| 研究実績の概要 |
During the fiscal year 2017 large number of datasets of the photoplethysmogram measured at green light (gPPG) were obtained, processed and analyzed. Analyzed data included reference data and the data measured with consideration of parameters such as subject’s age and the environment temperature in which data were recorded. To investigate essential properties of the gPPG dynamics, its chaotic characteristics were tested by the methods of nonlinear time series analysis, such as time-delay reconstruction, embedding dimension estimation by the false-nearest-neighbors method, the largest Lyapunov exponent, Wayland test translation error, deterministic nonlinear prediction, and the 0-1 chaos test. Particular attention was paid to the dependence of calculated indexes to such essential biological factor as subject’s age, and to the temperature and light as environmental factors. The obtained results demonstrated that based on applied methods clear differences can be found in the data corresponding to the different ages and that the effect of the observational noise is greater on the gPPG dynamics corresponding to elder people. It was also found that the level of observational noise may increase if the data were taken in a hot environment. Also work toward developing a modification of prediction technique, which can take into account information on local noise levels, was started on based on example of Lorenz model.
The results were presented at one international conference, one domestic symposium, and one peer-reviewed conference paper was accepted for publication in 2018.
|
| 現在までの達成度 (区分) |
現在までの達成度 (区分)
2: おおむね順調に進展している
理由
Properties of a large number of datasets were analyzed. Also, the factors, that can commonly variate at each data measurement, were investigated in order to find whether the changes of these factors may significantly increase noise effect on the gPPG dynamics and whether these factors must be taken into consideration as additional parameters for the denoising method. As the gPPG properties were comprehensively investigated, it creates a strong basis for developing prediction-based denoising method for the gPPG, planned for the fiscal year 2018.
|
| 今後の研究の推進方策 |
For the fiscal year 2018 results obtained at the previous year will be used for the development of prediction technique based on chaotic properties information, which can be used for locally-focused noise reduction. However, at first, locally-focused prediction will be developed on the example of noise-induced Lorenz and Rossler models, and then it will be adapted to the properties of the gPPG signal.
|
