Modeling Composite Degradation Processes in Lifetime Data Analysis

• Activity Number: 160
• Type: Topic Contributed
• Date/Time: Monday, August 5, 2013 : 10:30 AM to 12:20 PM
• Sponsor: Section on Risk Analysis
• Author(s): George Whitmore*+ and Mei-Ling Ting Lee
• Companies: McGill University and University of Maryland at College Park
• Keywords: analytical time scales ; composite degradation processes ; first hitting times ; statistical models ; stochastic processes ; threshold regression

Abstract

In many applications of lifetime data analysis, multiple incommensurate degradation processes operate in parallel to produce failure. For example, a mechanical component may be subject to wear and tear as well as a series of independent random physical shocks that taken together can cause failure. As another example, a COPD patient may experience a slow physiological degeneration of lung function as well as random assaults from infections and allergens that seriously compromise lung health. COPD exacerbations and lung failures are caused by this composition of stochastic processes. This talk will discuss strategies for melding disparate degradation processes into coherent models. Special attention will be given to composite processes that produce failure when the process reaches a critical boundary for the first time (so-called first hitting time models) and to composite analytical time scales. These composite process models have important real-world applications in assessing risks of failure.