Pharmacolibrary - DPH day 2025

Created by Tomas Kulhanek

pharmacolibrary

#modelica, #pharmacolibrary

Modeling Pharmacology Pharmacolibrary - Free Library to Model Pharmacology

Tomáš Kulhánek¹²³, Filip Ježek²⁴, Jiří Kofránek²³,

Marek Mateják²⁵, Stef Romes¹

¹Flemish Institute of Technology and Research (VITO), Mol, BE

²1st Faculty of Medicine, Charles University, Prague, CZ

³Creative Connections s.r.o, Prague, CZ

⁴University of Michigan, Ann Arbor, USA

⁵Institute of Clinical end Experimental Medicine, Prague, CZ

1st prize award for best library @ Modelica conference 2025

Institute of Pathophysiology

Creative Connections

small spin off company

innovation grants from proof of concept to pilot and pre-commercial validation

contract research

VITO

Flemish Institute for Technological Research

Mol, Belgium

Environmental unit

Digital Precision Health group

RAIDO

PRIBE - personal reference intervals

brave ZR8tgl5Xml

DigiTwin4PH - PGx digital twin

ONCOSCREEN

Pharmacology

Pharmacokinetics (PK) body → drug

Pharmacodynamics (PD) drug → body

Pharmacogenomics (PGx) genetics→drug&body

pharmacokinetics(PK): body → drug

Administration(absorption), Distribution, Metabolism, Elimination(excretion)

standard Modelica language

Commercial tools:

Dymola,

, …

Open-source:

standard Modelica library 4.0.0

custom libraries for specific domains

Physiolibrary 1.0.0 won 1st library award at Modelica conference 2014

Chemical 1.0.0 won 1st library award at Modelica conference 2015

Chemical 2.0

modeling physical chemistry

Physiolibrary 3.0.0

www.physiolibrary.org

Dymola qmDZxxtrfj

Physiomodel

firefox SHzAoizS4q

Complex model of physiology in Modelica and Physiolibrary

https://physiomodel.org

hemodynamics of cardiovascular system

Dymola 9CIA5zGnUy

respiration

Dymola sdXGmJf24o

blood-gas exchange

Dymola qmDZxxtrfj

why Modelica?

model of complex system may lead to implementation errors

A C Guyton, T G Coleman, H J Granger, Circulation: overall regulation, Annu Rev Physiol.1972;34:13-46.doi: 10.1146/annurev.ph.34.030172.000305.

Correction of schema after reimplementation in 2010

J.Kofránek, et al., Restoration of Guyton‘s Diagram for Regulation of the Circulation ...Physiol Res. 2010;59(6):897-908.doi: 10.33549/physiolres.931838.

Guyton's original scheme part of circulatory dynamics

equivalent implementation in component and acausal modeling language

Frank-Starling law is much more visible in component model - "force of the heart muscle depends on pressure of incoming blood".

Readability & Plausability

small models - equation based

VirtualBoxVM 4UK0GmE2mR

medium/complex models- composition of smaller models as components (icons) in diagrams

Dymola 9CIA5zGnUy

connectors

Administration (absorption) - route

parenteral (e.g. intravenous, intraarterial (rare), main parameters: F bioavailability (almost 100%), adminMass, adminDuration):

enteral (e.g. oral, main parameters: F bioavailability, adminMass, Tlag, ka absorption rate)

Administration (absorption kinetics)

instanteous (e.g. intravenous, intraarterial, main parameters: bolus injection adminMass):

Zero-order (e.g. continuous infusion, main parameter: adminDuration):

First-order (rate proportional to the amount of drug remaining to be absorbed, main parameters: adminMass, ka (absorption rate))

Distribution

simplified - ideal mixing, no volume flow rate, compartment (Vd volume of distribution)

$q_m =\frac{dM}{dt}$

$c=\frac{M}{V_d}$

and 1st order kinetics

q_m=Cl_a*c_a-Cl_b*c_b

Distribution - physiology based

compartment - with concentration mixing based on volumetric flow through organs and tissues:

$\frac{dM}{dt} = q_{v_{a}}*actualStream(c_a)+q_{v_{b}}*actualStream(c_b)+q_m$

fixed volumetric flow component:

Metabolism

generally, first pass

active drug → inactive drug (paracetamol)

inactive drug → active drug (aspirin)

active drug → more active drug

make drug more toxic (ethanol)

make drug more water soluble

1st order kinetics between compartments

paracetamol

acetaminophen → sulfate_conjugate

for more complex - use Chemical library (https://www.physiolibrary.org)

Elimination

via kidney or liver using clearance rate Cl

q_m=Cl*c

computed half life

$t_{1/2} = \frac{ln(2) \times V_d}{Cl}$

1-compartment PK model

$c = c_0 \times e^{-Cl.t}$

or derivative form:

