Description

BRIEF DESCRIPTION OF THE FIGURES

[0083] The invention will now be described in more detail with regard to the accompanying figures. The figures show one way of implementing the present invention and is not to be construed as being limiting to other possible embodiments falling within the scope of the attached claim set.

[0084] FIG. 1 shows patient examples for various settings of PEEP and the corresponding changes in physiological parameters,

[0085] FIG. 2 is a schematic drawing of a mechanical ventilation system according to the present invention,

[0086] FIG. 3 is a drawing of the decision support for providing decision support to an associated mechanical ventilation system (not shown),

[0087] FIG. 4 shows graphs of the physiological model parameters, ventilator settings, simulated patient changes, and clinical preference functions (CPF) as a function of the PEEP,

[0088] FIG. 5 shows graphs of PEEP models with learning of gas exchange,

[0089] FIG. 6 shows a graph of a PEEP model of lung compliance,

[0090] FIG. 7 shows a graph of other PEEP models of lung mechanics and metabolism,

[0091] FIG. 8 shows a flow chart with two PEEP changing strategies; recruitment or so-called crawl,

[0092] FIG. 9 shows the timing of the two PEEP changing strategies, recruitment or so-called crawl, from FIG. 8,

[0093] FIG. 10 shows the learning of gas exchange model parameters applying a Bayesian approach,

[0094] FIG. 11 shows a simulated patient case for a patient in a control ventilator mode where a preferred embodiment of the invention is applied to select PEEP over three steps using the crawl PEEP selection strategy, and

[0095] FIG. 12 is a schematic system-chart representing an outline of operations of the method according to the invention.

DETAILED DESCRIPTION OF AN EMBODIMENT

[0096] FIG. 1 shows patient examples for various settings of PEEP and the corresponding changes in physiological parameters. When changes in ventilator settings of PEEP is modified, patient physiological parameters change cause corresponding changes in some gas exchange and lung mechanics parameters as seen in FIG. 1, and mathematical models can no longer accurately predict changes in the these clinical variables. A decision support system using mathematical models for simulation and in calculation of optimal ventilator settings will then not be able to provide correct advice.

[0097] In a preferred embodiment, for the mechanical ventilation system with decision support for PEEP setting, the invention can be implemented as illustrated in FIG. 2. The mechanical ventilation system or respiration of an associated patient 5 is further adapted for providing decision support for mechanical ventilation, the system comprising ventilation means 11 capable of mechanical ventilating said patient with air and/or one or more medical gases, .e.g. oxygen, the ventilation means having a plurality of settings, V_SET, comprising a positive end expiratory pressure (PEEP) setting.

[0098] Additionally, control means 12 are provided, the ventilator means being controllable by said control means by operational connection thereto. The control means may be integrated on a computer system operationally connected to the ventilation means 11 and the measurements 11a, 11b, and/or 11c.

[0099] Measurement means, 11a, and 11b, are arranged for measuring the inspired gas and/or the respiratory feedback of said patient in the expired gas in response to the mechanical ventilation, the measurement means being capable of delivering first data D1 to said control means. The measurement means are shown as separate entities but could alternatively form part of the ventilator means 11 without significantly change the basic principle of the present invention. The measurement means are capable of delivering first data D1 to the control means 12 by appropriate connection, by wire, wirelessly or by other suitably data connection.

[0100] The control means 12 CON is also capable of operating the ventilation means by providing ventilator assistance so that said patient 5 is breathing spontaneously and/or breathing fully controlled by the ventilating means 11. As schematically indicated, a clinician, or other medical personnel, may survey and ultimately control the ventilation system 10. The control means is adapted for using both the first data D1, and second data D2 indicative of the respiratory feedback in the blood provided by measurement means 11c, in one or more physiological models MOD descriptive of, at least, lung mechanics, and/or gas exchange in the lungs of the patient, the physiological models comprising a number of model parameters MOD_P.

[0101] In one embodiment of the invention, second data may particularly be described as data originating from other sources than the mechanical ventilator itself (this could be sensor, blood gases, doctor input etc.).

[0102] In one variant of the invention, the second data D2 could be estimated or guessed values being indicative of the respiratory feedback in the blood of said patient, preferably based on the medical history and/or present condition of the said patient. Thus, values from previously (earlier same day or previous days) could form the basis of such estimated guess for second data D2. In other variants, the second data can be provided from measurement means 11c on a continuous basis from actual measurements.

