Trinomial probabilistic modeling of full, marginal, and no decompression sickness

Loading...
Thumbnail Image

Date

2018

Journal Title

Journal ISSN

Volume Title

Repository Usage Stats

82
views
77
downloads

Abstract

Decompression sickness (DCS) is a condition associated with reductions in ambient pressure during underwater diving. Determining the risk of DCS from a dive exposure remains an active area of research, to develop safe decompression schedules to mitigate the occurrences of DCS. This thesis develops a probabilistic model for the trinomial outcome of full, marginal, and no DCS. These models determine probabilities of the various outcomes for a given dive schedule. Six variants of exponential-exponential (EE) and linear-exponential (LE) decompression models were used for optimization of model parameters to best fit dive outcomes from empirical data of 3,322 exposures. Using the log likelihood difference test, the LE1 model was determined to provide the best fit for the data when considering full events along with marginal DCS events as separate, hierarchical events with a weighting of 0.1. The LE1 trinomial marginal model can be used to better understand decompression schedules, expanding upon binomial models which treat full DCS as an event and marginal DCS as a non-event. Future work could investigate whether the use of marginal DCS cases improves probabilistic DCS models. Model improvement could include the addition of a fourth outcome, where full DCS is split and categorized as serious or mild DCS, creating a tetranomial model with serious, mild, marginal, and no DCS outcomes for comparison with the presently developed model.

Description

Provenance

Citation

Citation

Andriano, Nicholas Ryan (2018). Trinomial probabilistic modeling of full, marginal, and no decompression sickness. Master's thesis, Duke University. Retrieved from https://hdl.handle.net/10161/17038.

Collections


Dukes student scholarship is made available to the public using a Creative Commons Attribution / Non-commercial / No derivative (CC-BY-NC-ND) license.