Biomedical solutions to problems in intensive care model based therapeutics: Adding quality but not cost to care

(1)

Biomedical Solutions to Problems in

Intensive Care

Model-Based Therapeutics: Adding Quality but not Cost to Care Measuring the Un-Measurable to Protocolise and Improve Care

Patient-Specific “One Method Fits All” Care Decentralizing Patient Care to the Bedside

(2)

Why bother? … Economics 101

• Health care grows by about 0.24% of GDP per year

– Over the last 20 years that’s ~5% (more) of GDP (NZ$7B and A$70B-ish, more) – Imagine what a “free” 5% of GDP would be worth to govts these days!

• Critical care is ~10% of all health care costs, which are in turn (currently)

~10% of GDP

• Critical care has several difficult problems reducing cost and improving /

protocolising care in several core areas despite obvious potential improvements in outcome if they could be sussed

– Mechanical Ventilation (MV), CVS diagnosis and treatment, Glycemic control to name 3 breads + butter

• The current growth of costs, in part demographic, is not sustainable

• Expectations are also rising faster than our ability provide the expected care

(3)

The inevitable why and what to do ??

•

Why

? Productivity improvements driven by technology solutions that

have occurred in many other areas haven’t reached medicine

– Or education for that matter! So, I straddle 2 unproductive sectors!

•

The Difficulty(ies)

?

– Improving productivity is easy, just reduce care and spread resources over more patients. This already occurs to an extent if you look at patient-nurse ratios in ICUs in the US and EU

– There is an inevitable increase in demand to “do more” often “with less” that is not sustainable or really possible without giving something useful up

– Increasing protocolisation helps, but typically provides a “one size fits all”

evidenced based approach that cannot necessarily improve care for everyone and thus doesn’t meet increasing expectations.

– Patient-specific care could improve things within a “one method fits all”

approach, but we already heard that there aren’t enough resources to spend the time to customise care for each patient individually.

• Some say we wouldn’t necessarily know how anyway!

(4)

A vision of the future?

(5)

The lack of technology itself isn't an issue!

Ventilators

A HUGE number of sensors

Infusion pumps: Deliver insulin and other medications to IV lines

Computers

(6)

Interestingly, no one really notices it all…

(7)

(8)

What’s missing? Technology is not well tied

to clinical use and outcome!

A HUGE number of sensors

Infusion pumps: Deliver insulin and other medications to IV lines

Computers Ventilators

(9)

The real problem

• 1: A wealth of numerical data that don’t necessarily have direct clinical

meaning or do not provide a “clear physiological picture”

– The numbers change moment to moment

– They require a “mental model” to sort into a picture of what is happening – Clinical staff are not trained to think about numbers like the engineers who

designed the equipment and thus much information is essentially lost

– All this creates an aura of confusion/uncertainty that suppresses critical thinking – Simplification is needed so clinical caregivers can “rule the technology” to

improve outcomes.

• 2: ICU patients are difficult to manage because are highly variable

– in care

– in response to care

• Yielding 3: The greater the variability arising from either the patient or the interpretation of the data…

– the more difficult the patient’s management – the more variable the care

– the more likely a lesser outcome

(10)

What about Protocolised Care?

•

Goal:

To reduce the

iatrogenic

component due to variability in care

•

BUT

applicable to groups with well-known clinical pathways

• “One size fits all approach”

• Reduces variability in how care is given,

but

…

• Not all patients are the same so it cannot take into account inter-

and intra- patient variability in response to care!

• What is needed is a patient-specific “One

method

fits all approach”

(11)

Less is more: 2 Kinds of Variability

Model-based methods

can provide

patient-specific

care

(12)

Summary of the Problem or

The end of the beginning!

•

Goals

:

– Break cycle of low productivity growth

– Increase productivity significantly without simply working harder or doing less for each patient

•

Will require

: doing patient-care much differently, but, in the absence of

the “cures all” drug, with the same technology tools to hand

• This is actually a huge ask and requires something more revolutionary

and disruptive than evolutionary

– Yet, in medicine “evolution” is the preferred route of change for many good historical reasons

•

So, how to “evolve in a revolutionary fashion”?