[0103] The control means is further arranged for simulating the effect on one, or more, model parameters MOD_P of the physiological models for a suggested value of the positive end expiratory pressure (PEEP) setting for the ventilation means, and thereby provide decision support in relation to said suggested PEEP value for the clinician. The suggested value of PEEP can be inputted by the clinician and/or suggested from the decision support part of the system 10 itself.

[0104] The effect on the model parameters MOD_P may be outputted to an appropriate human-machine interface (not shown) for displaying the result, e.g. a computer with a screen therefore. Alternatively or additionally, the model parameters MOD_P output may be communicated to the decision support part of the system 10 for use in connection with mechanical ventilation of patients, optionally for diagnostic purposes.

[0105] FIG. 3 is a schematic drawing of the decision support system 20 for providing decision support to an associated mechanical ventilation system (not shown). PEEP models, cf. FIG. 2, reference 20, describe the expected and/or measured effect of PEEP on physiological model parameters MOD_P, these physiological models parameters may include parameters of gas exchange (such as shunt, V/Q mismatch, compliance), haemodynamics, metabolism etc. These PEEP models 20 then allow use of physiological models to simulate the effect of PEEP on the effect of changes in other mechanical ventilator settings by input to models of for example gas exchange, lung mechanics, blood acid-base, circulation and metabolism. Then the optimal setting of remaining ventilator settings can be found through model simulation of clinical outcome variables and optimisation of settings according to for example a set of clinical preference functions (CPFs), or a rule set. The present invention allows any implementation of model based decision support (the so-called decision support core in FIG. 2) to be applied, the preferred embodiment being a fully model based approach with models of relevant physiological systems and models of clinical preferences. By simulating the effect of PEEP on model parameters MOD_P, indicated as MOD_P_1, MOD_P_2, . . . MOD_P_i, . . . MOD_P_n, and further simulating how changes in model parameters affect optimal values of other ventilator settings in a hierarchical manner, it is possible to simulate for all plausible values of PEEP, indicated as PEEP_1, PEEP_2, . . . PEEP_i, . . . PEEP_n, either iteratively going through all possible values or by a numerical method searching through the PEEP values, thus locating the optimal PEEP value which can be advised to the clinician. The decision support core is therefore applied in several layers, each layer representing one PEEP value (see FIG. 3).

[0106] It may be noted that one PEEP value maybe have more than one model parameter value MOD_P associated to it, and only for illustrative purposes, a one-to-one relation between these two is shown in FIG. 3.

[0107] PEEP models can for example be of linear form with only expected changes in physiological parameters over a certain range of PEEP (see patient example with expected changes in shunt, deltaPO2, deltaPCO2, row 1, FIG. 4). The a priori expectation can also be that parameters remain constant with PEEP (see patient example with expected changes in VO2, VCO2, compliance, row 2, FIG. 4). With prediction of the effect of PEEP on physiological parameters, said parameters can then for a given value of PEEP be used in physiological models to predict changes in clinical outcome variables such that the outcome for a given set of mechanical ventilator setting may be evaluated in relation to clinical preferences. Then it is possible to obtain the optimal balance of competing clinical goals such as normalising patient acid-base status but preventing deleterious high settings of volume and pressures. As an example in data from a mechanically ventilated patient, FIG. 4 row 3 show changes in the optimal level of mechanical ventilator settings (tidal volume, respiratory frequency and inspired oxygen fraction) for different values of PEEP using the models of PEEP effect shown the top two rows. It can be seen that an increase in PEEP leads to reduction in tidal volume (Vt) which is necessary to prevent lung injury from high pressures. This is due to the compliance, which is predicted to stay constant leading to increases in ventilator pressures with PEEP. To counter the reduction in ventilation with tidal volume respiratory frequency (Rf) is increased. Finally, inspired oxygen level (FiO2) is reduced due to the predicted improvements in Shunt and APO2. The fourth row of FIG. 4 shows predicted changes in clinical outcome variables with changes in PEEP. Despite tidal volume is reduced, the clinical outcome variable, plateau pressure (Pplat) increases with large changes in PEEP whilst pH and oxygenation (SaO2 and SmvO2) relatively constant, in a clinical perspective. The last two rows of FIG. 4 show examples of clinical preference functions representing different clinical goals and their predicted change with PEEP, which can be used for providing advice on the optimal level of PEEP.

