H1: Introduction
General
Bio-environmental control
- Influence of climate on micro-climate
- Measurements around humans, animals & plants
-
Modelling of biosystems
-
- Extension from bio-environment (climate) to wide range of measurable variables
General scheme of model based monitoring and control (!)
-
-
1
,A living organism is a CITD-system
The concept of CITD (!)
- Complex, individual, time variant, dynamic
- Important difference between biological & physical systems
- Ex: energy prod in humans & animals (biological) ↔ temp sensors (physical)
Complex
- Ex: energy pathways in human athlete
o Anaerobic a-lactic system
o Anaerobic lactic system
o Aerobic system
Individual
- No 2 organisms are the same
- Ex: 2 cars can be exactly the same, 2 people can’t
Time variant
- Body reacts different on different moments of the day / month / ….
-
o Time instant 2: few hours later than
time instant 1
o Time instant 3: one week later than
time instant 1
- Ex: physiological consequences of training
Dynamic
- Steady state does not exist in biosystem
Methodology
Type of models (!)
Data-based models (black box)
- Simple linear regression, correlation…
- Ex: relationship between body composition & performance variables
- Limitations: outliers, you can’t extrapolate (only use in range data)
Mechanistic models (white box)
- Momentum, heat, ….
- Tell more about system, explain more
- Ex: heat prod in animals, simulation of energy & AZ use in pigs
2
,Data-based mechanistic models (grey box models, DMB)
- In between
- Data has meaning
- Ex: bio-environmental control
- Many processes occur in non-perfect mixed fluids in which environmental variables vary in
time & space
Definitions (!)
- System: part of (virtual) reality that is isolated from environment, always in function of
objective, object that we separate from environment → subject to inputs & measure outputs
- Bio-system: system of which central part is living organism that is influenced (much more
than objects) by its environment, difficult to separate
- Inputs: external signals that influence system & can be manipulated by user
- Outputs: measurable signals that are interesting for user, in function of objective
- Disturbing variables: external signals that influence system & that cannot be manipulated by
user
Ex: monitoring patients at intensive care
-
Ex: formula one racing
-
3
, Ex: monitoring and modelling physical condition of a cyclist
- Physical condition = combination of several aspects of physical state which implies certain
sportive performance
- How can we model condition? We need
o Indicator variable for condition: heart rate
o Variables that influence indicator variable: delivered power, mental status, diet …
- Cyclist = complex system (CITD)
o Not all variables have same influence
o Search for dominant variables which influence system most
o All other variables are considered disturbance variables
o Power = dominant variable that has most effect on heart rate in regard to condition
o → Modelling condition = modelling heart rate response to power
o → Data collection: maximal effort test
o Heart rate response to intermediate power can be approximated with simple linear
model, eventually heart rate flattens out = max heart rate
o Conclusion
▪ Complex systems have dominant variables, relationship between dominant
variables can be simple
▪ Simple modelling techniques can be used
▪ → (Black-box model)
- Cyclist = individual system (CITD)
o Same power input has different heart rate response due to different condition
o Different condition = different linear relationship between heart rate & power
o Model structure is same but slope & intercept
o Conclusion
▪ Need modelling technique that uses same structure
▪ Can alter only parameter values
▪ → Black-box model
- Cyclist = time variant system (CITD)
o 3 zones
▪ Low intensity
▪ Intermediate intensity
▪ High intensity
o Conclusion
▪ Need modelling technique that uses same
structure
▪ Can alter parameter values during cycling
▪ Should be able to model recursively = model calculation needs low amount
of computational power → no mechanistic models possible
▪ → Black-box model
- Cyclist = dynamic system (CITD)
o Part of data with most system dynamics had most accurate estimated parameters
o Conclusion
▪ Watch out during modelling!
