Heart rate variability: Knowing more about HRV analysis and fatigue in transport studies

The use of ECG signal and derived HRV (Heart Rate Variability) analysis is a well-known technique for detecting different levels of fatigue for objective evaluation in human activities (e.g. car-driver state monitoring). This work takes a step further in detecting the first signals of fatigue without any hard methods (usually car-drivers are forced not to rest for many hours). So, based on data coming from the ECG signal for 24 experiments and the same number of different car-drivers driving for 3 hours starting in good conditions, some correlations between fatigue and heart physiology has been explored through data mining methods. Finally, one classifier based on a particular entropy evaluation has been used due to its very good behaviour (True-Positives > 75 % and ROC area > 90 %). This work, using not the classifier itself but its behaviour when the parameter known as "blending" ("blending" defines a different "neighbour" concept) is changed, shows how the entropy between the computed "five minutes" driving windows (each window is defined by a group of 15 previously selected variables) is more independent of the neighbour when these time-windows are near to two hours driving. The work concludes that the entropy is more stable when drivers reach two hours driving and this way will be promising. Consequently, it is proposed further studies in the future based on this entropy concept too, but now integrating additional factors, e.g. age and circadian cycles, which can complete and improve the HRV analysis, including different scenarios or applications out of the safety in the transport studies.