(13)

Engineering-based solutions?

•

When in doubt, apply manly force".

(The 1st Rule of Mechanical

Engineering; 1996; a ”colleague”)

•

To heal something that doesn’t work or that makes too much

noise, it is necessary and enough to hit on it with something

that works better or that is noisier".

(Shadoks Logic, 1968;

Jacques Rouxel and René Borg)

•

Apply

finesse

to create patient-specific solutions

– Or ‘Age and craft beat youth and speed every time’ (Unknown, a long long time ago)

(14)

What can an Engineer do about it?

Navier-Stokes equations:

Computational fluids analysis

Computational fluids analysis Mechanical stress analysis_{Mechanical stress analysis}

Building structural analysis Building structural analysis

Finite-element equations, Newton’s laws of motion:

Engineering

analysis is

used in many

different

applications

Rocket and satellite motion Rocket and satellite motion

(15)

...each

application area

is described by a

set of equations

representing the

physical world...

Navier-Stokes equations:

Rocket and satellite motion Rocket and satellite motion

(16)

What can an Engineer do about it?

Navier-Stokes equations:

Thermodynamics Thermodynamics

These systems

of equations are

often analysed

on computer to

help design and

(17)

Navier-Stokes equations:

Thermodynamics Thermodynamics

...and results are

used to make

safer and more

(18)

Model-based Therapeutics (MBT)?

What we do in

model-based therapeutics is

(19)

First, we describe

the physical

(20)

Next, we build up a

mathematical

(21)

Finally, we use computational

analysis to solve these

equations to help us design

and implement new, safer

(22)

Doctors clinical experience and intuition

Where does this model go?

 Insulin

 Glucose

 Sedation

 Steroids and vaso-pressors

 Inotropes

 And many many more …

• Glucose levels • Cardiac output • Blood pressures • SPO2 / FiO2 • HR and ECG

• And many more…

 Insulin Sensitivity

 Sepsis detection

 Circulation resistance

 A better picture of the patient-specific physiology in real-time at the bedside

 Optimise glucose

control

 Manage ventilation

 Diagnose and treat

CVS disease

 And many other

(23)

Where does this model go?

 Insulin

 Glucose

 Sedation

 Steroids and vaso-pressors

 Inotropes

 And many many more …

• Glucose levels • Cardiac output • Blood pressures • SPO2 / FiO2 • HR and ECG

• And many more…

Physiological Models

And Algorithms

 Insulin Sensitivity

 Sepsis detection

 Circulation resistance

 A clear picture of the patient-specific

physiology in real-time at the bedside

 Optimise glucose

control

 Manage ventilation

 Diagnose and treat

CVS disease

 And many other

(24)

What we do with these models in Chch and beyond

BG: Metabolism

CVS: Heart and Circulation

MV: Pulmonary Mechanics

P.vc V.vc Q .s ys P.ao V.ao E.ao R .s ys L.av R.av Q.av Q.tc E.vc L.tc

P.ra R.tc _P.pa

V.pa P.rv V.rv P.lv V.lv E.lv L.mt R.mt Q.mt Q.pv E.pu Q .p ul P.la P.pu V.pu R .p ul L.pv E.rv R.pv E.pa

P.th Thoracic Cavity E.pcd

(25)

Clear Physiological Picture?

• We can measure from clinical data:

– Lung Elastance: is added PEEP stretching the lung or recruiting more volume?

– Lung Volume: is added PEEP recruiting more volume? Enough?

– How have these things changed over time?

• What we get:

– Patient status

– Monitored over time (what’s changing? Getting better?)

– Response to therapy

(26)

Clear Physiological Picture?

• “Not your father’s 1/compliance!”

• A dynamic measure of “system elastance” in response to pressure and flow patterns (separated from resistance)

– Captures COPD for example as seen by suddenly decreasing

elastance as trapped volume is opened to inflowing gases – which is effectively an auto-PEEP

– A dynamic measure that is patient-specific

– It is not a super-syringe or tissue (ex vivo) equivalent!

