Master thesis presentation- The Art of Heart Rate Variability: Driver Fatigue Application

Welcome to a master thesis presentation by Fabio Forcolin: The Art of Heart Rate Variability: Driver Fatigue Application.
Examiner: Bengt Arne Sjöqvist

Time: Tuesday, June 2nd at 10.00
Place: Lunnerummet (room 3311), Hörsalsvägen 11, 3rd floor

Abstract
A substantial percentage of accidents are caused by drivers falling asleep at wheel. Among the different ways of detecting sleepiness at wheel, physiological measurements start changing at an earlier stage of fatigue what is crucial for accident avoidance. This motivated to study how heart rate and its variability (HRV) relate to drivers sleepiness. The study was conducted on a population of 80 drivers that drove 3 times (morning, afternoon, night) for 80 to 90 minutes, with subjective sleepiness evaluations every 5 minutes, leading to over 3500 epochs to analyze. In order to derive HRV indices, outlier detection and spectral transformation need to be applied to the data. Different technics are available for these purposes with no consensus about their suitability. In this thesis, all relevant HRV indices were derived using the main technics for outliers detection and spectral transformation. Two methods were used for outliers detection, one based on a heart beat interval differing in more than 30% the average of the 4 beats before and the other consisting in an interval deviating 5 std from mean. Regarding spectral transformation, DFT, AR-model and LS Periodogram were considered. The agreement between methods was evaluated using Bland-Altman plots and Student T-Tests. Both, outlier detection and spectral transformation choices, showed to have a significant influence on the value of the HRV indices calculated. Spectral transformation showed a much higher influence.