Modeling and Simulation in Biomedicine Jos Aarts', Dietmar M6ller2, Rogier van Wijk van Brievingh3
'Dept of Medical Infornatics, Hogeschool Midden Nederland, 3833 AM Leusden, The Netherlands 2Product Group Anaesthesia, Drigerwerk W-2400 Lubeck, Gemny of Electrical Engineering, Delft University of Technology, 2600 GA Delft, The Netherlands 3Faculty Abstract A group ofresearchers and educators in The Netherlands, Germany and Czechoslovakia have developed and adpted mathematical computer models ofphenomena in thefield ofphysiology and biomedicinefor use in higher education. The models are graphical and highly interactive, and are aUl written in TurboPasca1r or the mathematical sbnulation language PSIrm. An educational shell has been developed to lunch the models. The shell allows students to interact with the models and teachers to edit the models, to add new models and to monitor the achievemnts of the students. The models and the shell have been implemented on a MS-DOSrm personal computer. This paper describes thefeatures of the modeling package and presents the modeling and simulation ofthe heart muscle as an
exmple. Introduction The work msulted from afelt need to make available the resultsof mathematical modeling in bimedical rach for ts in biomedicine. Researchers were invited to make their models available for educational use. The invitation grew into a project with the following objectives: - togathermodelswhichwouldcoverawidefieldofbiomedicine; - to select a sndardied programming language in which the models would be implemented allowing an interative use and a graphical presentation; - to develop a shell to control the models; - topmduce atextbook to supportthe useof the models [1]. Twenty models were contibuted. TurboPascal and PSI were chosenasprogramminng ruages.TboPascal hasbeen selected because ofthe familiaity of many researchers with this language. PSI has been chosen as the major nxxeling software because it allows users who are unexperienced withprg todevelop models in a modular fasion with an immediate response to the modification ofparameters or strucure. Nicknamed "BIOPSr', a special version of PSI has been developed for the project The shell not only allows student interation butalso the eaher to access and edit the models and monitor student progess. A textbook has been produced to pport the software [2]. The student is assumed to have a basic knowledge of human physiology. 0195-4210/91/$5.00 C 1992 AMIA, Inc.
900
The educational goals can be described as follows: - Thesentaievesbyexpeimentingabetterunderstandingof th theoy behind the models; - The student is able to simulate experiments which cannt be perfonned on biomedical systems in vitro or in vivo; - Teacher independent leaming is enhanced; - Theadvancedstdetis stimlated todevelop simple modelsof his own based on the examples presewted. The models
The models are structured according to the following format: defaut, exercise, what can you do on your own. The default presents the model with parameters appopriate for an average situaton for example in many models the parameters for healdty pesons are used. The exercises enable the student to deal with an abnormal situation; often the student acts as part of a feedback system to storetheoriginal situation by entering valuesforcertain parameters. '"What can you do on our own" represents the case in which the student can add extensions to existing models or new models on thebasisofquestio raised in thetextbook.The models represent applications in the areas of electrophysiology, circulan, cleance, comparunental analysis and physition, r control ological systems. See Table 1 fora complete listing of the models and the auos. The simulation software
ThemodelshavebeenwritteninTurboPascalandPSI.Pasalisthe well known general purpose p rammng language in which variables mustbe declard TurboPascal comes with an extensive set of tools for gracal presenion. PSI, deveoped at the Delft University of Technology, is a blockoriented program which allows easy solving of (non-)linear, firstorder differential equaions [3]. PSI is highly interactive and easy to use and fits in general the need of naive users. The philosophy ofPSIis thaanysimulonmodelhastobeconstructedfiombasic elements lie a integratr, gain, divider, miltiplier, timer, tablelookup facilites, etc. Each element, caled a block in PSL has one ouW and can have several inputs. As a consequence of the availability ofjust one output the name of a block can reprsent the valueofitsouput.uThestuctureandparametersarestoredintables and the caulation of the required output is achieved by interpre-
tation and performed by calculating all blocks according to some calculation scheme. The solution of the equations describing the models generally follows the standard procedures of numerical integration. Figure 1 shows atypical example of the description of a model in PSI.
