1 of 33

HumMod: A Model of Integrative Physiology

Robert L. Hester, Ph.D.

Professor

Department of Physiology

University of Mississippi Medical Center

rhester@umc.edu

CEO, HC Simulation, LLC

robert@hcsimulation,com

2 of 33

Acknowledgements

Thomas G. Coleman, Ph.D. (1938-2021)
William A Pruett, Ph.D.
John Clemmer, Ph.D.
Thomas E. Lohmeier, Ph.D.

Funding

National Science Foundation (NSF EPS 0903787)
National Institutes of Health (PO1 HL51971)
NSF SBIR

3 of 33

Modeling physiology

The Oxford dictionary defines physiology as:

‘the science of the functions of living organisms and their parts’

Modeling physiology often allows only one variable (A) to change while carefully controlling every other variable in an attempt to understand how (A) affects the system

4 of 33

This is Physiology

Factor

Response

5 of 33

This is Also Physiology

Factor 1

Response

Factor 2

Factor 3

6 of 33

This is Getting Closer to Integrative Physiology

Factor 1

Response

Factor 2

Factor 3

7 of 33

Guyton/Coleman 1972

Guyton AC, Coleman TG, Granger HJ. Circulation: overall regulation

.Annu Rev Physiol. 1972;34:13-46.

8 of 33

~ 10,500 variables and parameters

HumMod

9 of 33

Mathematical Model

HumMod

Deterministic physiology model
>8,000 variables and ~2000 parameters
14 organ systems

left/right heart, lungs, left/right kidney, liver, skin, skeletal and respiratory muscles, GI Tract, bone, adipose, and brain

Endocrine system

Renin-angiotensin-aldosterone system, ADH/vasopressin, atrial natriuretic peptide, insulin/glucagon, cortisol, thyroid hormones, sex hormones, parathyroid hormone, norepinephrine/epinephrine.

Nervous system

Carotid and atrial baroreflex, central and peripheral chemoreceptors
Pre- and post-ganglionic efferent nerves

10 of 33

Medullary Pathways Mediating the Effects of Baroreflex on Sympathetic Activity

CARDIAC

PRESSURES

RENAL

FLUID�EXCRETION

RVLM

BLOOD VOLUME

ARTERIAL

PRESSURE

Medulla Oblongata

RENAL

SYMPATHETIC

NERVE

ACTIVITY

ARTERIAL

BARO-RECEPTORS

11 of 33

Baroreceptor Activation Therapy (BAT)

Programming

System

Implantable

Pulse Generator

Baroreflex

Activation Leads

12 of 33

Medullary Pathways Mediating the Effects of Baroreflex on Sympathetic Activity

CARDIAC

PRESSURES

RENAL

FLUID�EXCRETION

RVLM

BLOOD VOLUME

ARTERIAL

PRESSURE

Medulla Oblongata

RENAL

SYMPATHETIC

NERVE

ACTIVITY

ARTERIAL

BARO-RECEPTORS

13 of 33

Days

mmHg

bpm

% control

Baroreflex Activation

Clemmer et al. AJP Heart 2018

14 of 33

Medullary Pathways Mediating the Effects of Baroreflex on Sympathetic Activity

CARDIAC

PRESSURES

RENAL

FLUID�EXCRETION

RVLM

BLOOD VOLUME

ARTERIAL

PRESSURE

Medulla Oblongata

RENAL

SYMPATHETIC

NERVE

ACTIVITY

ARTERIAL

BARO-RECEPTORS

15 of 33

Baroreflex Activation

Time (days)

Mean Arterial Pressure (mmHg)

Animal Studies

16 of 33

Baroreflex Activation

Time (days)

Mean Arterial Pressure (mmHg)

Animal Studies

17 of 33

bpm

HR

Control

1 wk

2 wk

3 wk

Recovery

Clemmer et al. AJP Heart 2018

Control

1 wk

2 wk

3 wk

Recovery

MAP

BAT Decreases Heart Rate

Control

1 wk

2 wk

3 wk

Recovery

Time (weeks)

ANP

18 of 33

bpm

HR

Control

1 wk

2 wk

3 wk

Recovery

Clemmer et al. AJP Heart 2018

Control

1 wk

2 wk

3 wk

Recovery

MAP

Response When No Decrease in Heart Rate

Control

1 wk

2 wk

3 wk

Recovery

Time (weeks)

ANP

UO

Blood

Pressure

ANP

19 of 33

“In Silico Clinical Trials”

Virtual Population

20 of 33

To test the hypothesis that HumMod can simulate a virtual population which is comparable to an experimental population.

Hypothesis

21 of 33

Methods - HumMod

TGF Effect

Tubuloglomerular Feedback

[Na+] (mmol/L)

TGF Effect

Afferent Arteriolar Conductance

Effect (xNormal)

22 of 33

Parameter Variation

TGF Effect

Tubuloglomerular Feedback

[Na+] (mmol/L)

TGF Effect

Afferent Arteriolar Conductance

Effect (xNormal)

Variation cardiovascular and renal model coefficients was +/- 5% baseline

23 of 33

Baseline BP in 1000 Virtual Hypertensive Patients

Mean Blood Pressure

24 of 33

Mean Blood Pressure

After Treatment

Baseline

25 of 33

ΔBP in 1000 Virtual Hypertensive Patients after Treatment

Change in Mean Blood Pressure

26 of 33

Predictive testing

The goal is to leverage observation about network abnormalities into clinical challenges that can separate responding and nonresponding phenotypes

Challenges:

Water challenge
Epinephrine
Beta blockade
Etc.

27 of 33

Predict response to treatment?

By analyzing the underlying physiology the following predictive tests may predict responders and non-responders.

Low water challenge

250 mL over 5 minutes testing urine clearance in two hours
Responders Urine H₂O AUC = 240±20 ml
Non-Responders Urine H₂O AUC = 115±12 ml

High water challenge

1 L over 5 minutes testing urine clearance at two hours
Responders Urine H₂O AUC = 930±50 ml
Non-Responders Urine H₂O AUC = 815±85 ml

28 of 33

HumMod

We have a large integrative physiological model
For cardiovascular and renal physiology, the model closely mimics real experimental data
HumMod can be used to understand medical treatments
By varying underlying parameters we can create virtual populations
These virtual population closely match experimental data

Current work focuses on parameterizing model to match human clinical data

29 of 33

Limitations

HumMod is limited in several physiological aspects, such as immunology, genetics
Human data is limited, clinical trials may provide more data
The ability to make multiple time dependent measurements in humans is lacking

30 of 33

Thank you for your attention

Any questions?

rhester@umc.edu

31 of 33

32 of 33

Glomerular filtration rate (mL/min)

Plasma renin activity

(GU/mL)

Mean arterial pressure

(mmHg)

Dogs

Virtual dogs n=6092

Calibrated dogs n=60

33 of 33

HumMod

Dogs

Base

BAT

Base

BAT