COMPUTER - ASSISTED INFUSION OF MUSCLE RELAXANTS

1
COMPUTER - ASSISTED INFUSION OF MUSCLE RELAXANTS

• Dr Valérie BILLARD

2
NEUROMUSCULAR BLOCKERS (NMB) EXPECTED EFFECTS
Required
larynx abdominal orthopedic eye, neuro...
EFFECT NM blockade
NMB - drug ? -dosage ?
Unexpected - early motor testing - respiratory
failure - light anaesthesia
3
NMB MEASURED EFFECTS
Twitch TOF Tetanos DBS
T1/Tinitial T4/T1 PTC
NMB - drug ? -dosage ?
Muscle (AP, OO)
?
Expected effect
Visual Force transducer Accelerometry EMG
4
NMB Simple closed-loop systems
ANESTHESIOLOGIST
PATIENT
MONITOR

INFUSION DEVICE
CONTROLLER
5
Simple closed loop systems the controller

• Properties
• Output dependent on the control opération
• rapidly achieve a stable control
• protected from electrical interference and noise
• easy to monitor and to operate
• Principle based upon the error (e measured -
  target)
• Proportional Rate K . Weight . e
• Proportional Integral Rate Kp.weight.e
  Ki.weight.(SeP)
• Proportional Integral Derivative Rate
  K1.eK2.SeK3.de/dt

6
Fuzzy logic control

• Control accepting qualitative data as
  small,big...
• Input error E and change in the error
• Output controller or change in the controller
• Ex. IF error 0 and change in error is positive
  small, THEN output is negative small .

7
Closed loop systems the performances
8
FROM THE DOSE TO THE EFFECT PK -PD RELATIONSHIP
EFFECT (predicted)
NMB DOSE
PK
PD
CONCENTRATION (effect site)
CONCENTRATION (plasma)
Ke0
9
TARGET THE  MEASURABLE  PREDICTED EFFECT
USING PKPD
NMB DOSE
EFFECT (predicted)
PK
PD
CONCENTRATION (effect site)
CONCENTRATION (plasma)
Ke0
10
PK -PD RELATIONSHIP PERFORMANCES
EFFECT (measured)
ERROR
EFFECT (predicted)
NMB DOSE
PK
PD
CONCENTRATION (effect site)
CONCENTRATION (plasma)
Ke0
11
PKPD MODEL ERROR ON THE PK

• Wrong drug (rare!)
• Wrong model (elimination from central
  compartment)
• PK parameters not adjusted to the current patient
• Age
• elderly (CL1 æ Vdss ä)
• infants (CL1 and Vdss ä)
• Obesity (ideal weight vs. real weight)
• Renal or liver failure
• Variability

12
PKPD MODEL ERROR ON THE PD

• PD model inadapted
• Emax vs. others ?
• other muscle or measure than in the model
• PD parameters not adjusted to the patient
• Age EC50 lower in infants
• Burning
• Interactions (volatile )
• Wrong Ke0 hypothermia, age
• Variability

13
HOW TO DECREASE THE ERROR?

• Adjust the PK and PD model to covariates
• Clinical research and publications
• Library of models
• Enter a measured value to adjust the model
  Bayesian forecasting
• Take globally account of the patient covariates
• Could change over time

14
FROM THE DOSE TO THE EFFECT PK -PD RELATIONSHIP
EFFECT (measured)
EFFECT (predicted)
NMB DOSE
PD
PK
CONCENTRATION (effect site)
CONCENTRATION (plasma)
Ke0
15
Bayesian approach

• Comes from Bayes description of conditional
  probability
• Combines
• the amount of information given by a population
  model
• with 1 or few pieces of information coming from a
  patient
• to improve the accuracy of the model to describe
  this patient
• Has been used mainly by adding a measured
  concentration to PK model and applied to
  antibiotics, lidocaine, theophylline,
  antineaplasic agents,...

16
Bayesian adaptation using Stanpump

• Available for atracurium, vecuronium, rocuronium
• Only for target blockade less than 95
• Adjust the PK model to a measured value of effect
• This value is entered manually (open loop)
• Then adjust the target in order to have minimal
  change

17
CONCLUSION

• The effects of muscle relaxants could be measured
  
• This measured effect can
• act as input in closed loop system where output
  is dose
• become a target for CCI based on PKPD model
• be compared to the target to adapt the model to
  the patient
• PK model mainly interindividual variability
• PD model mainly intraindividual variability
• The relevant clinical effects corresponding to
  these measures remain to be known