– Can differentiate ARDS and COPD, as well as changes in resistance

(R) due to tube blockage as all are seen dynamically in different ways in the PV data

(27)

Clear Physiological Picture?

• All at high resolution so we can clearly see changes over

time as conditions change and patient variability rears its head to change things

• None can be measured now with the same resolution

• A direct measurement of something you can titrate to (as

the model makes it visible) since it reflects recruitment vs

resistance vs overstretch directly for that patient.

(28)

– Pulmonary and System resistances that change for sepsis (Rsys) and pulmonary embolism (Rpu)

– Changes in SV (from pressure only measurements, and no cheap surrogate!) in response to inotropes

– Patient status monitored over time (what’s changing?)

• More “Un-Measurable” values that can be used to better

diagnose and guide treatment of CVS dysfunction

Clear Physiological Picture?

P.ra R.tc _P.pa

(29)

Clear Physiological Picture?

– Real-time insulin sensitivity (SI) in response to glucose and insulin administration

– SI changes with patient condition (e.g. sepsis) and over time sometimes quite dramatically (e.g. onset of atrial fibrillation)

– Ability to forecast changes in SI so we can dose to account for future variability and reduce hypoglcyemia.

– Patient status monitored over time (what’s changing?)

– Far less hypoglycemia, optimised care and improved outcomes

• SI is our un-measurable quantity, and is the dynamics

system balance that guides response to care

(30)

Un-Measurables?

• Many clinical decisions are partly blind as they can only measure surrogates of

the disease state

– Thus, they rely on clinical staff intuition and experience more than “firm data” – Outcome is variability and reduced quality of care in a more hectic world

• Models offer a clear physiological picture that makes diagnosis, treatment and

evaluation of response far clearer, and thus less variable

– Available to everyone from the Sr Specialist to Junior Nurse

– Clear pictures = easy diagnosis and treatment decisions with no 2nd_guesses

– Made visible by models and data  patient-specific models (and time specific!)

• They do this in a patient-specific fashion by linking patient-specific data from all

(31)

Un-measurables and Endpoints

• Importantly, chosen well, these metrics are direct markers of health and response related to core ICU therapies, and can thus be used to protocolise using specific values to create and guide patient-specific care

– i.e. One method fits all (since patient-specific implies different “sizes”)

•

These are patient-specific treatment metrics that allow more

complete insight into patient state than directly measured

endpoints

(32)

Short Case Examples in MBT

(33)

(34)

A wish list

• If I add PEEP will I stretch the lung more or recruit more lung units?

• What extra volume can I recruit with a change in PEEP?

• Did my recruitment maneuver work? How well, exactly?

• Is patient condition changing?

• Does PEEP need to be changed?

• Broadly, the answers are obvious, yet patient-specific variability

over time and different interpretations or mental models to evaluate

that data means that significant uncertainty creeps into each

decision.

(35)

• Elastance = 1/Compliance • Falling elastance as pressure

rises implies you recruit volume faster than pressure rises == good!

• Minimal Elastance (Maximum Compliance) was observed at PEEP 15cmH2O

• The inflection line is identified as +5~10 % above minimal Elastance.

• Measured by model and PV

data from the vent, it is far more accurate than any estimate or inflection point approximation

Example – Variable PEEP with Average Respiratory System Elastance

(36)

Examples – Variable PEEP with Average Respiratory

System Elastance (all were at PEEP = 10 cmH2O)

Pt 2: (Trauma)

Minimal Elastance PEEP = 15cmH2O

Inflection PEEP = 6~9cmH2O

Pt 8:

(Aspiration)

Minimal Elastance PEEP = 25cmH2O

Inflection PEEP = 12~18cmH2O

Pt 6:

(Intra-abdominal sepsis, CHF)

Inflection PEEP = 7.5~10cmH2O

Pt 10:

(Legionnaires, COPD)

(37)

