Survival forests under test: Impact of the proportional hazards assumption on prognostic and predictive forests for amyotrophic lateral sclerosis survival
Restricted accessResearch articleFirst published online May, 2020
Survival forests under test: Impact of the proportional hazards assumption on prognostic and predictive forests for amyotrophic lateral sclerosis survival
We investigate the effect of the proportional hazards assumption on prognostic and predictive models of the survival time of patients suffering from amyotrophic lateral sclerosis. We theoretically compare the underlying model formulations of several variants of survival forests and implementations thereof, including random forests for survival, conditional inference forests, Ranger, and survival forests with L1 splitting, with two novel variants, namely distributional and transformation survival forests. Theoretical considerations explain the low power of log-rank-based splitting in detecting patterns in non-proportional hazards situations in survival trees and corresponding forests. This limitation can potentially be overcome by the alternative split procedures suggested herein. We empirically investigated this effect using simulation experiments and a re-analysis of the Pooled Resource Open-Access ALS Clinical Trials database of amyotrophic lateral sclerosis survival, giving special emphasis to both prognostic and predictive models.
ChiòALogroscinoGHardimanO, et al.Prognostic factors in ALS: A critical review. Amyotroph Lateral Scler2009; 10: 310–323. DOI: 10.3109/17482960802566824.
2.
KimuraFFujmuraCIshidaS, et al.Progression rate of ALSFRS-R at time of diagnosis predicts survival time in ALS. Neurology2006; 66: 265–267. DOI: 10.1212/01.wnl. 0000194316.91908.8a.
3.
ZoccolellaSBeghiEPalaganoG, et al.Analysis of survival and prognostic factors in amyotrophic lateral sclerosis: A population based study. J Neurol Neurosurg Psychiatry2008; 79: 33–37. DOI: 10.1136/jnnp.2007.118018.
4.
Fujimura-KiyonoCKimuraFIshidaS, et al.Onset and spreading patterns of lower motor neuron involvements predict survival in sporadic amyotrophic lateral sclerosis. J Neurol Neurosurg Psychiatry2011; 82: 1244–1249. DOI: 10.1136/ jnnp-2011-300141.
5.
Beaulieu-JonesBKGreeneCSthe Pooled Resource Open-Access ALS ClinicalTrials. Semi-supervised learning of the electronic health record for phenotype stratification. J Biomed Inform2016; 64: 168–178. DOI: 10.1016/j.jbi.2016.10.007.
6.
MandrioliJRosiEFiniN, et al.Changes in routine laboratory tests and survival in amyotrophic lateral sclerosis. Neurol Sci2017; 38: 2177–2182. DOI: 10.1007/s10072-017-3138-8.
7.
OngMLTanPFHolbrookJD. Predicting functional decline and survival in amyotrophic lateral sclerosis. PLoS ONE2017; 12: e0174925DOI: 10.1371/journal.pone. 0174925.
8.
PfohlSRKimRBCoanGS, et al.Unraveling the complexity of amyotrophic lateral sclerosis survival prediction. Front Neuroinf2018; 12: 12DOI: 10.3389/fninf. 2018.00036.
9.
BrooksBRSanjakMRingelS, et al.The Amyotrophic Lateral Sclerosis Functional Rating Scale: Assessment of activities of daily living in patients with amyotrophic lateral sclerosis. Arch Neurol1996; 53: 141–147.
10.
CedarbaumJMStamblerNMaltaE, et al.The ALSFRS-R: A revised ALS functional rating scale that incorporates assessments of respiratory function. J Neurol Sci1999; 169: 13–21.
11.
HothornTJungHH. RandomForest4Life: A random forest for predicting ALS disease progression. Amyotrop Lateral Scler Frontotemporal Degener2014; 15: 444–452. DOI: 10.3109/21678421.2014.893361.
12.
KüffnerRZachNNorelR, et al.Crowdsourced analysis of clinical trial data to predict amyotrophic lateral sclerosis progression. Nat Biotechnol2015; 33: 51–57. DOI: 10.1038/nbt.3051.
13.
FangTKhleifatAAMeurgeyJH, et al.Stage at which riluzole treatment prolongs survival in patients with amyotrophic lateral sclerosis: A retrospective analysis of data from a dose-ranging study. Lancet Neurol2018; 17: 416–422. DOI: 10.1016/S1474-4422 30054-1.
14.
SeiboldHZeileisAHothornT. Model-based recursive partitioning for subgroup analyses. Int J Biostat2016; 12: 45–63. DOI: 10.1515/ijb-2015-0032.
15.
SeiboldHZeileisAHothornT. Individual treatment effect prediction for amyotrophic lateral sclerosis patients. Stat Methods Med Res2018; 27: 3104–3125.
16.
IshwaranHKogalurUBBlackstoneEH, et al.Random survival forests. Ann Appl Stat2008; 2: 841–860.
17.
