Évaluation de prédictions dynamiques : quelques méthodes et applications au pronostic de la démence
Journal de la société française de statistique, Tome 157 (2016) no. 2, pp. 1-18.

L’utilisation de données longitudinales pour calculer des prédictions individuelles de risque dites “dynamiques” gagne actuellement en popularité. Les prédictions sont dites dynamiques car leur calcul est actualisé au fur et à mesure que l’information sur le profil de santé des sujets évolue au cours de leur suivi. Cet article papier présente des méthodes pour quantifier et comparer des capacités pronostiques pour ce type de prédictions. Une évaluation basée sur les deux concepts de discrimination et de calibration est suggérée et une approche non paramétrique de pondération par l’inverse de la probabilité de censure est présentée pour l’inférence. Cette approche permet de s’adapter naturellement à la présence de données censurées et de risques concurrents, deux situations fréquentes en recherche médicale. Quelques résultats asymptotiques sont présentés. Des tests et des régions de confiance en sont dérivés. Une application sur des prédictions du risque de démence chez les personnes âgées est discutée. Les prédictions sont basées sur les mesures répétées de deux tests psychométriques et sont issues de deux modèles précédemment estimés sur les données de la cohorte Paquid. Leurs capacités pronostiques sont quant à elles évaluées et comparées avec les données externes de la cohorte des Trois-Cités.

The computation of dynamic predictions, using longitudinal data, has recently become popular. The term dynamic emphasizes that the predictions can be updated when the information on the subjects’ health profiles increases with follow-up time. This paper presents methods to quantify and compare prognostic abilities of such dynamic predictions. The methods aim to evaluate the calibration and discrimination performances of the dynamic predictions. We focus on methods which handle censoring and which can be applied in the competing risks setting. Inverse probability of censoring weighting estimators are suggested. Tests and both pointwise and simultaneous confidence intervals are derived from asymptotic results. We illustrate the methods by comparing predictions of dementia in the elderly, accounting for the competing risk of death. Predictions are computed from two models which model the risk of dementia given repeated measures of two psychometric tests. The models are estimated on the Paquid cohort and prediction abilities are evaluated and compared using external data from the Three-City cohort.

Mot clés : calibration, censure, discrimination, données répétées, modèle conjoint, risques concurrents
Keywords: calibration, censoring, discrimination, repeated measurements, joint-model, competing risks
@article{JSFS_2016__157_2_1_0,
     author = {Blanche, Paul},
     title = {\'Evaluation de pr\'edictions dynamiques~: quelques m\'ethodes et applications au pronostic de la d\'emence},
     journal = {Journal de la soci\'et\'e fran\c{c}aise de statistique},
     pages = {1--18},
     publisher = {Soci\'et\'e fran\c{c}aise de statistique},
     volume = {157},
     number = {2},
     year = {2016},
     zbl = {1358.62086},
     language = {fr},
     url = {http://www.numdam.org/item/JSFS_2016__157_2_1_0/}
}
TY  - JOUR
AU  - Blanche, Paul
TI  - Évaluation de prédictions dynamiques : quelques méthodes et applications au pronostic de la démence
JO  - Journal de la société française de statistique
PY  - 2016
SP  - 1
EP  - 18
VL  - 157
IS  - 2
PB  - Société française de statistique
UR  - http://www.numdam.org/item/JSFS_2016__157_2_1_0/
LA  - fr
ID  - JSFS_2016__157_2_1_0
ER  - 
%0 Journal Article
%A Blanche, Paul
%T Évaluation de prédictions dynamiques : quelques méthodes et applications au pronostic de la démence
%J Journal de la société française de statistique
%D 2016
%P 1-18
%V 157
%N 2
%I Société française de statistique
%U http://www.numdam.org/item/JSFS_2016__157_2_1_0/
%G fr
%F JSFS_2016__157_2_1_0
Blanche, Paul. Évaluation de prédictions dynamiques : quelques méthodes et applications au pronostic de la démence. Journal de la société française de statistique, Tome 157 (2016) no. 2, pp. 1-18. http://www.numdam.org/item/JSFS_2016__157_2_1_0/