• Dynamically over a breath at every pressure point = Edrs = dynamic elastance

• Identifies change of Respiratory Elastance within a breathing cycle

• Falling Edrs indicates volume rises

faster than pressure = Recruitment

• Rising Edrs indicates Overstretch

more than recruitment

• Flat Edrs (at minimum) would

thus be theoretically ideal

• Can be monitored every breath

• Edrs potentially provides higher resolution in monitoring and more detailed information where a

constant value cannot

Example – Variable PEEP with Dynamic

Respiratory System Elastance

Edrs drops = recruiting

Edrs rises = stretching not recruiting

Best PEEP thus between 5-10 cmH2O

(38)

Examples – Variable PEEP with Dynamic Respiratory

System Elastance (all were at PEEP = 10 cmH2O)

Pt 2: (Trauma)

Pt 8: (Aspiration)

Pt 6: (Intra-abdominal sepsis, CHF)

Pt 10:

(39)

Some other answers …

• Clear ability to monitor

patient outcome and

response to therapy

Elastance increase

El

as

ta

n

ce

(c

m

H2

O

/L

)

P

EE

P

(

cm

H2

O

)

(40)

Some other answers … volume response to PEEP

• Clear ability to monitor patient outcome and response to therapy

dFRC volume rises 150mL over 0.9 hours

dFRC volume constant over 0.8-0.9 hours

(41)

Potential Clinical Use and Outcome?

• A clear physiological picture can help guide therapy by adding more and better information that is not normally available

• Can we guide PEEP and MV based on Edrs or Elung profiles/values to get beter clinical outcomes (LoMV or number of desaturation events)?

(42)

(43)

A wish list

• What will happen if I add more insulin?

• What is the hypoglycemia risk for this insulin dose?

– Over time?

– When should I measure next to be sure?

• How good is my control? Does it need to be better?

• Should I change nutrition? What happens if someone else has

changed it? How should I then change my insulin dose?

– Many if not all protocols are “carbohydrate blind” and thus BG is a

very poor surrogate of response to insulin

(44)

Standard infuser equipment adjusted by nurses

Patient management Measured data

Feedback control

“Nurse-in-the-loop” system. Standard ICU equipment and/or low-cost commodity hardware.

Decision Support System

I en L I ex I K I L G c I G b G I G V G u x V t u t Q t I n t I n t I t I n I t Q t Q n t Q t I n Q V t PN CNS EGP P P d t Q t Q t G S t G p G ) ( ) 1 ( ) ( )) ( ) ( ( ) ( ) ( 1 ) ( ) ( 1 ) ( )) ( ) ( ( ) ( ) , min( ) ( 1 ) ( ) ( ) ( . . max 2 2 .                        

Identify and utilise “immeasurable” patient parameters For insulin sensitivity

(45)

ICU bed setup

Nutrition pumps: Feed patient through nasogastric tube, IV routes or meals Glucometers:

Measure blood sugar levels

Infusion pumps: Deliver insulin and other medications to IV lines. Sub-cut insulins may also be used.

(46)

Blood Glucose levels

Blood Glucose levels

Variability, not physiology or medicine…

Controller Controller

Fixed dosing systems

Typical care

Fixed dosing systems

Typical care

Adaptive control

Engineering approach

Adaptive control

Engineering approach

Variability flows through to BG control

Variability flows through to

BG control Variability stopped at _controller

Variability stopped at controller

Models offer the opportunity to identify, diagnose and manage variability directly, to guaranteed risk levels. Models offer the opportunity to identify, diagnose and

manage variability directly, to guaranteed risk levels.

Fixed protocol treats everyone much the same

Controller identifies and manages patient-specific

variability

Controller identifies and manages patient-specific

variability Patient

response to insulin Patient response to

(47)

Models, Variability and Risk

BG [mg/dL] Time 4.4 6.5 Insulin sensitivity Blood glucose tnow

Stochastic model shows the bounds (5th_{– 95}th_percentile)

for insulin sensitivity variation over next 1-3 hours from the initially identified level

For a given feed+insulin intevention an output BG distribution can be forecast using the model

tnow+(1-3)hr

95th 75th 50th 25th 5th 5th 25th 50th 75th 95th

5th_{, 25}th_{, 50}th _{(median), 75}th_{, 95}th

percentile bounds for SI(t) variation

based on current value

Insulin sensitivity

Blood glucose

tnow

Stochastic model shows the bounds (5th_{– 95}th _percentile)

tnow+(1-3)hr

95th 75th 50th 25th 5th 5th 25th 50th 75th 95th Stochastic model predicts SI Stochastic model predicts SI

Forecast BG percentile bounds:

A predicted patient response!

