| 研究実績の概要 |
In the fiscal year 2018, the data analysis was continued on the photoplethysmogram data collected in the previous year of this project. The method of noise-contamination weighted prediction was proposed and tested on the chaotic Lorenz model and experimental data. It was revealed that while the proposed method of noise contamination weighted prediction demonstrated certain improvement in prediction quality in Lorenz model and part of experimental data, it could not provide stable improvement for the prediction-based reconstruction of noisy parts of the trajectory for all the variety of collected data. This indicated that the method should rely on the most fundamental properties of the data that are preserved for every type of experimental photoplethysmograms. Therefore further investigation of fundamental photoplethysmogram properties and photoplethysmogram dynamics' structures was conducted to achieve a better understanding of the properties that remain unchanged regardless of the subject's age and experimental settings. Additional attention was paid to the problem of the data preprocessing, which was found to have the potential to affect the results of the chaotic properties calculation significantly. As a result preprocessing method with minimum effect on the original data properties was found. The obtained results were reported in two peer-reviewed articles published in well-recognized international journals, one peer-reviewed conference article, presented in 3 talks at international conferences and in one presentation at the domestic symposium.
|