[0108] The decision support depend on the PEEP selection strategy, as the effect of PEEP is different when set as part of an iterative set of increments in PEEP compared to an iterative set of decrements in PEEP following a recruitment manoeuvre. This is due to the hysteresis of pressure-volume relationship of human lungs. As PEEP is the ventilator maintaining a pressure during expiration during which the alveoli may collapse (de-recruit), PEEP is best set during expiration after a large increase of pressure to maintain open recently recruited alveoli as smaller levels of pressure are necessary to keep these lung units open (hysteresis). However, as recruitment maneuvers can be detrimental in some patients and the identification of whom may benefit from a recruitment manoeuvre is difficult, this approach to setting PEEP is primarily performed by clinical experts. Whilst these experts do perform recruitment maneuvers, little evidence supports an overall successful strategy. Therefore, at most institutions, PEEP changes are performed in small steps without first performing a recruitment maneuver, with the benefit of PEEP here being represented by the recruitment of alveoli due to small increases in pressure during inspiration and the PEEP maintaining these lung units open. As a consequence of the hysteresis of the lung pressure-volume relationship the optimal PEEP found during an incremental (“crawl”) PEEP strategy is not the same as would have been found during a decremental (recruitment) strategy. The present invention therefore includes separate flows for setting PEEP depending on the PEEP selection strategy (see FIG. 6). In a preferred embodiment the system will have two sets of PEEP models representing each PEEP selection strategy, in the case where clinicians apply both strategies in the same patient. The clinician may chose the strategy via one, or more, PEEP selection strategy modules 30, cf. FIG. 2. It may be advantageous if the system can provide advice on the optimal PEEP selection strategy based on the potential for lung recruitment in the patient, which may be assessed from the current values of the model parameters, the expected gain in these values from increases in pressure, and/or the measured changes in model parameters as a response to a change in pressure. For example, if a change in PEEP during a crawl strategy causes a large reduction in shunt indicating recruitment of alveoli, this may indicate large benefit from a recruitment maneuver and a shift in strategy. Also large levels of shunt per se, may indicate possible benefit from a recruitment maneuver. Based on the strategy selected by the clinician and/or the system and calculated decision support, the clinician or the system modifies the PEEP setting and other settings on the ventilator.

[0109] Changes in patient physiological variables are measured by measurement means for assessing ventilation, flow, pressure and volumes, and/or inspired and expired contents of O2 and CO2, and/or blood gas contents (O2 and CO2) and/or changes in acid base status of blood. These measurements are collected and processed. In a preferred embodiment, the data processing evaluates the changes in measurements and delay model parameter estimation until steady state is detected, this allowing the change in PEEP to take effect and avoiding the system from providing new advice when the previous advice has not had an effect.

[0110] When the patient is in steady state or the system has in other ways evaluated that the PEEP has had an effect, the changes in patient physiological parameters can be measured. These may be available from a single point measurement, such as parameters for cardiac output, lung compliance, lung resistance, respiratory drive, anatomical dead space, alveolar deadspace, Effective shunt etc. However, some physiological parameters require the clinician to perturb the patient physiology for accurate measurement. For example, gas exchange parameters (e.g. shunt, low V/Q, high V/Q, dPO2, dPCO2, End expiratory lung volume) can be accurately measured from a stepwise variation in inspired oxygen fraction and measurement of several variables at steady state, cf. WO 2000 45702 and WO 2012069051, which are hereby incorporated by reference in their entirety, these two references describing the so-called automatic lung parameter estimator (‘ALPE’) of the present applicant, without and with carbon dioxide measurements, respectively. An example is shown in FIG. 5a, where downward triangles and x's are measurements at low PEEP and upward triangles and +'s are measurements at a high PEEP value. The fit of a physiological model of gas exchange is shown by solid line for oxygen and o for carbon dioxide (right and plot), and a dashed line for oxygen and a square for carbon dioxide at low and high PEEP, respectively. It can be seen that the patient response to PEEP is an improvement in oxygenation shown by the leftwards shift of the FetO2-SaO2 curve and a worsening in carbon dioxide excretion, as caused by increase in high V/Q lung regions), seen as larger value of arterial partial pressure of carbon dioxide (PaCO2).

