The article deals with the problems of improving modern human-machine interaction systems. Such systems are called biocybernetic systems. It is shown that a significant increase in their efficiency can be achieved by stabilising their work according to the automation control theory. An analysis of the structural schemes of the systems showed that one of the most significantly influencing factors in these systems is a poor “digitization” of the human condition. “Digitization” here is the identification of a person as a participant in the interaction with a cybernetic or cyber-physical system. The main problem of a biocybernetic system construction is the non-stationarity of such human characteristics as time of the reaction to external disturbances, physical or nervous fatigue, the ability to perform the required amount of work, etc. At the same time, as a rule, there is no objective assessment of this non-stationarity. Under these conditions, ensuring the controllability and efficiency of biocybernetic systems is a very difficult task. It is proposed to solve this problem with the help of electrocardiogram signals: the most accessible and accurate information about a human’s current state. Herein, several examples of such solutions and the results of theoretical studies and experiments are discussed.