▪ Calculate model only is there is enough dynamics
▪ Modelling technique should be flexible
▪ → Black-box model
▪ Big challenge: come to Grey-box model
4
General
Bio-environmental control
- Influence of climate on micro-climate
- Measurements around humans, animals & plants
-
Modelling of biosystems
-
- Extension from bio-environment (climate) to wide range of measurable variables
General scheme of model based monitoring and control (!)
-
-
1
,A living organism is a CITD-system
The concept of CITD (!)
- Complex, individual, time variant, dynamic
- Important difference between biological & physical systems
- Ex: energy prod in humans & animals (biological) ↔ temp sensors (physical)
Complex
- Ex: energy pathways in human athlete
o Anaerobic a-lactic system
o Anaerobic lactic system
o Aerobic system
Individual
- No 2 organisms are the same
- Ex: 2 cars can be exactly the same, 2 people can’t
Time variant
- Body reacts different on different moments of the day / month / ….
-
o Time instant 2: few hours later than
time instant 1
o Time instant 3: one week later than
time instant 1
- Ex: physiological consequences of training
Dynamic
- Steady state does not exist in biosystem
Methodology
Type of models (!)
Data-based models (black box)
- Simple linear regression, correlation…
- Ex: relationship between body composition & performance variables
- Limitations: outliers, you can’t extrapolate (only use in range data)
Mechanistic models (white box)
- Momentum, heat, ….
- Tell more about system, explain more
- Ex: heat prod in animals, simulation of energy & AZ use in pigs
2
,Data-based mechanistic models (grey box models, DMB)
- In between
- Data has meaning
- Ex: bio-environmental control
- Many processes occur in non-perfect mixed fluids in which environmental variables vary in
time & space
Definitions (!)
- System: part of (virtual) reality that is isolated from environment, always in function of
objective, object that we separate from environment → subject to inputs & measure outputs
- Bio-system: system of which central part is living organism that is influenced (much more
than objects) by its environment, difficult to separate
- Inputs: external signals that influence system & can be manipulated by user
- Outputs: measurable signals that are interesting for user, in function of objective
- Disturbing variables: external signals that influence system & that cannot be manipulated by
user
Ex: monitoring patients at intensive care
-
Ex: formula one racing
-
3
, Ex: monitoring and modelling physical condition of a cyclist
- Physical condition = combination of several aspects of physical state which implies certain
sportive performance
- How can we model condition? We need
o Indicator variable for condition: heart rate
o Variables that influence indicator variable: delivered power, mental status, diet …
- Cyclist = complex system (CITD)
o Not all variables have same influence
o Search for dominant variables which influence system most
o All other variables are considered disturbance variables
o Power = dominant variable that has most effect on heart rate in regard to condition
o → Modelling condition = modelling heart rate response to power
o → Data collection: maximal effort test
o Heart rate response to intermediate power can be approximated with simple linear
model, eventually heart rate flattens out = max heart rate
o Conclusion
▪ Complex systems have dominant variables, relationship between dominant
variables can be simple
▪ Simple modelling techniques can be used
▪ → (Black-box model)
- Cyclist = individual system (CITD)
o Same power input has different heart rate response due to different condition
o Different condition = different linear relationship between heart rate & power
o Model structure is same but slope & intercept
o Conclusion
▪ Need modelling technique that uses same structure
▪ Can alter only parameter values
▪ → Black-box model
- Cyclist = time variant system (CITD)
o 3 zones
▪ Low intensity
▪ Intermediate intensity
▪ High intensity
o Conclusion
▪ Need modelling technique that uses same
structure
▪ Can alter parameter values during cycling
▪ Should be able to model recursively = model calculation needs low amount
of computational power → no mechanistic models possible
▪ → Black-box model
- Cyclist = dynamic system (CITD)
o Part of data with most system dynamics had most accurate estimated parameters
o Conclusion
▪ Watch out during modelling!
▪ Calculate model only is there is enough dynamics
▪ Modelling technique should be flexible
▪ → Black-box model
▪ Big challenge: come to Grey-box model
4
