Modelling and simulation of complex human physiology systems Tom - - PowerPoint PPT Presentation

Modelling and simulation of complex human physiology systems

Tomáš Kulhánek

Institute of Pathological Physiology, First Faculty of Medicine, Charles Univerzity in Prague, Czech Republic CESNET z.s.p.o. (NGI-CZ) EGI Champion

EGI CF13, Manchester, 10 April 2013

Physiology

• Physiology is a branch of biology which studies all

functions of the living body, including mechanical, physical and biochemical.

• Integrative physiology (excercise science) study

how human body responds to external and internal stimuli

• Computational physiology developes and uses

computational framework to support study of physiological systems

What we do

• Institute of Pathological Physiology

–

Interdisciplinary team (~10 people)- physicians, mathematicians, computer scientists, biomedical engineers, painters/graphical designers, …

–

mathematical modeling of human physiology, Software system for simulation application, Graphical design, Educational portal www.physiome.cz/atlas

Simple Circulation DEMO

Disciplines

 Modeling  Simulation

– computation

 Visualization

Guyton's 1972 diagram corrected

HumMod

Guyton, A. C., Coleman, T. G., & Grander, H. J. (1972). Circulation: Overall regulation. Ann. Rev. Physiol., 41,

• str. 13-41.

Robert Hester, Alison Brown, Leland Husband, Radu Iliescu, William Andrew Pruett, Richard L Summers, and Thomas

• Coleman. Hummod: A modeling environment for

the simulation of integrative human physiology. Frontiers in Physiology, 2(12), 2011. www.hummod.org Kofránek, Jiří, Mateják, Marek, Privitzer, Pavol, Tribula, Martin: HumMod-Golem Edition – Large Scale Model of Physiological Systems for Web Based Medical

• Simulator. In Proceedings of Medicine 2.0 Conference,

Stanford, 2011, Stanford University, 2011 (Lary Chu, ed.). Internet Proceedings www.physiome.cz/hummod

Tools

Simulink - Block diagram environment for multidomain simulation and Model-Based Design. Modelica - is an open standard, object-oriented, declarative, multi-domain modeling language for component-oriented modeling of complex systems. CellML is an open standard based on the XML markup language. to store and exchange computer-based mathematical models. JSim is a Java-based simulation system for building quantitative numeric models and analyzing them with respect to experimental reference data.

Challenges throughout disciplines

 Modeling

– Sustainability

• MATLAB/Simulink – Causal block oriented

modeling, model update and reuse is difficult – update all causalities

• Modelica – acausal object oriented modeling, model

update is simple, causalities solved by tools – Understandable and utilizable for domain experts

• SAPHIR Exemplar project based on Guyton 1972

inspired model instead of newer HumMod → VPH

• IUPS PHYSIOME – uses CellML language, set of

models mainly based on Guyton 1972

Modelica implementation: Guyton 1972 → HumMod 2011

Chalanges throughout disciplines

 Simulation

– Automatize conversion from model to simulator

• MATLAB/Simulink → ActionScript, .NET
• Modelica → C, FMI, .NET

Chalanges throughout disciplines

 Visualization

– Survive technology/platform death

• Native application (MS DOS, MS Windows, … )
• Web based simulators (Adobe Flash, MS Silverlight,

HTML5) – Understandable for student and resercher

• Education
• Research

Chalanges throughout use cases

 Education …

– Simulation takes ~ seconds

 Research

– Hypothesis testing

– Identification of model parameter

• Compute model parameters that can't be measured

– Validation and verification of model

• Mathematical simulation corresponds to clinical and

experimental data – 1 000~1 000 000 instances of simulation, some of them can be parallelized

Computation system - BOINC

Model evaluation – Binaries

• n MS Windows platform

Computation service – SOAP web service BOINC – desktop grid - volunteer computing grid DC-API – SZTAKI desktop grid API based upon BOINC Computation nodes – BOINC clients

client

identification and model parameter estimation

visualization server

simulation

Web service simulation control Data processing REST, HTTP, JSON Database

simulation simulation

... Simulation tasks on dedicated cluster

Identification Algorithm

client

Moving computation to NGI-CZ/EGI

visualization server

simulation

Web service simulation control Data processing REST, HTTP, JSON Database

simulation simulation

...

Identification Algorithm

Curent and future research

Education Virtual Patient ... – LifeScience Boost identification of parameters among partners in pathological physiology ... Identification of dynamics of hematopoietic stem cells engrafted in submyeloablatively irradiated mice … Validation and verification of acidbase balance model against real patients data Analysis of human voice, psychoacoustic tests … (cooperation with Academy of Performing arts) – Computer Science Agent based simulation ... Domain specific language for sustainable simulation application – MetaProgrammingSystem (http://www.jetbrains.com/mps/), Language oriented programming …

Summary

Challenges: understandability, sustainability Modeling – acausal modeling in Modelica Simulation – generated by tools Visualization – web based Research – promote results … Education – virtual pacient ...

Thank you for your attention!

Tomáš Kulhánek tomaton@centrum.cz