Decompression sickness predictive models for unsafe human exposure.

Area of Science:

• Diving Medicine
• Physiological Modeling
• Risk Assessment

Background:

• Existing decompression sickness (DCS) prediction models perform well for routine Navy diving.
• Extrapolation of these models to higher-risk exposures, such as emergency conditions, has been problematic.
• A need exists for improved DCS incidence prediction in severe diving scenarios.

Purpose of the Study:

• To develop and validate improved models for predicting decompression sickness (DCS) incidence.
• To calibrate models using a comprehensive dataset of high-incidence DCS cases from U.S. Navy diving trials.
• To evaluate a modified linear-exponential risk model incorporating relative supersaturation squared.

Main Methods:

• Assembled a calibration dataset of 3,300 single exposures with 200 DCS cases.
• Evaluated a variant of the linear-exponential risk model with instantaneous risk defined as relative supersaturation squared.
• Assessed model goodness of fit using maximum likelihood and comparative analysis of observed versus predicted cases.
• Employed multimodel inferences, weighting four well-fitting models using the Akaike Information Criterion.

Main Results:

• Four distinct models demonstrated good fit to the calibration data.
• Two models utilized the established risk definition, while two incorporated the new squared supersaturation risk definition.
• Satisfactory model parameters were identified for each risk definition, with variations in gas kinetics treatment.
• Multimodel inference provided a combined model with weighted parameters.

Conclusions:

• The developed combined model shows promise for predicting DCS incidence in emergency diving preparations.
• The model is recommended for scenarios involving compressed-air exposures with a potential DCS incidence of 40% or higher.
• This research enhances the ability to prepare for and mitigate risks associated with severe diving exposures.