[1] Aisen, PS; Andrieu, S.; Sampaio, C.; Carrillo, M.; Khachaturian, ZS; Dubois, B.; Feldman, HH; Petersen, R.C.; Siemers, E.; Doody, RS Report of the task force on designing clinical trials in early (predementia) AD, Neurology, Volume 76 (2011) no. 3, pp. 280-286 | DOI

[2] Andersen, P. K.; Borgan, Ø.; Gill, R. D.; Keiding, N. Statistical Models Based on Counting Processes, New york : Springer Verlag, New York, 1993 | Zbl

[3] Aalen, Odd O; Johansen, Søren An empirical transition matrix for non-homogeneous Markov chains based on censored observations, Scandinavian Journal of Statistics (1978), pp. 141-150 | Zbl

[4] Amieva, Hélène; Le Goff, Mélanie; Millet, Xavier; Orgogozo, Jean Marc; Pérès, Karine; Barberger-Gateau, Pascale; Jacqmin-Gadda, Hélène; Dartigues, Jean François Prodromal Alzheimer’s disease : successive emergence of the clinical symptoms, Annals of neurology, Volume 64 (2008) no. 5, pp. 492-498

[5] Blanche, Paul; Proust-Lima, Cécile; Loubère, Lucie; Berr, Claudine; Dartigues, Jean-François; Jacqmin-Gadda, Hélène Quantifying and comparing dynamic predictive accuracy of joint models for longitudinal marker and time-to-event in presence of censoring and competing risks, Biometrics, Volume 71 (2015) no. 1, pp. 102-113 | Zbl

[6] Brier, Glenn W Verification of forecasts expressed in terms of probability, Monthly weather review, Volume 78 (1950) no. 1, pp. 1-3

[7] Beyersmann, Jan; Termini, Susanna Di; Pauly, Markus Weak convergence of the wild bootstrap for the Aalen–Johansen estimator of the cumulative incidence function of a competing risk, Scandinavian Journal of Statistics, Volume 40 (2013) no. 3, pp. 387-402 | Zbl

[8] Camm et al, A.J. Guidelines for the management of atrial fibrillation : The task force for the management of atrial fibrillation of the European Society of Cardiology (ESC), European heart journal, Volume 31 (2010), pp. 2369-2429

[9] Datta, Somnath; Bandyopadhyay, Dipankar; Satten, Glen A Inverse Probability of Censoring Weighted U-statistics for Right-Censored Data with an Application to Testing Hypotheses, Scandinavian Journal of Statistics, Volume 37 (2010) no. 4, pp. 680-700 | Zbl

[10] Dartigues, J.F.; Gagnon, M.; Barberger-Gateau, P.; Letenneur, L.; Commenges, D.; Sauvel, C.; Michel, P.; Salamon, R. The Paquid epidemiological program on brain ageing, Neuroepidemiology, Volume 11 (1992) no. 1, pp. 14-18

[11] Fournier, Marie-Cécile; Dantan, Etienne; Blanche, Paul R 2 -curve in the dynamic prediction context, in preparation (2016)

[12] Freedman, Andrew N; Seminara, Daniela; Gail, Mitchell H; Hartge, Patricia; Colditz, Graham A; Ballard-Barbash, Rachel; Pfeiffer, Ruth M Cancer risk prediction models : a workshop on development, evaluation, and application, Journal of the National Cancer Institute, Volume 97 (2005) no. 10, pp. 715-723

[13] Gerds, Thomas A; Cai, Tianxi; Schumacher, Martin The performance of risk prediction models, Biometrical Journal, Volume 50 (2008) no. 4, pp. 457-479 | Zbl

[14] Geskus, Ronald B Cause-Specific Cumulative Incidence Estimation and the Fine and Gray Model Under Both Left Truncation and Right Censoring, Biometrics, Volume 67 (2011) no. 1, pp. 39-49 | Zbl

[15] Gill, Richard D. Lectures on survival analysis., Bakry, Dominique (ed.) et al., Lectures on probability theory. Ecole d’Eté de Probabilités de Saint-Flour XXII-1992. Summer School, 9th- 25th July, 1992, Saint-Flour, France. Berlin : Springer-Verlag. Lect. Notes Math. 1581, 115-241 , 1994 | Zbl