I2r-
Table 1: The models and their authors. EI.cfrophysiology
I
RP van Wijk van BrievinWi
Respirabon Respiration regulation The BAIN model for the capnogram
EW Kruyt, A Berkenbodch, RJGM de Zwart (Leiden) JEW Beneken, DPF MOler (Eindhoven, LLbeck) DPF Moller (LLbeck)
Urodynarrics of the lower utinary tract
WA van Duyl (Rotterdam)
Conararlnontal anlaysis
Rtid volumes Pharmacokinetics Optimal experiment design in pharmacokinetics The cerebrospinal fiuid irculation model
II.W F
Blodc
Type
lnput1
Input2
PA RP 1165 1180
CON CON DIV DIV
1160 1175
1145 1145
Parl
Par2
Par3
0.0000
1 .OOOOE+1 0
400 100
5.000E+06
AORTFLOW
1140
1 .OOOE+05 100
339
0.0000 1180
lp
..lI
MEEesrTv
T4.
Change ftn
JChoose
ewf
Pumping and wall mechanics of the left ventricle
PSI has been implemented as well on mainframe and MS-DOS personal computers. A special version of PSI, BIOPSI, has been developed for the project. BIOPSI carries all the features ofPSI but allows only the construction of models consisting of not more than 50 blocks. BIOPSI can be used with thecurrentgraphics standards of the PC.
900
I1
11
Entering a password the teacher can access student files and, modify and edit, if necessary, the models, exercises and help screens and add new models to the shell.
U van Briemen, JHM van Eijndhoven, DA de Jong (Rotterdam)
Weenen, DL Ypey (Delft, Leiden)
INL
man" 2
Figure 2: Pull-down menu of the shell.
DL Ypey, AA Verveen, A van Dujn (Leiden) A van Dujn, DL Ypey, J de Goede (Leiden) RP van Wijk van Brievingh (Delft) RP van W9k van Brievingh, W de Leeuw van
SPL
Pon
(Delft)
FBM Mh (Twente) DPF Moller (LObeck) J Potucek (Prahi)
Pfrysioogical control systems Blood glucose regulaton Regtlation of gastric acbvity Musde contrl Thermoregulation
MEANAOFL
tArm-ma
BJ Jansen, JECM Aarts, MJ kAt (Utrecht, Leusden, Maasticht)
stabiEzation
AORTFLOW
M. - f',-+ r, -F.
KH Wesseing, DPF MolUer (Amsterdam, Lnbedc)
Clearance Renal function and blood pressue
960
f',-,A 4 -,I.
FBM Min (Twente)
J Potucek (Prahi) F Pasveer. RP van Wijk van Brievmngh
GAM
-
(Delft)
The heart as a pump: the carcdovascular system The baroreflex-controled drculation with emphaus on nt baromodulation hypothesis Pumping and wall mecharics of toe left ventride Heart rate regulation during physical boad Catheter-manometer system
GA] GA
VI--4 .-I
RP van Wijk van Brievingh (Delft)
CircL, don
880
T-#--
RP van Wik van Brievingh, IA Garda AIves (Delt)
The excitable membrare: Hodgkin-Huxley model The speakic conduction system of the heart Ebectrodes for bidoolectric signals
930 CT
pull-down menu. Clicking the name of a model gives access to another pull-down menu containing the model in default or the simulations (See figure 2). The results ofthe exercises are stored in a file and can be used for future evaluation by the tewher. The shell has extensive help facilities.
As an example we will treat briefly in this section the implementation ofa model of the wall mechanicsof the left ventricle andhow the student interacts with this model and the simulations. In the model the left ventricular pump (heart muscle) is considered to be a muscular wall enclosing a cavity and the pump function is determined by the parameters heart rate, left atrial pressure, peripheral resistance and contractility. The model is described by four differential equations [4,5]. The model has been implemented in PSI (see figure 1) and consists of more than 100 blocks. The model in default shows the situation in steady state, i.e. the situation of healthy person in rest. The student is asked to give a rough esfimte of the relevant parameters (figure 3). IW":AORWT3IS nM: 28s
9.6902E-05
DI PLEUN
zIONs
LA 281
INI~OFL 0 81
Hit nyj 1ey to coetinue TIhE 8
Figure 1: An example of a model constructed of PSI-blocks. The figure represents a part of the modeling of the wall mechanics of the left ventricle. The educational shell An educational shell has been especially developed for the project. The shell allows the students to access and launch the models and do the simulation exercises. It also allows the teacher to edit the models and track the progress of the students. The shell is menu driven and has an hierarchical structure. After identifying himself the student accesses the model by means of a
901
:1CT:i8.83 662.2821
12839.87 9.675E-5
S
Figure 3: Graphical output of the model in default. The relevant parameters are plotted as a function of time.