[0111] The estimated physiological model parameters are input to the PEEP model learning component where the PEEP models are adapted to the patient response to PEEP. FIG. 5 shows a preferred embodiment for the PEEP learning models for gas exchange model parameters, and their adaptation to a measured response to a PEEP change. The PEEP models can be implemented as linear models. However, as the patient response is unknown a change in PEEP is preferably only assumed within a certain range beyond which model parameters are assumed to stay constant thus avoiding the decision support core to calculate too large changes in response to PEEP. This range can reflect the uncertainty in the assumed response and may be modified according to local clinical preferences and patient populations. When the models are updated so are these ranges reflecting the greater certainty regarding patient response to PEEP. To provide the first suggestion it is necessary to have an a priori expected response to PEEP. This may be patient population specific reflecting the different tendencies for specific patient populations to respond positively to PEEP. The model can be based on data as shown in FIG. 1. In FIG. 5 the a priori assumption is that shunt decreases with PEEP, low V/Q measured as dPO2 remains constant and high V/Q measured as dPCO2 increases. The patient responds somewhat differently which the models are adapted to, and these updated models may then be used by the decision support core to calculate a new optimal value of PEEP.

[0112] In addition to gas exchange, PEEP affects lung mechanics and in support ventilation modes PEEP may also affect metabolism. Lung mechanics are affected differently whether the patient is ventilated in a controlled ventilation mode where the ventilator fully initiates and carries out the breath or in support mode ventilation where the patient initiates the breaths and the ventilator supports the work of breathing. FIG. 6 shows a preferred embodiment of a PEEP learning model for compliance for mechanical ventilation in control mode where it is not necessary to take the patient's own effort into account. In control mode, PEEP would theoretically affect compliance differently whether PEEP is increased or decreased due to the hysteresis of the lung pressure-volume relationship. In a preferred embodiment, the compliance model for control mode ventilation therefore has a steeper slope for simulating a decrease in PEEP reflecting the risk of causing a lung derecruitment where lung units collapse.

[0113] FIG. 7a-b shows a preferred embodiment of a PEEP learning model for compliance for mechanical ventilation in support mode where it is necessary to take the patient's own effort into account. In support mode PEEP affects both lung mechanics and the respiratory work, lung mechanics by recruiting lung units, respiratory work by affecting the length tension relationship of the respiratory muscles thereby increasing/decreasing respiratory work. It is not possible from normally available bedside measurements to separate patient effort from lung mechanics. A preferred embodiment is therefore a combined model of effective lung compliance encompassing changes in patient respiratory work and lung mechanics with PEEP. FIG. 7a shows a model of pressure-volume relationship in pressure support. The preferred embodiment models PEEP as affecting this relationship through the parameter kc shown in FIG. 7b where changes in kc displaces the PS-Vt model. As such if increases in PEEP worsens the length tension relationship of respiratory muscles kc decreases resulting in lower tidal volume for the same pressure support.

[0114] In support mode, PEEP may also affect patient metabolism as too high PEEP levels force the patient into a rapid shallow breathing pattern where respiratory work is significantly increased. To prevent the decision support core from advising on such PEEP levels if this response is observed, a preferred embodiment includes models of patient metabolism increasing several fold at this PEEP level. FIGS. 7c and 7d illustrate preferred embodiments of these metabolism learning models, with a factor which is multiplied with oxygen consumption (VO2) or carbon dioxide production (VCO2) and is 1 for PEEP levels not causing a significant metabolic load.

[0115] FIG. 8 illustrates the algorithm flow for a preferred embodiment of the invention. In a preferred embodiment, the mechanical ventilation system first makes sure that gas exchange model parameters (simple ‘ALPE’ with no carbon dioxide and full ‘ALPE’ with carbon dioxide) are available and that the patient is stable with other ventilator settings set according to the decision support core. The flow is similar for the two possible PEEP strategies, with the most important difference being that learning models are describing the inspiratory limb and expiratory limb of the lung pressure-volume relationship for the crawl and the recruitment strategies, respectively. For each PEEP step, a target optimal PEEP is calculated, and a step in PEEP is then advised this step being the full PEEP step or a substep hereof, this substep being taken so as to minimize risk of adverse effects not known by the learning models at the current PEEP level. After each PEEP step, relevant learning models are updated where after one or several steps in FiO2 are calculated to learn the PEEP step effect on gas exchange and/or lung volumes. In a preferred embodiment the gas exchange models are updated in two steps, first following the initial step in PEEP, with this step giving a rough estimate of the gas exchange model parameters, these values being used for selecting the FiO2 target for subsequent steps.

