Lifetime analysis with monotonic degradation: a boosted first hitting time model based on a homogeneous gamma process

Area of Science:

• Statistics
• Survival Analysis
• Machine Learning

Background:

• First hitting time (FHT) methods are valuable in time-to-event analysis, modeling events as the endpoint of evolving processes.
• These models are particularly useful when the degradation path is known, especially for monotonic processes where degradation only worsens.

Purpose of the Study:

• To propose a novel boosting algorithm for FHT models.
• To incorporate a homogeneous gamma process to ensure monotonicity in degradation trends.
• To demonstrate the algorithm's predictive power and versatility across diverse applications.

Main Methods:

• Development of a boosting algorithm tailored for FHT models.
• Utilizing an underlying homogeneous gamma process to model monotonic degradation.
• Validation using real-world data from engineering and biomedical fields, alongside simulated data.

Main Results:

• The proposed boosting algorithm effectively accounts for monotonic degradation trends.
• The algorithm exhibits strong predictive performance in both real and simulated datasets.
• Successful application demonstrated in engineering (e.g., component lifetime) and biomedical (e.g., disease progression) contexts.

Conclusions:

• The novel boosting algorithm provides a robust approach for FHT analysis with monotonic degradation.
• The method's versatility makes it applicable to a wide range of time-to-event problems.
• This work enhances the utility of FHT models in analyzing processes with inherent monotonicity.