A predicted patient response!

SI percentile bounds

+

known insulin

+

system model

= ...

+

known insulin

+

system model

= ...

Iterative process targets this BG forecast to the range we want:

= optimal treatment found! Patient response forecast

(48)

Maximum 5% Risk of BG < 4.4 mmol/L

BG [mg/dL] Time 4.4 6.5 Insulin sensitivity Blood glucose tnow

Stochastic model shows the bounds (5th_{– 95}th_percentile)

tnow+(1-3)hr

95th 75th 50th 25th 5th 5th 25th 50th 75th 95th

5th_{, 25}th_{, 50}th _{(median), 75}th_{, 95}th

percentile bounds for SI(t) variation

based on current value

Insulin sensitivity

Blood glucose

tnow

Stochastic model shows the bounds (5th_{– 95}th _percentile)

tnow+(1-3)hr

95th 75th 50th 25th 5th 5th 25th 50th 75th 95th Stochastic model predicts SI

A predicted patient response!

+

known insulin

+

system model

= ...

= optimal treatment found!

Patient response forecast can be recalculated for

different treatments

Patient response forecast can be recalculated for

(49)

Why this approach?

• Model lets us guarantee and fix risk of hypo- and hyper- glycemia

• Giving insulin (and nutrition) is a lot easier if you know

the range of

what is likely to happen

.

• We know this and dose appropriately

• Allows clinicians to select a target band of desired BG and guarantee

risk of BG above or below

• We tend to fix a 5% risk of BG < 4.4 mmol/L which translates to less

than 1/10,000 (interventions) risk of BG < 2.2 mmol/L (should be about

2% by patient)

(50)

Some Results to Date

• Very tight

• Very safe

• Works over several countries and clinical practice styles

• Also been used in Belgium

• Measuring SI is very handy whether you do it with a model (STAR) or estimated by response

(SPRINT)

STAR Chch STAR Gyula SPRINT Chch SPRINT Gyula

Workload

# BG measurements: 1,486 622 26,646 1088

Measures/day: 13.5 12.8 16.1 16.4

Control performance

BG median [IQR] (mmol/L): _{[5.7 – 6.8]}6.1 _{[5.4 – 6.8]}6.0 _{[5.0 – 6.4]}5.6 _{[5.5 – 7.5]}6.30

% BG in target range 89.4 84.1 86.0 76.4

% BG > 10 mmol/L 2.48 7.7 2.0 2.8

Safety

% BG < 4.0 mmol/L 1.54 4.5 2.89 1.90

% BG < 2.2 mmol/L 0.0 0.16 0.04 0

# patients < 2.2 mmol/L 0 1 (started _hypo) 8 (4%) 0

Clinical interventions

Median insulin (U/hr): 3 2.5 3.0 3.0

(51)

So, because we know the risk …

• We get tight control safely

• We do it by identifying insulin sensitivity (SI) every intervention

– Measuring SI is a direct surrogate of patient response to all aspects of metabolism, and is not available without a (good) model

– Using just BG level is a very poor surrogate because it lacks insulin/nutrition context. Like trying to estimate kidney function from just urine output – it lacks context

• We can minimise interventions, measurements and clinical effort with confidence

and exact knowledge of the risk

• We know what to do when nutrition changes, and can change it directly if we

require!

• So, what’s the target you ask.. (not yet answered for MV case)

– All we know is that level is bad and so is variability with about 1M opinions as to what and how much….

(52)

cTIB = cumulative time in band

A measure of exposure / badness over time

• Measures both level and variability

• We examined 3 “intermediate ranges” that most would think are not at all different!