[0116] FIG. 9 illustrates the timing of two PEEP steps for each PEEP selection strategy according to the flow depicted in FIG. 8. For the crawl strategy, an initial model calibration is performed, followed by a PEEP advice. This advice is then, if accepted by the clinician, set on the ventilator. After some time when the system signal processing indicates steady state the effect of PEEP is learned and a new PEEP advice is calculated which the clinician again shall review and if it is accepted it is set on the ventilator, and so follows until the system advices that the current PEEP is optimal or the clinician decides otherwise. As a difference, in the recruitment strategy, a recruitment manoeuvre is performed by the clinician after model calibration, where after a new maximum PEEP value is the starting point and the effect of the recruitment at this PEEP level is learned before the first PEEP step is advised to the clinician.

[0117] The procedure for measurement of gas exchange parameters shown in FIG. 5a requires 3 or more measurement points for best accuracy which is time-consuming, in particular if it must be performed repeatedly, once at each PEEP level during an optimization of PEEP. A preferred embodiment therefor uses fewer steps in FiO2 to learn changes in gas exchange parameters using both the expected effect and the measured effect using an approach weighing the error for both to learn as much as possible from the measurements and where the range of measurements are limited using the expected effect. FIG. 10 shows a full gas exchange measurement at low PEEP (downwards triangle and x's for SaO2 measurements and downward triangle for PaCO2 measurement, solid line and o for model fit). Dotted line is expected response to PEEP increase with the three * being points selected as expected measurements in a range not measured with the upward triangle and + being the measured response. The dashed line illustrates a possible model fit utilising a weighted error based on fit to both expected and measured response. Any number of measured and/or expected points may be applied using either weighted/non-weighted error. If using few measurements, limitations in ranges of parameters can be used to allow a single solution. These ranges can be set so as to be conservative selecting the ranges that would require the highest levels of ventilator settings according to the decision support core so at so secure that if model parameters are inaccurate error is on the safe side.

[0118] FIG. 11 shows a simulated patient case for a patient in a control ventilator mode where a preferred embodiment of the invention is applied to select PEEP over three steps using the crawl PEEP selection strategy. The top row shows four of the PEEP learning models, one for shunt, low V/Q (dPO2), high V/Q (dPCO2) and lung mechanics (compliance). Solid line is the default expected response at baseline PEEP value, dashed line is the updated learning model after first step, with dash-dot being the final update of the PEEP learning models before the decision support core advices that current PEEP is the optimal level. Second row shows the optimal value of PEEP and other ventilator settings according to patient response to PEEP and the final row shows changes in mathematical clinical preference functions from a decision support core reflecting the optimal ventilator settings. Steps are not necessarily performed at full to the target, with smaller steps allowing the patient response to be learned in a safe manner.

[0119] FIG. 12 is a schematic system-chart representing an out-line of/in detail the operations of the method, or computer program product according to the invention, i.e. a method for operating an mechanical ventilation system 10 for respiration of an associated patient 5, as shown in FIG. 2, the system being adapted for providing decision support for mechanical ventilation, the method comprising: [0120] S1 providing ventilation means 11 capable of mechanical ventilating said patient with air and/or one or more medical gases, the ventilation means having a plurality of settings (V_SET) comprising a positive end expiratory pressure (PEEP) setting, [0121] S2 providing control means 12, the ventilator means being controllable by said control means by operational connection thereto, and [0122] S3 providing measurement means, 11a and 11b, arranged for measuring the inspired gas and/or the respiratory feedback of said patient in the expired gas in response to the mechanical ventilation, the measurement means being capable of delivering first data D1 to said control means,

[0123] wherein the control means is adapted for using both the first data D1 and second data D2 indicative of the respiratory feedback in the blood in physiological models, MOD, descriptive of, at least, lung mechanics, and/or gas exchange in the lungs of the patient, the physiological models comprising a number of model parameters, MOD_P, and

[0124] wherein the control means is further arranged for simulating the effect on one, or more, model parameters, MOD_P, of the physiological models for a suggested value of the positive end expiratory pressure (PEEP) setting for the ventilation means, and thereby provide decision support in relation to said suggested PEEP value.

[0125] The invention can be implemented by means of hardware, software, firmware or any combination of these. The invention or some of the features thereof can also be implemented as software running on one or more data processors and/or digital signal processors.