[16] Gail, Mitchell H; Pfeiffer, Ruth M On criteria for evaluating models of absolute risk, Biostatistics, Volume 6 (2005) no. 2, pp. 227-239 | Zbl

[17] Gneiting, Tilmann; Raftery, Adrian E Strictly proper scoring rules, prediction, and estimation, Journal of the American Statistical Association, Volume 102 (2007) no. 477, pp. 359-378 | Zbl

[18] Graf, E.; Schmoor, C.; Sauerbrei, W.; Schumacher, M. Assessment and comparison of prognostic classification schemes for survival data, Statistics in Medicine, Volume 18 (1999), pp. 2529-2545

[19] Hand, David J Evaluating diagnostic tests : the area under the ROC curve and the balance of errors, Statistics in Medicine, Volume 29 (2010) no. 14, pp. 1502-1510

[20] Hung, H.; Chiang, C.T. Estimation methods for time-dependent AUC models with survival data, Canadian Journal of Statistics, Volume 38 (2010) no. 1, pp. 8-26 | Zbl

[21] Hilden, Jørgen; Gerds, Thomas A A note on the evaluation of novel biomarkers : do not rely on integrated discrimination improvement and net reclassification index, Statistics in Medicine, Volume 33 (2014) no. 19, pp. 3405-3414

[22] Hilden, Jørgen Commentary : On NRI, IDI, and “good-looking” statistics with nothing underneath, Epidemiology, Volume 25 (2014) no. 2, pp. 265-267

[23] Huang, Ying; Sullivan Pepe, Margaret; Feng, Ziding Evaluating the predictiveness of a continuous marker, Biometrics, Volume 63 (2007) no. 4, pp. 1181-1188 | Zbl

[24] Hastie, Trevor; Tibshirani, Robert; Friedman, Jerome The Elements of Statistical Learning, Second edition, New York : Springer, 2009 | Zbl

[25] Kaplan, E.L.; Meier, P. Nonparametric estimation from incomplete observations, Journal of the American Statistical Association, Volume 53 (1958) no. 282, pp. 457-481 | Zbl

[26] Lin, D.Y.; Fleming, T.R.; Wei, L.J. Confidence bands for survival curves under the proportional hazards model, Biometrika, Volume 81 (1994) no. 1, pp. 73-81 | Zbl

[27] Lopez, Olivier Nonparametric estimation of the multivariate distribution function in a censored regression model with applications, Communications in Statistics-Theory and Methods, Volume 40 (2011) no. 15, pp. 2639-2660 | Zbl

[28] Martinussen, T.; Scheike, T.H. Dynamic regression models for survival data, Springer, 2006 | Zbl

[29] Nicolaie, MA; van Houwelingen, JC; de Witte, TM; Putter, H Dynamic Pseudo-Observations : A Robust Approach to Dynamic Prediction in Competing Risks, Biometrics, Volume 69 (2013) no. 4, pp. 1043-1052 | Zbl

[30] Parast, L.; Cheng, S-C; Cai, T. Landmark Prediction of Long-Term Survival Incorporing Short-Term Event Time Information, Journal of the American Statistical Association, Volume 107 (2012) no. 500, pp. 1492-1501 | Zbl

[31] Pepe, M.S. The statistical evaluation of medical tests for classification and prediction, Oxford University Press, USA, 2003 | Zbl

[32] Pepe, MS; Feng, Z; Gu, JW Comments on ‘Evaluating the added predictive ability of a new marker : From area under the ROC curve to reclassification and beyond’by MJ Pencina et al., Statistics in Medicine (DOI : 10.1002/sim. 2929), Statistics in Medicine, Volume 27 (2008) no. 2, pp. 173-181

[33] Pepe, Margaret S; Feng, Ziding; Huang, Ying; Longton, Gary; Prentice, Ross; Thompson, Ian M; Zheng, Yingye Integrating the predictiveness of a marker with its performance as a classifier, American Journal of Epidemiology, Volume 167 (2008) no. 3, pp. 362-368

[34] Pepe, M.; Janes, H. Methods for Evaluating Prediction Performance of Biomarkers and Tests, Risk Assessment and Evaluation of Predictions (Lee, M-L; Gail, G.; Cai, T.; Pfeiffer, R.; Gandy, A., eds.), Springer, 2013

