F7ABBAZD - Biomedical Data Analysis and Processing

Time series analysis, trends, mutual dependency, stationarity. Correlation function and covariance function. Algorithms of correlation function estimation. Impact of removing trends to autocorrelation function. Periodogram - relationship between corellogram and periodogram. Frequency spectrum, spectrum of random signals. Linear frequency filtering. AR, ARMA, and MA processes. Spectral analysis. FFT algorithm. Non-parametric methods of the frequency spectrum estimation. Positives and negatives of the specteal analysis. Repeated measurements and analysis of their properties. AR a ARMA model parameter identification. Prediction. Bivariance analysis of time series - cross-correlation and cross-covariance and their estimation. Bispectrum.

- A maximum of 2 absences from practical classes is permitted.

- Complete control tests; maximum of 50 points (two control tests worth 25 points each during the semester), with a minimum total of 25 points required.

- Complete a final test at the end of the semester covering the lecture material; maximum of 50 points.

- Grading will be based on the ECTS scale: F (<50), E (5059), D (6069), C (7079), B (8089), A (90100).

1. Time series analysis - fundamentals; trends, mutual dependency, stationarity. Correlation function and covariance function. Algorithms of correlation function estimation.

2. Impact of removing trends to autocorrelation function. Periodogram - relationship between corellogram and periodogram.

3. Frequency spectrum, spectrum of random signals. Linear frequency filtering.

4. AR, ARMA, and MA processes. Spectral analysis. FFT algorithm.

5. Non-parametric methods of the frequency spectrum estimation. Positives and negatives of the spectral analysis.

6. Repeated measurements and analysis of their properties.11. AR and ARMA model parameter identification.

7. Prediction. Bivariance analysis of time series - cross-correlation and cross-covariance. Estimation of cross-correlation and cross-covariance functions. Bispectrum.

1. Fourier transform. Frequency spectrum, frequency spectrum of random signals.

2. Time series filtering, time series decomposition.

3. Box-Jenkins methodology.

4. Control test (25 points). Interpolation and time series processing.

5. Cluster analysis.

6. Neural networks.

7. Control test (25 points).

to provide students with basic methods of statistical processing time series typical for life sciences

[1]Diggle P.J. Time Series. A Biostatistical Introduction. Clarendon Press. Oxford 1996

[2]Weiss S.M., Indurkhya N. Predictive Data Mining

• Biomedical Technology (compulsory elective course)
• Biomedical Technology (compulsory elective course)