Nonlinear systems identification by combining regression with bootstrap resampling

Area of Science:

• Nonlinear dynamics
• Statistical modeling
• Biomedical data analysis

Background:

• Accurate parameter estimation is crucial for understanding complex nonlinear systems.
• Traditional methods often struggle with short or noisy time series data.
• Unbiased estimation is essential for reliable model predictions.

Purpose of the Study:

• To propose a new, robust parameter estimation method for nonlinear systems.
• To address limitations of existing methods when dealing with challenging data.
• To demonstrate the applicability of the proposed method in diverse scenarios.

Main Methods:

• Utilizing the bootstrap method for unbiased parameter estimation in regression problems.
• Applying the novel method to benchmark chaotic models.
• Testing the method's efficacy on real clinical data from intermittent hormonal therapy for prostate cancer.

Main Results:

• The proposed bootstrap-based method provides unbiased and robust parameter estimates.
• Demonstrated practical applicability through successful estimation of chaotic models.
• Successfully applied to analyze intermittent hormonal therapy for prostate cancer using clinical data.

Conclusions:

• The new parameter estimation technique is effective for nonlinear systems, even with limited data.
• The bootstrap approach ensures robustness and reduces bias in estimations.
• The method holds significant potential for both theoretical modeling and clinical data analysis.