~ ~A A 1
The exercises related to the model simulate the situation of running (physical load), hemorrhagic shock and myocardial ischemia. The studentbehaves as part of a feedback system and is asked to restore the mean aortic pressure to its original value of 12 kPa since this guarantees undisturbed perfusion of the various organs. Figures 4, 5 and 6 show studentresponse in the simulation of myocardial ischemia. In myocardial ischemia contrctility decreases and heart rate and left atrial pressure must be adapted to keep the mean aortic pressure at its original level. mm:MMzeIssM allMl r
ZiuiiA 2
zeis 2|1b
Hit aew IsW to ooutiwa
essAN 9091
TINEa
Conclusions
The project has resulted in a software package with models, the educational shell and BIOPSI, accompanied by a textbook. It took four and a half years to complete the project; most work was devoted to the implementation ofthe models in one of the computer languages and the validation of these computer models. In the case of the left ventricle the speed and the stability of the simulation were important issues. Interesting aspects of the project which add to the value of computer simulation in biomedicine are: - The number of models contributed not only encompass classical work but also recent research in biomedicine; - The availability of BIOPSI allows easy implementation of simple models and adaptation of existing models; - The flexibility of the educational shell allows not only for, of course, student access butalso teacher access for monitoring student progress, for editing of the models, exercises and help screens, for modification of student interaction and for the addition of new models. The authors acknowledge the financial support of the Foundation 'Meducatie' for the development of the shell.
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~-I
t
ACT: 6734.68
56601
7479.617
5.6-5
8
Figure 4: In the case of myocardial ischemia the decrease of contractility causes a drop of the mean aortic pressure. Intro
Cdlo
Batlex
0wd Ami
oo 14
Sh
References
looIsohee
Isch HRote
Ish Pt
Coy HIwat
Coy Pires
.E ~t
P-U D la INPUT
Ent_r value betwen 6.3 and 1.6 or for default: 0.6
Fl
n
_E
Choos
ingtheheartrate. Figure5:Thestudentespondstothisby.dat Hit mu km te cautiute fl:omuT3S WZ1suI IEMW3~~ TIE1 NHN:60rnP PZUUA ziVn a UhvAL U
ACT:917.U5
942.581
842.983 6.6371-
5
Flgure6:Ihesystem'sresponsetotheinputofthestudent
902
[1] Aarts JECM, DPF Moller, RP van Wijk van Brievingh. Modeling and simulation in biomedicine on a personal computer. In: Petersen PC, B Onaral (eds.). Proceedings of the twelfth annual international conference ofthe IEEE Engineering in Medicine and Biology Society. New York: IEEE, 1990: 1258-1259. [2] Van Wijk van Brievingh RP, DPF Moller (eds.). Biomedical modeling and simulation on a PC - a workbench for physiology and biomedical engineering. New York: Springer Verlag, in press. [3] Van den Bosch PPJ. Interactive computer aided control system analysis and design. In: Jamshidi M and CJ Herget (eds.). Computer-aided control systems engineering. Amsterdam: North-Holland, 1985:229-242. [4] Arts MGJ, RS Reneman, PC Veenstra. A model of the mechanics of the left ventricle. Ann. Biomed. Eng. 1979;7:299-318. [5] Jansen BJ, JECM Aarts, MGJ Arts. Pumping and wail mechanics of the left ventricle. In: VanWijk van Brievingh RP, DPF Moller (eds.). Biomedical modeling and simulation on aPC - a workbench for physiology and biomedical engineering. New York: Springer Verlag, in press.