(53)

cTIB

• 1700 patients from SPRINT and before SPRINT, and both arms (high and low) of Glucontrol trial in 7 EU countries

• Is there a difference between 7 and 8 mmol/L or 3-4 mmol/L of variability???

• Yes, significantly so from day 2-3 onward

• Difference is more stark if you eliminate patients who have at least 1 hypo (BG < 2.2)

• We think the answer is clear and know how to safely achieve those goals

• Because you can calculate it in real time you can use it as an endpoint for a RCT

Day (1-14)

S

u

rv

iv

al

O

d

s

R

at

io

4.0 – 7.0 5.0 – 8.0 4.0 – 8.0

cTIB > 50%

cTIB > 60%

cTIB > 70%

(54)

(55)

Engineering + Medicine = Patient-Specific Care

• The main goal of models and engineering in critical care might readily

be summarised as:

– Turning a wealth of data and technology into a coordinated, predictive and, most importantly, patient-specific picture of the clinical situation by making key patient-specific parameters “visible” to enhance

monitoring and diagnosis, and guide/optimise care

• The

technology is there

what is missing is the “finesse” and elegant

solutions, but, we feel those are coming

– I.e. it’s not about the technology but how it’s used.

• MBT can provide patient-specific “

one method fits all

” care that

improves care, decentralises care to the bedside, and, in doing, reduces

cost and increases productivity

– PS: we didn’t say, but we implement these with cheap tablet computers

(56)

And the salient sign that it’s “right”

• The nurses have not thrown it out the window yet…

• And, in fact, appear to like these solutions …

•

It’s all about better tools to do a better job for patients with less

time, stress, effort, uncertainty or worry…

•

In a world where demand outstrips supply this the most important

(57)

Acknowledgements

Glycemia PG Researchers

Thomas Lotz

Jess Lin Aaron LeCompteJason Wong et al

Hans

Gschwendtner

Lusann

Yang

Amy Blakemore &

Piers Lawrence

Carmen Doran

Kate Moorhead Sheng-Hui Wang

Simone

Scheurle Uli

Goltenbott Normy Razak Chris Pretty

Jackie

Parente

Darren Hewett James Revie

Fatanah Suhaimi Ummu

Jamaludin

Leesa Pfeifer

Harry Chen Sophie Penning

Stephan Schaller

Sam Sah Pri

Brian

Juliussen Ulrike Pielmeier

Klaus

(58)

Math, Stats and Engineering Gurus

Dr Dom Lee

Dr Bob

Broughton Dr _{Paul Docherty}

Prof

Graeme Wake

The Danes

Prof Steen Andreassen

Dunedin

Dr Kirsten

McAuley Prof Jim Mann

Acknowledgements

Glycemia - 1

Geoff Shaw and Geoff Chase

Don’t let this happen to you!

Some guy named Geoff

The Belgians

Dr Thomas Desaive

Dr Jean-Charles Preiser

Hungarians

Dr Balazs Benyo

Belgium: Dr. Fabio Taccone, Dr JL Vincent, Dr P Massion, Dr R Radermecker Hungary: Dr B Fulesdi, Dr Z Benyo, Dr P Soos, Dr I Attila, and 12 others ...

(59)

Acknowledgements

(Neonatal) Glycemia - 2

And Dr Adrienne Lynn and all the clinical staff at Christchurch Women's Hospital, and all the clinical staff Waikato Hospital

Prof Jane Harding Ms Deb Harris RN Dr Phil Weston

(60)

Acknowledgements

Cardiovascular Systems

Dr. Christina Starfinger

Engineers, Math and Docs

Prof Geoff Chase Dr. Chris Hann Dr Geoff Shaw

The Belgians

Dr Thomas Desaive Dr. BernardLambermont Dr Philippe Kolh

David Stevenson

Claire Froissart

James Revie Stefan Heldmann

The Kiwi’s French and Germans

The Danes

Prof Steen

Andreassen Dr Bram SmithDr Bram Smith

Honorary

Danes

(61)

(62)

Acknowledgements

Agitation / Sedation

Dr. Geoff Shaw

Dr. Andrew Rudge

Carmen Doran 2nd_{Lt S. Hunt}

Dr. Franck

Agogue’ Dr. Dominic_Lee

Dr. Christina Starfinger

(63)

eTIME (Eng Tech and Innovation in Medicine) Consortia

(64)

Last but hardly least!