[35] Proust-Lima, Cécile; Amieva, Hélène; Dartigues, Jean-François; Jacqmin-Gadda, Hélène Sensitivity of Four Psychometric Tests to Measure Cognitive Changes in Brain Aging-Population–based Studies, American Journal of Epidemiology, Volume 165 (2007) no. 3, pp. 344-350

[36] Proust-Lima, Cécile; Dartigues, Jean-François; Jacqmin-Gadda, Hélène Joint modeling of repeated multivariate cognitive measures and competing risks of dementia and death : a latent process and latent class approach, Statistics in Medicine (2015)

[37] Proust-Lima, Cécile; Séne, Mbéry; Taylor, Jeremy MG; Jacqmin-Gadda, Hélène Joint latent class models for longitudinal and time-to-event data : A review, Statistical Methods in Medical Research, Volume 23 (2014) no. 1, pp. 74-90

[38] Rizopoulos, Dimitris Joint Models for Longitudinal and Time-to-event Data : With Applications in R, Boca Raton : Chapman & Hall/CRC, 2012 | Zbl

[39] Satten, G.A.; Datta, S. The Kaplan-Meier estimator as an inverse-probability-of-censoring weighted average, The American Statistician, Volume 55 (2001) no. 3, pp. 207-210 | Zbl

[40] Steyerberg, E.W. Clinical prediction models : a practical approach to development, validation, and updating, Springer, 2009

[41] The 3C Study Group Vascular factors and risk of dementia : design of the Three-City Study and baseline characteristics of the study population, Neuroepidemiology, Volume 22 (2003), pp. 316-325

[42] Tian, Lu; Cai, Tianxi; Goetghebeur, Els; Wei, LJ Model evaluation based on the sampling distribution of estimated absolute prediction error, Biometrika, Volume 94 (2007) no. 2, pp. 297-311 | Zbl

[43] Tsiatis, Anastasios A; Davidian, Marie Joint modeling of longitudinal and time-to-event data : an overview, Statistica Sinica, Volume 14 (2004) no. 3, pp. 809-834 | Zbl

[44] Tjur, Tue Coefficients of determination in logistic regression models–A new proposal : The coefficient of discrimination, The American Statistician, Volume 63 (2009) no. 4, pp. 366-372 | Zbl

[45] Taylor, Jeremy MG; Park, Yongseok; Ankerst, Donna P; Proust-Lima, Cecile; Williams, Scott; Kestin, Larry; Bae, Kyoungwha; Pickles, Tom; Sandler, Howard Real-time individual predictions of prostate cancer recurrence using joint models, Biometrics, Volume 69 (2013) no. 1, pp. 206-213 | Zbl

[46] Tsiatis, A.A. Semiparametric theory and missing data, New York : Springer Verlag, 2006 | Zbl

[47] Uno, H.; Cai, T.; Tian, L.; Wei, L.J. Evaluating prediction rules for t-year survivors with censored regression models, Journal of the American Statistical Association, Volume 102 (2007) no. 478, pp. 527-537 | Zbl

[48] van Houwelingen, Hans C Dynamic prediction by landmarking in event history analysis, Scandinavian Journal of Statistics, Volume 34 (2007) no. 1, pp. 70-85 | Zbl

[49] Vickers, Andrew J; Cronin, Angel M Everything you always wanted to know about evaluating prediction models (but were too afraid to ask), Urology, Volume 76 (2010) no. 6, pp. 1298-1301

[50] van Houwelingen, H.C; Putter, H. Dynamic Prediction in Clinical Survival Analysis, CRC Press, 2012 | Zbl

[51] Vickers, Andrew J Decision analysis for the evaluation of diagnostic tests, prediction models, and molecular markers, The American Statistician, Volume 62 (2008) no. 4

[52] Viallon, Vivian; Latouche, Aurélien Discrimination measures for survival outcomes : connection between the AUC and the predictiveness curve, Biometrical Journal, Volume 53 (2011) no. 2, pp. 217-236 | Zbl

[53] Zheng, Yingye; Heagerty, Patrick J Partly conditional survival models for longitudinal data, Biometrics, Volume 61 (2005) no. 2, pp. 379-391 | Zbl