The living systems (complexity, homeostatic systems) are a special systems of the third type of complexity in natural science and for such systems it is impossible to determine the stationary state in form of dx/dt=0 (deterministic approach) or in the form of invariance of distribution function f(x) for samples acquired in a row of, the any component xi of all vectors of state x=x(t) =(x1,x2,…,xm)T in m‐dimensional phase space of states. At the same time the mixing property doesn’t met (no invariant measures), the autocorrelation functions A(t) don’t tend to zero if t→∞, Lyapunov’s constants can continuously change the sign. Such systems of the third type (complexity) do not meet the condition of Glansdorff – Prigogine’s theorem, i.e. P ‐ the rate of increase of entropy E (P=dE/dt) doesn’t minimized near the point of maximum entropy E (i.e., at point of thermodynamic equilibrium). It is proposed to use the concept of quasi‐attractors to describe the complexity.

Traditional biological science (biophysics, systems analyses of biosystems) stationary mode of biosystems describes according to equation dx/dt=0 for the systems state vector x=x(t)=(x1, x2,…xm)T. But real biosystems demonstrated uninterrupted chaotic dynamics when dx/dt≠0 is always uninterrupted. The authors present two types of approaches to stationary mode investigation for biosystems. The first approach is based on the compartmental-cluster theory and the second approach is based on the theory of chaos-self-organization. The last is more convenient for real biosystems description because there are pragmatic results of its use. The compartmental-cluster approach may be used for real complex biosystems and the authors present some typical examples of such theory. The stationary mode of hierarchical neural networks were illustrated according to specific audi - analyzator. It was demonstrated that short intervals of tremogram demonstrate the real difference of distribution function parameters. As a result of such experiments – the classical statistics methods don’t usefulness for investigation of postural tremor. The tremogram, cardiogram, encephalogram are the systems of third type. The main idea consists of uninterrupted chaotic movements (glimmering property) of system’s vector in phase space of state and evolution of such system’s state vector in phase space of state. The glimmering property and evolution don’t have properties which can be modeled by traditional deterministic and stochastic approaches.