[0126] The individual elements of an embodiment of the invention may be physically, functionally and logically implemented in any suitable way such as in a single unit, in a plurality of units or as part of separate functional units. The invention may be implemented in a single unit, or be both physically and functionally distributed between different units and processors.

[0127] Although the present invention has been described in connection with the specified embodiments, it should not be construed as being in any way limited to the presented examples. The scope of the present invention is to be interpreted in the light of the accompanying claim set. In the context of the claims, the terms “comprising” or “comprises” do not exclude other possible elements or steps. Also, the mentioning of references such as “a” or “an” etc. should not be construed as excluding a plurality. The use of reference signs in the claims with respect to elements indicated in the figures shall also not be construed as limiting the scope of the invention. Furthermore, individual features mentioned in different claims, may possibly be advantageously combined, and the mentioning of these features in different claims does not exclude that a combination of features is not possible and advantageous.

[0128] In short, the invention relates to a mechanical ventilation system 10 for respiration of a patient 5, the system being adapted for providing decision support for mechanical ventilation. Control means 12 is adapted for using both first data D1 and second data D2, indicative of the respiratory feedback in the blood, in physiological models MOD descriptive of, at least, lung mechanics, and/or gas exchange in the lungs of the patient, the physiological models comprising a number of model parameters MOD_P. The control means is further arranged for simulating the effect on one, or more, model parameters MOD_P of the physiological models for a suggested value of the positive end expiratory pressure (PEEP) setting for the ventilation means, and thereby provide decision support in relation to said suggested PEEP value. The invention is advantageous for providing mathematical based models of changes in physiology in response to changes in ventilator settings of the PEEP thereby allowing mathematical physiological models to predict changes in clinical variables for a given value of PEEP.

GLOSSARY

[0129] Vt Respiratory volume in a single breath, tidal volume

[0130] RR respiratory frequency (RR) or, equivalently, duration of breath (including duration of inspiratory or expiratory phase)

[0131] pHa Arterial blood pH

[0132] PaCO2 Pressure of carbon dioxide in arterial blood,

[0133] SaO2 Oxygen saturation of arterial blood

[0134] PaO2 Pressure of oxygen in arterial blood

REFERENCES

[0135] 1. The Acute Respiratory Distress Syndrome (ARDS) Network (2000) Ventilation with lower tidal volumes as compared with traditional tidal volumes for acute lung injury and the acute respiratory distress syndrome. N Engl. J Med. 342:1301-1308. [0136] 2. Rees SE, Allerød C, Murley D, Zhao Y, Smith B W, Kjaergaard S, Thorgaard p, Andreassen S (2006) Using physiological models and decision theory for selecting appropriate ventilator settings. J Clin Monit Comput. 20:421-429. [0137] 3. Karbing D S, Allerød C, Thomsen L P, Espersen K, Thorgaard P, Andreassen S, Kjaergaard S, Rees S E (2012) Retrospective evaluation of a decision supports system for controlled mechanical ventilation. Med Biol Eng Comput. 50:43-51. [0138] 4. Kwok H F, Linkens D A, Mahfouf M, Mills G H (2004). SIVA: A Hybrid Knowledge-and-Model-Based Advisory System for Intensive Care Ventilators. IEEE TRANSACTIONS ON INFORMATION TECHNOLOGY IN BIOMEDICINE. 8:161-172. [0139] 5. Tehrani D F and Roum J H (2008). Intelligent decision support systems for mechanical ventilation. Artificial Intelligence in Medicine. 44:171-182. [0140] 6. Mélot C. Contribution of multiple inert gas elimination technique to pulmonary medicine—5 (1994). Thorax. 49:1251-1258. [0141] 7. Murley D, Rees S, Rasmussen B, Andreassen S (2005). Decision support of inspired oxygen selection based on Bayesian learning of pulmonary gas exchange parameters. Artificial Intelligence in Medicine. 34:53-63. [0142] 8. Rees S E and Andreassen S. Mathematical models of oxygen and carbon dioxide storage and transport: the acid base chemistry of blood (2005). Crit Rev Biomed Eng. 33:209-264. [0143] 9. Otis A B, Fenn W O, Rahn H. Mechanics of breathing in man (1950). J Appl Phys. 2: 592-607.

[0144] All of the above patent and non-patent literature are hereby incorporated by reference in their entirety.