'Dept of Medical Infornatics, Hogeschool Midden Nederland, 3833 AM Leusden, The Netherlands 2Product Group Anaesthesia, Drigerwerk W-2400 Lubeck, Gemny of Electrical Engineering, Delft University of Technology, 2600 GA Delft, The Netherlands 3Faculty Abstract A group ofresearchers and educators in The Netherlands, Germany and Czechoslovakia have developed and adpted mathematical computer models ofphenomena in thefield ofphysiology and biomedicinefor use in higher education. The models are graphical and highly interactive, and are aUl written in TurboPasca1r or the mathematical sbnulation language PSIrm. An educational shell has been developed to lunch the models. The shell allows students to interact with the models and teachers to edit the models, to add new models and to monitor the achievemnts of the students. The models and the shell have been implemented on a MS-DOSrm personal computer. This paper describes thefeatures of the modeling package and presents the modeling and simulation ofthe heart muscle as an
exmple. Introduction The work msulted from afelt need to make available the resultsof mathematical modeling in bimedical rach for ts in biomedicine. Researchers were invited to make their models available for educational use. The invitation grew into a project with the following objectives: - togathermodelswhichwouldcoverawidefieldofbiomedicine; - to select a sndardied programming language in which the models would be implemented allowing an interative use and a graphical presentation; - to develop a shell to control the models; - topmduce atextbook to supportthe useof the models [1]. Twenty models were contibuted. TurboPascal and PSI were chosenasprogramminng ruages.TboPascal hasbeen selected because ofthe familiaity of many researchers with this language. PSI has been chosen as the major nxxeling software because it allows users who are unexperienced withprg todevelop models in a modular fasion with an immediate response to the modification ofparameters or strucure. Nicknamed "BIOPSr', a special version of PSI has been developed for the project The shell not only allows student interation butalso the eaher to access and edit the models and monitor student progess. A textbook has been produced to pport the software [2]. The student is assumed to have a basic knowledge of human physiology. 0195-4210/91/$5.00 C 1992 AMIA, Inc.
900
The educational goals can be described as follows: - Thesentaievesbyexpeimentingabetterunderstandingof th theoy behind the models; - The student is able to simulate experiments which cannt be perfonned on biomedical systems in vitro or in vivo; - Teacher independent leaming is enhanced; - Theadvancedstdetis stimlated todevelop simple modelsof his own based on the examples presewted. The models
The models are structured according to the following format: defaut, exercise, what can you do on your own. The default presents the model with parameters appopriate for an average situaton for example in many models the parameters for healdty pesons are used. The exercises enable the student to deal with an abnormal situation; often the student acts as part of a feedback system to storetheoriginal situation by entering valuesforcertain parameters. '"What can you do on our own" represents the case in which the student can add extensions to existing models or new models on thebasisofquestio raised in thetextbook.The models represent applications in the areas of electrophysiology, circulan, cleance, comparunental analysis and physition, r control ological systems. See Table 1 fora complete listing of the models and the auos. The simulation software
ThemodelshavebeenwritteninTurboPascalandPSI.Pasalisthe well known general purpose p rammng language in which variables mustbe declard TurboPascal comes with an extensive set of tools for gracal presenion. PSI, deveoped at the Delft University of Technology, is a blockoriented program which allows easy solving of (non-)linear, firstorder differential equaions [3]. PSI is highly interactive and easy to use and fits in general the need of naive users. The philosophy ofPSIis thaanysimulonmodelhastobeconstructedfiombasic elements lie a integratr, gain, divider, miltiplier, timer, tablelookup facilites, etc. Each element, caled a block in PSL has one ouW and can have several inputs. As a consequence of the availability ofjust one output the name of a block can reprsent the valueofitsouput.uThestuctureandparametersarestoredintables and the caulation of the required output is achieved by interpre-
tation and performed by calculating all blocks according to some calculation scheme. The solution of the equations describing the models generally follows the standard procedures of numerical integration. Figure 1 shows atypical example of the description of a model in PSI.