(65)

(66)

(67)

CVS Monitoring

P.ra R.tc _P.pa

(68)

A wish list

• How is the patient responding?

• I added inotropes and the PiCCO shows no real change in CO but

what I really want to know is what is the stroke volume (SV)?

– Did the inotropes increase SV or just HR?

• What is systemic or pulmonary resistance (i.e. is there an emerging

acute dysfunction?)?

• Is patient condition changing?

(69)

Case Study: Post-Mitral Valve Surgery

0 50 100 150 Patient 1 E nd d ia st ol ic v ol um e (m l) 0 0.5 1 1.5 2 V as cu la r r es is ta nc e (m m H g

s/m

l) 0 1 2 3 4 E nd s ys to lic e la st an ce (m m H g/ m l)

0 2 4 6 8 10 12 0 5 10 15 20 P ul m on ar y ve in p re ss ur e( m m H g) Time (hours) Patient 2

0 2 4 6 8 10 12

Time (hours)

Patient 3

0 2 4 6 8 10 12

Time (hours)

Patient 4

0 2 4 6 8 10 12

Time (hours) Average Left ventricle Right ventricle Systemic Pulmonary Left ventricle Right ventricle

0 2 4 6 8 10 12

(70)

Patient 4

• Measured SV and Pao (aortic pressure) from typical sensors

• Decreased left and right ventricle contractility and increased systemic resistance noticed

• Contributed to a decrease in measured stroke volume and increase in measured aortic pressure.

• The combination of these factors caused left ventricle dilation and is symptomatic of patients with decompensated hearts, where an increase in left ventricle afterload after valve

replacement leads to a decline in ejection fraction.

• Overall, a very clear picture emerges of a failure to respond to the surgery and the weakened contractile state of the left

ventricle does not appear to be able to compensate for this apparent increase in afterload and reduced pulmonary pressure as the left ventricle dilates

0 50 100 150 Patient 1 E nd d ia st ol ic v ol um e (m l) 0 0.5 1 1.5 2 V as cu la r r es is ta nc e (m m H g

s/m

0 2 4 6 8 10 12

Time (hours)

Patient 3

0 2 4 6 8 10 12

Time (hours)

Patient 4

0 2 4 6 8 10 12

Time (hours) 0 50 100 150 Patient 1 E nd d ia st ol ic v ol um e (m l) 0 0.5 1 1.5 2 V as cu la r r es is ta nc e (m m H g

s/m

0 2 4 6 8 10 12

Time (hours)

Patient 3

0 2 4 6 8 10 12

Time (hours)

Patient 4

0 2 4 6 8 10 12

(71)

Patient 1

• Measured SV and Pao (aortic pressure) from typical sensors

• In contrast Patient 1 responds well

• Clear differentiation in patient-specific response 0 50 100 150 Patient 1 E nd d ia st ol ic v ol um e (m l) 0 0.5 1 1.5 2 V as cu la r r es is ta nc e (m m H g

s/m

0 2 4 6 8 10 12

Time (hours)

Patient 3

0 2 4 6 8 10 12

Time (hours)

Patient 4

0 2 4 6 8 10 12

Time (hours) 0 50 100 150 Patient 1 E nd d ia st ol ic v ol um e (m l) 0 0.5 1 1.5 2 V as cu la r r es is ta nc e (m m H g

s/m

0 2 4 6 8 10 12

Time (hours)

Patient 3

0 2 4 6 8 10 12

Time (hours)

Patient 4

0 2 4 6 8 10 12

(72)

Another factor at play is “culture”

Built by engineers



rationalism

Managed by doctors



empiricism

Cockpit view A380 Critically ill patient

72