Optimizing dynamic predictions from joint models using super learning
© 2024 The Authors. Statistics in Medicine published by John Wiley & Sons Ltd..
Joint models for longitudinal and time-to-event data are often employed to calculate dynamic individualized predictions used in numerous applications of precision medicine. Two components of joint models that influence the accuracy of these predictions are the shape of the longitudinal trajectories and the functional form linking the longitudinal outcome history to the hazard of the event. Finding a single well-specified model that produces accurate predictions for all subjects and follow-up times can be challenging, especially when considering multiple longitudinal outcomes. In this work, we use the concept of super learning and avoid selecting a single model. In particular, we specify a weighted combination of the dynamic predictions calculated from a library of joint models with different specifications. The weights are selected to optimize a predictive accuracy metric using V-fold cross-validation. We use as predictive accuracy measures the expected quadratic prediction error and the expected predictive cross-entropy. In a simulation study, we found that the super learning approach produces results very similar to the Oracle model, which was the model with the best performance in the test datasets. All proposed methodology is implemented in the freely available R package JMbayes2.
Medienart: |
E-Artikel |
---|
Erscheinungsjahr: |
2024 |
---|---|
Erschienen: |
2024 |
Enthalten in: |
Zur Gesamtaufnahme - volume:43 |
---|---|
Enthalten in: |
Statistics in medicine - 43(2024), 7 vom: 30. März, Seite 1315-1328 |
Sprache: |
Englisch |
---|
Beteiligte Personen: |
Rizopoulos, Dimitris [VerfasserIn] |
---|
Links: |
---|
Themen: |
Brier score |
---|
Anmerkungen: |
Date Completed 18.03.2024 Date Revised 18.03.2024 published: Print-Electronic Citation Status MEDLINE |
---|
doi: |
10.1002/sim.10010 |
---|
funding: |
|
---|---|
Förderinstitution / Projekttitel: |
|
PPN (Katalog-ID): |
NLM367604868 |
---|
LEADER | 01000caa a22002652 4500 | ||
---|---|---|---|
001 | NLM367604868 | ||
003 | DE-627 | ||
005 | 20240318234128.0 | ||
007 | cr uuu---uuuuu | ||
008 | 240125s2024 xx |||||o 00| ||eng c | ||
024 | 7 | |a 10.1002/sim.10010 |2 doi | |
028 | 5 | 2 | |a pubmed24n1334.xml |
035 | |a (DE-627)NLM367604868 | ||
035 | |a (NLM)38270062 | ||
040 | |a DE-627 |b ger |c DE-627 |e rakwb | ||
041 | |a eng | ||
100 | 1 | |a Rizopoulos, Dimitris |e verfasserin |4 aut | |
245 | 1 | 0 | |a Optimizing dynamic predictions from joint models using super learning |
264 | 1 | |c 2024 | |
336 | |a Text |b txt |2 rdacontent | ||
337 | |a ƒaComputermedien |b c |2 rdamedia | ||
338 | |a ƒa Online-Ressource |b cr |2 rdacarrier | ||
500 | |a Date Completed 18.03.2024 | ||
500 | |a Date Revised 18.03.2024 | ||
500 | |a published: Print-Electronic | ||
500 | |a Citation Status MEDLINE | ||
520 | |a © 2024 The Authors. Statistics in Medicine published by John Wiley & Sons Ltd. | ||
520 | |a Joint models for longitudinal and time-to-event data are often employed to calculate dynamic individualized predictions used in numerous applications of precision medicine. Two components of joint models that influence the accuracy of these predictions are the shape of the longitudinal trajectories and the functional form linking the longitudinal outcome history to the hazard of the event. Finding a single well-specified model that produces accurate predictions for all subjects and follow-up times can be challenging, especially when considering multiple longitudinal outcomes. In this work, we use the concept of super learning and avoid selecting a single model. In particular, we specify a weighted combination of the dynamic predictions calculated from a library of joint models with different specifications. The weights are selected to optimize a predictive accuracy metric using V-fold cross-validation. We use as predictive accuracy measures the expected quadratic prediction error and the expected predictive cross-entropy. In a simulation study, we found that the super learning approach produces results very similar to the Oracle model, which was the model with the best performance in the test datasets. All proposed methodology is implemented in the freely available R package JMbayes2 | ||
650 | 4 | |a Journal Article | |
650 | 4 | |a Brier score | |
650 | 4 | |a cross-entropy | |
650 | 4 | |a precision medicine | |
650 | 4 | |a prognostic models | |
650 | 4 | |a survival analysis | |
650 | 4 | |a time-varying covariates | |
700 | 1 | |a Taylor, Jeremy M G |e verfasserin |4 aut | |
773 | 0 | 8 | |i Enthalten in |t Statistics in medicine |d 1984 |g 43(2024), 7 vom: 30. März, Seite 1315-1328 |w (DE-627)NLM012664596 |x 1097-0258 |7 nnns |
773 | 1 | 8 | |g volume:43 |g year:2024 |g number:7 |g day:30 |g month:03 |g pages:1315-1328 |
856 | 4 | 0 | |u http://dx.doi.org/10.1002/sim.10010 |3 Volltext |
912 | |a GBV_USEFLAG_A | ||
912 | |a GBV_NLM | ||
951 | |a AR | ||
952 | |d 43 |j 2024 |e 7 |b 30 |c 03 |h 1315-1328 |