$c=\frac{M}{V_d}$

$\frac{dM}{dt} = - Cl \times c$

1-compartment PK model in Modelica (text variant)

1-compartment PK model in component diagram

equations are behind each component

connecting components → equations among variables

compartment

$q_m =\frac{dM}{dt}$

$c=\frac{M}{V_d}$

elimination

q_m=Cl*c

connectors

$qm_1 +qm_2 = 0\$

c_1 = c_2

1-compartment model

parameters for intravenous administration of gentamicin, type of antibiotic against bacterial infections

equivalent 1-compartment model (in equations)

VirtualBoxVM 4UK0GmE2mR

gentamicin - 1 day simulation

gentamicin - 1 day simulation logscale

pk 1c simlog

2-compartment model

parameters for intravenous administration of midazolam, a short-acting benzodiazepine used for sedation, anesthesia, ...

midazolam - 12 h simulation

logscale - 12h simulation

logscale - 12h with peripheral

pk2csimlog

3-compartment model

fentanyl - 10h simulation

fentanyl - 10h simulation logscale

pk3csim

physiology based

compartment takes physiological features into account

cardiac output
lymphatic system flow
detail organ tissue

PBPK model

PBPK simulation

5496 models of drugs by Anatomical Therapeutical Chemical (ATC) classification - knowledge base extracted and combined from Pubmed, Drugbank using LLM

ATC level 1

ATC level 3

ATC level 5

Drug model variants

caffeine

1-compartment model - caffeine

1 cup = 100 mg of caffeine in 8 hours

caffeine in 24 hours

pharmacodynamics (PD): drug → body

What effect has a drug to body

linear effect

effectlin

Emax effect

effectemax

sigmoid Emax effect

effectsigmoid

irreversible effect

time dependent (R response, C concentration, k rate of irreversible effect)

$\frac{dR}{dt}=-k\times C\times R$

with recovery growth factor ks

$\frac{dR}{dt}=k_s \times R - k\times C\times R$

PK PD model Midazolam on EEG activity

midazolam

PK PD simulation

PK PD model digoxin on hemodynamics of cardiovascular system

PKPD HemodynamicsMeurs flatNorm

PK PD simulation

pkpd sim1

pkpd sim2

pkpd sim4

pkpd sim3

pressure volume (PV) diagram

PV shift from right to left → less maximal volume, slight increase in systolic pressure.

Digoxin is a cardiac glycoside derived from the foxglove plant Digitalis lanata. It is primarily used in the treatment of various heart conditions, notably atrial fibrillation, atrial flutter, and sometimes heart failure that cannot be controlled by other medications. Digoxin is approved and widely used in clinical practice today.

Pharmacogenomics (PGx)

What is individual variability of PK and PD based on genotype and phenotype?

genotype

phenotype

SLC22A2

OTC2_phenotype

alleles

*1, *1

808T, *1

clearance

normal

intermediate

CYP2C19

CYP2C19_phenotype

alleles

*1, *1

*2, *1

*17, *1

metabolizer

normal

poor

rapid

PK PGx model

gentamicin

PK PGx simulation (gentamicin)

population PGx model

pgmodelexperiment

population PGx simulation

pkpgx experiment sim

optimal dose of lithium carbonate

brave OZ24t5RrJ2

https://egolem.online/pharma

iron metabolism visualisation

brave 5rh0hlzHNL

eGolem powered by bodylight.js

https://egolem.online/irm/ https://bodylight.physiome.cz

howto find a drug

VirtualBoxVM quF8UzRISk

workflow - 1. export model to FMI unit 2. simulate in Python & fmpy & 3. get data from external data (PK-DB.com) 4. compare/fit/...

VirtualBoxVM TIsoOjFLH2

CA125 concentration based on tumor growth, interactive simulator - change growth factor (logistic function) and CA125 production rate

Conclusion

Pharmacolibrary: https://github.com/creative-connections/Pharmacolibrary

pharmacokinetic: 5496 drugs models - most generated

pharmacodynamics: various effects

pharmacogenomics: various genotype/phenotype influence

current and future research and development: contribute to democratize pharmacology knowledge, use PBPK for quantitative in vitro to in vivo extrapolation (QIVIVE), validate PK models of drugs, add PK of dietary supplements and toxicokinetics of chemicals, cover published PD and common pharmacogenes,

Pharmacolibrary / GITHUB

https://github.com/creative-connections/Pharmacolibrary

acknowledgment

ONCOSCREEN project, funded by the European Union’s Horizon Europe grant No. 101097036.

subaward to R01 HL173346. funded by NIH

co-simulation of multiple models via FMI in static web app with bodylight.js

https://egolem.online/demo/

comprehensibility

Pharmacolibrary 25.06 (beta)

connectors and components:

effects

PGx

40-50 pharmacogenes commercially available in testing panels

majority cytochrome P450 enzymes ( CYP2D6,CYP2C19,...)