Expert System for Sleepiness Monitoring

Introduction: The purpose of the monitoring system is to determine the level of human sleepiness without interfering with the performance of actions, give recommendations for further action so that prevent falling asleep. The secondary objectives of the study are to orientate in the development of sleepiness, optimize the set of measured parameters, develop expert decision rules in a situation where most parameters are absent numerical value, but are verbal assessments given by experts.
Materials and methods: To achieve the goals, the monitoring system (MS) is designed as an expert a system that works with decision trees, production laws.
For understanding the development (pathogenesis) of sleepiness and the parameters to be measured the topological method of functioning systems is used to select an optimized set modeling (TM). Subjective information is an input evaluated by expert ratings on point scales without a numerical equivalent. Decision making applies fuzzy logic and linguistic variable theory. Anticipatory alarm is realized by EEG signals.
Results: The MS “Non-intrusive human fatigue assessment” of sleepiness has been developed. The MS consists of 3 blocks: sleepiness determination, recommendations, anticipatory alarm. Blocks evaluate the objective and subjective information using the fuzzy set and the linguistic variable theory. Anticipatory alarm block that prevents a person from falling asleep, receives an EEG signal from 1 electrode in the forehead area. A single electrode EEG lead provides monitoring application of the system in practice. Realized MS interface system for the user (operator or driver) in the form of a mobile Application, where the user is supposed to enter questionnaire data, interactively perform fatigue detection test activities and visualize fatigue assessment, recommendations and alarm warnings. On the other hand, the implemented Expert system visualization interface is intended for direct interaction with each of the modules of the ES complex.
Conclusions: The application of topological modeling is justified by choosing a set of parameters optimized for sleepiness by their informativeness and detection options. Theory of expert systems with decision trees and product laws allow the level of sleepiness to be determined. Linguistic variables theory allows the evaluation of subjective information that does not contain numeric assessments. Adaptation of the system to the field of use is discussed by creating specific recommendations for car drivers and for the work of operators. Recommendations for reducing mental fatigue and alarming state assessments during monitoring and for a pre-flight survey tailored to operators or performing risky work are separately provided. On the other hand, monitoring for the detection of drowsiness has been implemented for the use of drivers, which is based on the data of wearable sensors with the possibility of providing a preventive alarm signal.