Singular Value Decomposition filtering, of vehicle component present in actograms recorded over moving platforms.

GALLARDO, JOSE; DIEZ, JOAQUIN; VIGO, DANIEL; NICOLA SIRI, LEONARDO

Tucuman

Congreso; III Latin American Federation of Biophysical Societies (LAFeBS) ? IX IberoAmerican Congress of Biophysics ? XLV Reunion Anual SAB 2016; 2016

Latin American Federation of Biophysical Societies, Sociedad Argentina de Biofisica

Driving a ground bus is a complex task. Drivers drowsiness and fatigueimpair travel safety. Buses are drived by couples of bus-drivers, whichalternate each between driving and non-driving stages 2 hours length.The actigraphy quantifies body kinetic activity in an actogram, in timessupplemented with a Sleep Log compiled by the subject; actographsattached to the subject are used. In order to identify patterns of resting/ activity on the studied subjects, actograms can be analyzed manuallyby a blind expert or by computed algorithms.Currently, actograms are recorded on motionless platforms, because ofcontamination by vehicular signals (as is the case for actigraphy on bus-drivers during travelling). In a previous work (1) this situation waspresent.In this work, attenuation of the vehicle component was obtained by fil-tering actograms by Singular Value Decomposition. Further, we devel-oped a method to automatically establish the intensity of kynetic activ-ity developed for each driver, specially during the non-driving period,to obtain the ?effective? resting period during the non-driving stages.We performed a retrospective study on 20 actograms recorded (1) frombus-drivers during 10 long distance travels. Concordance between rest-ing / activity periods obtained by actigraphy and driving / non-drivingstages informed in their Sleep Logs, was analyzed by Bland-Altman.Effective rest durations during non-driving periods were computed andaveraged over the 20 drivers, and results compared with previous re-ported by an blind expert (1) by using the Z statistic. Non significantdifferences (NS) were considered when p ≥ 0, 05.