I2r-
Table 1: The models and their authors. EI.cfrophysiology
I
RP van Wijk van BrievinWi
Respirabon Respiration regulation The BAIN model for the capnogram
EW Kruyt, A Berkenbodch, RJGM de Zwart (Leiden) JEW Beneken, DPF MOler (Eindhoven, LLbeck) DPF Moller (LLbeck)
Urodynarrics of the lower utinary tract
WA van Duyl (Rotterdam)
Conararlnontal anlaysis
Rtid volumes Pharmacokinetics Optimal experiment design in pharmacokinetics The cerebrospinal fiuid irculation model
II.W F
Blodc
Type
lnput1
Input2
PA RP 1165 1180
CON CON DIV DIV
1160 1175
1145 1145
Parl
Par2
Par3
0.0000
1 .OOOOE+1 0
400 100
5.000E+06
AORTFLOW
1140
1 .OOOE+05 100
339
0.0000 1180
lp
..lI
MEEesrTv
T4.
Change ftn
JChoose
ewf
Pumping and wall mechanics of the left ventricle
PSI has been implemented as well on mainframe and MS-DOS personal computers. A special version of PSI, BIOPSI, has been developed for the project. BIOPSI carries all the features ofPSI but allows only the construction of models consisting of not more than 50 blocks. BIOPSI can be used with thecurrentgraphics standards of the PC.
900
I1
11
Entering a password the teacher can access student files and, modify and edit, if necessary, the models, exercises and help screens and add new models to the shell.
U van Briemen, JHM van Eijndhoven, DA de Jong (Rotterdam)
Weenen, DL Ypey (Delft, Leiden)
INL
man" 2
Figure 2: Pull-down menu of the shell.
DL Ypey, AA Verveen, A van Dujn (Leiden) A van Dujn, DL Ypey, J de Goede (Leiden) RP van Wijk van Brievingh (Delft) RP van W9k van Brievingh, W de Leeuw van
SPL
Pon
(Delft)
FBM Mh (Twente) DPF Moller (LObeck) J Potucek (Prahi)
Pfrysioogical control systems Blood glucose regulaton Regtlation of gastric acbvity Musde contrl Thermoregulation
MEANAOFL
tArm-ma
BJ Jansen, JECM Aarts, MJ kAt (Utrecht, Leusden, Maasticht)
stabiEzation
AORTFLOW
M. - f',-+ r, -F.
KH Wesseing, DPF MolUer (Amsterdam, Lnbedc)
Clearance Renal function and blood pressue
960
f',-,A 4 -,I.
FBM Min (Twente)
J Potucek (Prahi) F Pasveer. RP van Wijk van Brievmngh
GAM
-
(Delft)
The heart as a pump: the carcdovascular system The baroreflex-controled drculation with emphaus on nt baromodulation hypothesis Pumping and wall mecharics of toe left ventride Heart rate regulation during physical boad Catheter-manometer system
GA] GA
VI--4 .-I
RP van Wijk van Brievingh (Delft)
CircL, don
880
T-#--
RP van Wik van Brievingh, IA Garda AIves (Delt)
The excitable membrare: Hodgkin-Huxley model The speakic conduction system of the heart Ebectrodes for bidoolectric signals
930 CT
pull-down menu. Clicking the name of a model gives access to another pull-down menu containing the model in default or the simulations (See figure 2). The results ofthe exercises are stored in a file and can be used for future evaluation by the tewher. The shell has extensive help facilities.
As an example we will treat briefly in this section the implementation ofa model of the wall mechanicsof the left ventricle andhow the student interacts with this model and the simulations. In the model the left ventricular pump (heart muscle) is considered to be a muscular wall enclosing a cavity and the pump function is determined by the parameters heart rate, left atrial pressure, peripheral resistance and contractility. The model is described by four differential equations [4,5]. The model has been implemented in PSI (see figure 1) and consists of more than 100 blocks. The model in default shows the situation in steady state, i.e. the situation of healthy person in rest. The student is asked to give a rough esfimte of the relevant parameters (figure 3). IW":AORWT3IS nM: 28s
9.6902E-05
DI PLEUN
zIONs
LA 281
INI~OFL 0 81
Hit nyj 1ey to coetinue TIhE 8
Figure 1: An example of a model constructed of PSI-blocks. The figure represents a part of the modeling of the wall mechanics of the left ventricle. The educational shell An educational shell has been especially developed for the project. The shell allows the students to access and launch the models and do the simulation exercises. It also allows the teacher to edit the models and track the progress of the students. The shell is menu driven and has an hierarchical structure. After identifying himself the student accesses the model by means of a
901
:1CT:i8.83 662.2821
12839.87 9.675E-5
S
Figure 3: Graphical output of the model in default. The relevant parameters are plotted as a function of time.
