20 November 2013 in Computational and Mathematical Methods in Medicine
Survival analysis focuses on modeling and predicting the time to an event of interest. Many statistical models have been proposed for survival analysis. They often impose strong assumptions on hazard functions, which describe how the risk of an event changes over time depending on covariates associated with each individual. In particular, the prevalent proportional hazards model assumes that covariates are multiplicatively related to the hazard. Here we propose a nonparametric model for survival analysis that does not explicitly assume particular forms of hazard functions. Our nonparametric model utilizes an ensemble of regression trees to determine how the hazard function varies according to the associated covariates. The ensemble model is trained using a gradient boosting method to optimize a smoothed approximation of the concordance index, which is one of the most widely used metrics in survival model performance evaluation. We implemented our model in a software package called GBMCI (gradient boosting machine for concordance index) and benchmarked the performance of our model against other popular survival models with a large-scale breast cancer prognosis dataset. Our experiment shows that GBMCI consistently outperforms other methods based on a number of covariate settings. GBMCI is implemented in R and is freely available online.
26 March 2024 in IEEE Transactions on Instrumentation and Measurement
Although the envelope-spectrum-based methods for bearing fault diagnosis have been widespread in the scientific community, their application to autonomous diagnosis is hindered by the specified selection of informative frequency bands and the threshold calculation. This paper therefore proposes a novel autonomous diagnosis method via Fault-Induced Envelope Spectrum (FIES) and Moving Peaks-Over-Threshold (MPOT) approach. A fault-induced filter is first designed to reveal all the informative bands of the Spectral Coherence (SCoh) rather than only a specified band. Then, the FIES is used to extract each fault signature, which weights and integrates along the spectral frequency axis of the SCoh. Subsequently, the MPOT is proposed to calculate a frequency-dependent threshold for the FIES, which not only concentrates the heavy-tailed statistical characteristics of faults, but also removes the influence of the non-stationary statistical characteristics for the threshold. Finally, the healthy indicator and suspected fault indicator are compared to warn users of the possible risk, meanwhile making a decision for autonomous diagnosis. The effectiveness of proposed method is verified by the experimental data. Results are found superior to two existing envelope-spectrum-based methods, which is more practical in terms of autonomous fault diagnosis and health monitoring.
10 January 2025 in IEEE Transactions on Instrumentation and Measurement
Although the condition monitoring of sophisticated equipment plays a crucial role in practical applications, the fault symptom is often overwhelmed by the harsh working conditions with compound and unknown interfering noise. To address this challenge, this paper introduces a novel signal decomposition scheme, termed the Complex Symplectic Geometry Mode Decomposition (CSGMD), that leverages the instantaneous complex envelope to ensure the complex statistical structure hidden in a signal. First, the applicable condition of the symplectic similar transform theory is extended into the complex domain to facilitate the exploitation of the phase information. Relying on phase-corrected Short-Time Fourier Transform, the CSGMD achieves a time-frequency signal decomposition while enhancing the interpretation of the physical significance of the feature extraction. Target component reconstruction is then executed by using the statistical indicators, resulting in the reconstructed components that are enriched with the original characteristic information. The effectiveness of the proposed method is finally demonstrated with both simulated and measured signals. Through the comparative experiments of the state-of-the-art methods, its superiority of feature extraction and signal decomposition is further verified with non trivial cases of complex and strong noise interference.