DeepPAMM: Deep Piecewise Exponential Additive Mixed Models for Complex Hazard Structures in Survival Analysis

Kopper, Philipp; Wiegrebe, Simon; Bischl, Bernd; Bender, Andreas; Rügamer, David

doi:10.48550/arxiv.2202.07423

Cited by 3 publications

(1 citation statement)

References 35 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…High-dimensional models, including DeepConvSurv [6] and CapSurv [7], showcase the newfound ability to integrate and interpret information from diverse data modalities, such as images, using convolutional neural networks. Furthermore, the incorporation of deep learning in parametric models, as observed in DeepPAMM [8] and DPWTE [9], advances survival analysis frameworks that leverage probabilistic models for more accurate predictions. DeepHit [10] introduces an innovative discrete-time approach to manage time-to-event data through classification techniques.…”

Section: Introductionmentioning

confidence: 99%

Deep Survival Analysis for Interpretable Time-Varying Prediction of Preeclampsia Risk

Eberhard,

Gray,

Bates

et al. 2024

Preprint

View full text Add to dashboard Cite

Objective: Survival analysis is widely utilized in healthcare to predict the timing of disease onset. Traditional methods of survival analysis are usually based on Cox Proportional Hazards model and assume proportional risk for all subjects. However, this assumption is rarely true for most diseases, as the underlying factors have complex, non-linear, and time-varying relationships. This concern is especially relevant for pregnancy, where the risk for pregnancy-related complications, such as preeclampsia, varies across gestation. Recently, deep learning survival models have shown promise in addressing the limitations of classical models, as the novel models allow for non-proportional risk handling, capturing nonlinear relationships, and navigating complex temporal dynamics. Methods: We present a methodology to model the temporal risk of preeclampsia during pregnancy and investigate the associated clinical risk factors. We utilized a retrospective dataset including 66,425 pregnant individuals who delivered in two tertiary care centers from 2015-2023. We modeled the preeclampsia risk by modifying DeepHit, a deep survival model, which leverages neural network architecture to capture time-varying relationships between covariates in pregnancy. We applied time series k-means clustering to DeepHit's normalized output and investigated interpretability using Shapley values. Results: We demonstrate that DeepHit can effectively handle high-dimensional data and evolving risk hazards over time with performance similar to the Cox Proportional Hazards model, achieving an area under the curve (AUC) of 0.78 for both models. The deep survival model outperformed traditional methodology by identifying time-varied risk trajectories for preeclampsia, providing insights for early and individualized intervention. K-means clustering resulted in patients delineating into low-risk, early-onset, and late-onset preeclampsia groups-notably, each of those has distinct risk factors. Conclusion: This work demonstrates a novel application of deep survival analysis in time-varying prediction of preeclampsia risk. Our results highlight the advantage of deep survival models compared to Cox Proportional Hazards models in providing personalized risk trajectory and demonstrating the potential of deep survival models to generate interpretable and meaningful clinical applications in medicine.

show abstract

Section: Introductionmentioning

confidence: 99%

Deep Survival Analysis for Interpretable Time-Varying Prediction of Preeclampsia Risk

Eberhard,

Gray,

Bates

et al. 2024

Preprint

View full text Add to dashboard Cite

show abstract

Plasmodium falciparum gametocyte carriage in longitudinally monitored incident infections is associated with duration of infection and human host factors

Andolina

Ramjith

Rek

et al. 2023

Sci Rep

View full text Add to dashboard Cite

Malaria transmission depends on the presence of Plasmodium gametocytes that are the only parasite life stage that can infect mosquitoes. Gametocyte production varies between infections and over the course of infections. Infection duration is highly important for gametocyte production but poorly quantified. Between 2017 and 2019 an all-age cohort of individuals from Tororo, eastern Uganda was followed by continuous passive and routine assessments. We longitudinally monitored 104 incident infections from 98 individuals who were sampled once every 28 days and on any day of symptoms. Among infections that lasted ≥ 3 months, gametocyte appearance was near-universal with 96% of infections having detectable gametocytes prior to clearance. However, most infections were of much shorter duration; 55.7% of asymptomatic infections were detected only once. When considering all asymptomatic infections, regardless of their duration, only 36.3% had detectable gametocytes on at least one time-point prior to parasite clearance. Infections in individuals with sickle-cell trait (HbAS) were more likely to have gametocytes detected (Hazard Rate (HR) = 2.68, 95% CI 1.12, 6.38; p = 0.0231) and had gametocytes detected at higher densities (Density Ratio (DR) = 9.19, 95% CI 2.79, 30.23; p = 0.0002) compared to infections in wildtype (HbAA) individuals. Our findings suggest that a large proportion of incident infections is too short in duration and of too low density to contribute to onward transmission.

show abstract

Plasmodium falciparum gametocyte production in longitudinally monitored incident infections is associated with duration of infection and human host factors

Andolina

Ramjith

Rek

et al. 2022

Preprint

View full text Add to dashboard Cite

Malaria transmission depends on the presence of Plasmodium gametocytes that are the only parasite life stage that can infect mosquitoes. Gametocyte production varies between infections and over the course of infections. Infection duration is highly important for gametocyte production but poorly quantified. Between 2017–2019 an all-age cohort of individuals from Tororo, eastern Uganda was followed by continuous passive and routine assessments. We longitudinally monitored 104 incident infections from 98 individuals. Among infections that lasted ≥ 3 months, gametocyte production was near-universal with 96% of infections producing gametocytes prior to clearance. However, most infections were of much shorter duration; 55.7% of asymptomatic infections were detected only once. When considering all asymptomatic infections, regardless of their duration, only 36.3% initiated gametocyte production prior to clearance. Infections in individuals with sickle-cell trait (HbAS) were more likely to produce gametocytes (Hazard Rate (HR) = 2.68, 95% CI: 1.12, 6.38; p = 0.0231) and produced gametocytes at higher densities (Density Ratio (DR) = 9.19, 95% CI: 2.79–30.23; p = 0.0002) compared to wildtype (HbAA) individuals. Our findings suggest that a large proportion of incident infections is too short in duration and of too low density to contribute to onward transmission.

show abstract

DeepPAMM: Deep Piecewise Exponential Additive Mixed Models for Complex Hazard Structures in Survival Analysis

Cited by 3 publications

References 35 publications

Deep Survival Analysis for Interpretable Time-Varying Prediction of Preeclampsia Risk

Deep Survival Analysis for Interpretable Time-Varying Prediction of Preeclampsia Risk

Plasmodium falciparum gametocyte carriage in longitudinally monitored incident infections is associated with duration of infection and human host factors

Plasmodium falciparum gametocyte production in longitudinally monitored incident infections is associated with duration of infection and human host factors

Contact Info

Product

Resources

About