~ ~A A 1
The exercises related to the model simulate the situation of running (physical load), hemorrhagic shock and myocardial ischemia. The studentbehaves as part of a feedback system and is asked to restore the mean aortic pressure to its original value of 12 kPa since this guarantees undisturbed perfusion of the various organs. Figures 4, 5 and 6 show studentresponse in the simulation of myocardial ischemia. In myocardial ischemia contrctility decreases and heart rate and left atrial pressure must be adapted to keep the mean aortic pressure at its original level. mm:MMzeIssM allMl r
ZiuiiA 2
zeis 2|1b
Hit aew IsW to ooutiwa
essAN 9091
TINEa
Conclusions
The project has resulted in a software package with models, the educational shell and BIOPSI, accompanied by a textbook. It took four and a half years to complete the project; most work was devoted to the implementation ofthe models in one of the computer languages and the validation of these computer models. In the case of the left ventricle the speed and the stability of the simulation were important issues. Interesting aspects of the project which add to the value of computer simulation in biomedicine are: - The number of models contributed not only encompass classical work but also recent research in biomedicine; - The availability of BIOPSI allows easy implementation of simple models and adaptation of existing models; - The flexibility of the educational shell allows not only for, of course, student access butalso teacher access for monitoring student progress, for editing of the models, exercises and help screens, for modification of student interaction and for the addition of new models. The authors acknowledge the financial support of the Foundation 'Meducatie' for the development of the shell.
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~-I
t
ACT: 6734.68
56601
7479.617
5.6-5
8
Figure 4: In the case of myocardial ischemia the decrease of contractility causes a drop of the mean aortic pressure. Intro
Cdlo
Batlex
0wd Ami
oo 14
Sh
References
looIsohee
Isch HRote
Ish Pt
Coy HIwat
Coy Pires
.E ~t
P-U D la INPUT
Ent_r value betwen 6.3 and 1.6 or for default: 0.6
Fl
n
_E
Choos
ingtheheartrate. Figure5:Thestudentespondstothisby.dat Hit mu km te cautiute fl:omuT3S WZ1suI IEMW3~~ TIE1 NHN:60rnP PZUUA ziVn a UhvAL U
ACT:917.U5
942.581
842.983 6.6371-
5
Flgure6:Ihesystem'sresponsetotheinputofthestudent
902
[1] Aarts JECM, DPF Moller, RP van Wijk van Brievingh. Modeling and simulation in biomedicine on a personal computer. In: Petersen PC, B Onaral (eds.). Proceedings of the twelfth annual international conference ofthe IEEE Engineering in Medicine and Biology Society. New York: IEEE, 1990: 1258-1259. [2] Van Wijk van Brievingh RP, DPF Moller (eds.). Biomedical modeling and simulation on a PC - a workbench for physiology and biomedical engineering. New York: Springer Verlag, in press. [3] Van den Bosch PPJ. Interactive computer aided control system analysis and design. In: Jamshidi M and CJ Herget (eds.). Computer-aided control systems engineering. Amsterdam: North-Holland, 1985:229-242. [4] Arts MGJ, RS Reneman, PC Veenstra. A model of the mechanics of the left ventricle. Ann. Biomed. Eng. 1979;7:299-318. [5] Jansen BJ, JECM Aarts, MGJ Arts. Pumping and wail mechanics of the left ventricle. In: VanWijk van Brievingh RP, DPF Moller (eds.). Biomedical modeling and simulation on aPC - a workbench for physiology and biomedical engineering. New York: Springer Verlag, in press.
