What does this research mean for the field?

Extending the PyCox deep learning survival analysis framework to accommodate time-varying covariates improves computational efficiency while maintaining estimation accuracy comparable to standard time-varying Cox regression. Novelty: ClaimNovelty.METHODOLOGICAL. Consensus alignment: ConsensusAlignment.NEUTRAL.

What question did this study set out to answer?

To extend the PyCox deep learning survival analysis framework to handle counting-process data structures with time-varying covariates, mitigating immortal time bias in observational healthcare studies.

September 6, 2026JAMIA OpenOpen Access

Deep Learning Survival Analysis with Time-Varying Covariates: Extending PyCox to Assess COVID-19 Antiviral Treatments and Long-COVID Associations

View Full Paper

Ask AI

Bookmark

Authors

SBSeo Hyon Baik HXHaotian Xian FBFitsum Baye

Discussion

Member takes

Overview

Retrospective cohort study demonstrates reduced long-COVID risk from antivirals in older adults, suggesting modified deep learning matches standard accuracy with superior speed.

Key Points

To extend the PyCox deep learning survival analysis framework to handle counting-process data structures with time-varying covariates, mitigating immortal time bias in observational healthcare studies.
Modified PyCox to accommodate longitudinal counting-process records and evaluated it on 2,246,913 Medicare beneficiaries aged ≥65 diagnosed with COVID-19 between January and September 2022 (2,785,807 total records).
Compared modified PyCox against standard time-varying Cox regression and original time-fixed PyCox to evaluate associations between early antiviral treatment (nirmatrelvir or molnupiravir) and long-COVID.
Assessed model performance using concordance index, integrated Brier score, integrated negative binomial log-likelihood, and time-dependent AUC.
In the cohort (19.5% receiving nirmatrelvir, 2.6% receiving molnupiravir, and 14.0% developing long-COVID), modified PyCox produced hazard ratios concordant with time-varying Cox regression for nirmatrelvir (HR 0.878 vs 0.874) and molnupiravir (HR 0.918 vs 0.909), whereas time-fixed PyCox estimated stronger associations (HR 0.822 and 0.879).
Modified PyCox achieved higher time-dependent AUCs (0.536–0.609) than time-fixed PyCox (0.481–0.519) while demonstrating comparable overall discrimination and calibration to traditional Cox modeling.
Modified PyCox achieved the fastest computational runtime, training in 1 minute 14 seconds compared to 3 minutes 18 seconds for original PyCox and 5 minutes for traditional Cox regression.

Cite This Study

Baik et al. (2026) studied this question.

synapsesocial.com/papers/6a9d1e3328139818eab20e2b https://doi.org/10.1093/jamiaopen/ooag186

View Full Paper

Ask AI

Bookmark

September 6, 2026JAMIA OpenOpen Access

Deep Learning Survival Analysis with Time-Varying Covariates: Extending PyCox to Assess COVID-19 Antiviral Treatments and Long-COVID Associations

View Full Paper

Ask AI

Bookmark

Authors

SBSeo Hyon Baik HXHaotian Xian FBFitsum Baye

Discussion

Member takes

Overview

Retrospective cohort study demonstrates reduced long-COVID risk from antivirals in older adults, suggesting modified deep learning matches standard accuracy with superior speed.

Key Points

To extend the PyCox deep learning survival analysis framework to handle counting-process data structures with time-varying covariates, mitigating immortal time bias in observational healthcare studies.
Modified PyCox to accommodate longitudinal counting-process records and evaluated it on 2,246,913 Medicare beneficiaries aged ≥65 diagnosed with COVID-19 between January and September 2022 (2,785,807 total records).
Compared modified PyCox against standard time-varying Cox regression and original time-fixed PyCox to evaluate associations between early antiviral treatment (nirmatrelvir or molnupiravir) and long-COVID.
Assessed model performance using concordance index, integrated Brier score, integrated negative binomial log-likelihood, and time-dependent AUC.
In the cohort (19.5% receiving nirmatrelvir, 2.6% receiving molnupiravir, and 14.0% developing long-COVID), modified PyCox produced hazard ratios concordant with time-varying Cox regression for nirmatrelvir (HR 0.878 vs 0.874) and molnupiravir (HR 0.918 vs 0.909), whereas time-fixed PyCox estimated stronger associations (HR 0.822 and 0.879).
Modified PyCox achieved higher time-dependent AUCs (0.536–0.609) than time-fixed PyCox (0.481–0.519) while demonstrating comparable overall discrimination and calibration to traditional Cox modeling.
Modified PyCox achieved the fastest computational runtime, training in 1 minute 14 seconds compared to 3 minutes 18 seconds for original PyCox and 5 minutes for traditional Cox regression.

Cite This Study

Baik et al. (2026) studied this question.

synapsesocial.com/papers/6a9d1e3328139818eab20e2b https://doi.org/10.1093/jamiaopen/ooag186

Also Consider

Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context:

View Full Paper

Ask AI

Bookmark