Page 22 - Control Theory in Biomedical Engineering

10 Control theory in biomedical engineering

Fig. 5 Data-driven modeling approach principle. (The figure has been adopted from
Solomatine, D., See, L.M., Abrahart, R.J., 2008. Data-driven modeling: concepts,
approaches and experiences. In: Practical Hydroinformatics. Springer Berlin Heidelberg,
Berlin, Heidelberg, pp. 17–30.)

constructed using a basic and natural law intimately tied to many other inter-
disciplinary areas, such as biophysics, and biochemistry, and involves elec-
trical and mechanical analogues. Different classes of models exist and can
be described as follows: lumped models vs. distributed parameter models,
continuous-time vs. discrete-rime models, deterministic vs. stochastic
models, parametric vs. nonparametric models and so on. Lumped models
are generally described by ODEs, whereas distributed parameter systems
are described by partial differential equations (PDEs). For example, readers
can find different models of diabetes and blood flow dynamics in the follow-
ing survey papers (Makroglou et al., 2006; Cobelli et al., 2009; Kokalari
et al., 2013) including lumped and distributed parameter models.
A review on recent progress on modeling and characterizing early epidemic
growth patterns from infectious disease including the simplest model SIR is
found in Chowell et al. (2016).
Discrete models are characterized by a countable number of states where
both input and output signals are discrete signals. These types of systems are
easily implementable in physiological applications since experimental real
results are already discrete. In literature, several physiological systems are
described by discrete models such as ECG signals (Huang et al., 2019). Dis-
crete models such as logical, finite state machine or Boolean networks have
also an important impact in modeling process for physiological systems.
A stochastic model represents a set of random variables that can be con-
tinuous or discrete (Kulkarni, 2016). In fact, a stochastic measure contains
uncertainty. Thus, the results of stochastic systems are impossible to predict
even when the dynamics and the initial states are given. These types of