HothornTZeileisA. partykit: A modular toolkit for recursive partytioning in R. J Mach Learn Res2015; 16: 3905–3909.
18.
HothornTLausenBBennerA, et al.Bagging survival trees. Stat Med2004; 23: 77–91.
19.
WrightMNZieglerA. ranger: A fast implementation of random forests for high dimensional data in C++ and R. J Stat Softw2017, pp. 77.
20.
DatemaFRMoyaAKrauseP, et al.Novel head and neck cancer survival analysis approach: Random survival forests versus Cox proportional hazards regression. Head Neck2012; 34: 50–58.
21.
AtheySTibshiraniJWagerS. Generalized random forests. Ann Stat2019; 47: 1148–1178.
22.
Hothorn T and Zeileis A. Transformation forests. Technical report, arXiv 1701.02110, v2, 2017, https://arxiv.org/abs/1701.02110.
23.
SchlosserLHothornTStaufferR, et al.Distributional regression forests for probabilistic precipitation forecasting in complex terrain. Ann Appl Stat.. Epub ahead of print 21 February 2019, https://arxiv.org/abs/1804.02921.
24.
CiampiAThiffaultJNakacheJP, et al.Stratification by stepwise regression, correspondence analysis and recursive partition: A comparison of three methods of analysis for survival data with covariates. Comput Stat Data An1986; 3: 185–204.
25.
SegalMR. Regression trees for censored data. Biometrics1988; 44: 35–47.
26.
WrightMNDankowskiTZieglerA. Unbiased split variable selection for random survival forests using maximally selected rank statistics. Stat Med2017; 36: 1272–1284.
27.
ShimokawaAKawasakiYMiyaokaE. Comparison of splitting methods on survival tree. Int J Biostat2015; 11: 175–188.
28.
MoradianHLarocqueDBellavanceF. L1 splitting rules in survival forests. Lifetime Data Anal2017; 23: 671–691.
29.
HothornTMöstLBühlmannP. Most likely transformations. Scand J Stat2018; 45: 110–134.
LinYJeonY. Random forests and adaptive nearest neighbors. J Am Stat Assoc2006; 101: 578–590.
32.
McLainACGhoshSK. Efficient sieve maximum likelihood estimation of time-transformation models. J Stat Theory Prac2013; 7: 285–303.
33.
FriedmanJH. Multivariate adaptive regression splines. Ann Stat1991; 19: 1–67.
34.
MorganJNSonquistJA. Problems in the analysis of survey data, and a proposal. J Am Stat Assoc1963; 58: 415–434.
35.
BreimanL. Random forests. Mach Learn2001; 45: 5–32.
36.
BreimanL. Statistical modeling: The two cultures. Stat Sci2001; 16: 199–231.
37.
FuWSimonoffJS. Survival trees for interval-censored survival data. Stat Med2017; 36: 4831–4842.
38.
DrouinAHockingTDLavioletteFMaximum margin interval trees. In: GuyonILuxburgUVBengioS, et al.(eds). Advances in neural information processing systems 30, Red Hook, NY: Curran Associates, Inc., 2017, pp. 4947–4956.
39.
GrafESchmoorCSauerbreiW, et al.Assessment and comparison of prognostic classification schemes for survival data. Stat Med1999; 18: 2529–2545.
40.
RoyMHLarocqueD. Prediction intervals with random forests. Stat Methods Med Res.. Epub ahead of print 2019. DOI: 10.1177/0962280219829885.
41.
ProbstPWrightMNBoulesteixAL. Hyperparameters and tuning strategies for random forest. Wiley Interdiscip Rev: Data Min Knowl Discovery2019, pp. 9DOI:10.1002/widm.1301.
42.
HarrellFECaliffRMPryorDB, et al.Evaluating the yield of medical tests. JAMA1982; 247: 2543–2546.
43.
SchmidMHielscherTAugustinT, et al.A robust alternative to the Schemper-Henderson estimator of prediction error. Biometrics2011; 67: 524–535.
44.
SchmidMWrightMNZieglerA. On the use of Harrell's C for clinical risk prediction via random survival forests. Expert Syst Appl2016; 63: 450–459. DOI: 10.1016/j.eswa.2016.07.018.
45.
MogensenUIshwaranHGerdsT. Evaluating random forests for survival analysis using prediction error curves. J Stat Softw2012; 50: 1–23.
46.
R Core Team. R: A language and environment for statistical computing, Vienna, Austria: R Foundation for Statistical Computing, 2019.
Ishwaran H and Kogalur UB. randomForestSRC: Fast unified random forests for survival, regression, and classification (RF-SRC), https://CRAN.R-project.org/package=randomForestSRC. R package version 2.9.0 (2019).
50.
Hothorn T, Hornik K, Strobl C, et al. party: A laboratory for recursive partytioning, http://party.R-forge.R-project.org. R package version 1.3-3 (2019).
