A sequential threshold cure model for genetic analysis of time-to-event data.

Area of Science:

• Quantitative genetics
• Statistical modeling
• Aquaculture breeding

Background:

• Classical survival models are inadequate when populations contain nonsusceptible (cured) individuals.
• Nonsusceptible individuals are common in challenge testing, crucial for aquaculture breeding schemes.
• Cure models address this by accounting for a fraction of nonsusceptible individuals.

Purpose of the Study:

• To propose a novel mixed cure model for time-to-event data, specifically for sequential binary records.
• To evaluate the model's ability to differentiate underlying traits like susceptibility and endurance.
• To assess the impact of ignoring nonsusceptible individuals on selection accuracy in breeding programs.

Main Methods:

• A mixed cure model was developed for time-to-event data represented as sequential binary records.
• A simulation study generated survival data based on underlying liabilities for susceptibility and endurance.
• The model's performance was evaluated under varying degrees of phenotypic confounding and genetic correlation.

Main Results:

• The proposed mixed cure model successfully distinguished between susceptibility and endurance traits, even with significant phenotypic confounding.
• Applying classical survival models led to non-negligible errors, particularly when selection targeted susceptibility.
• Model accuracy was highest with substantial genetic variation in endurance and low genetic correlation between traits.

Conclusions:

• The mixed cure model offers a more accurate method for analyzing time-to-event data in the presence of nonsusceptible individuals.
• This approach is valuable for aquaculture breeding schemes and accurately utilizes time-to-event data.
• The model also demonstrated success with zero-inflated longitudinal binary data.