pith. sign in

arxiv: 2606.23871 · v1 · pith:ZRXY2VVRnew · submitted 2026-06-22 · 💻 cs.LG · q-bio.QM· stat.ME· stat.ML

Federated Survival Analysis in Healthcare: A Multi-Model Evaluation on Cross-Institutional Heterogeneous Breast Cancer Data

Natalia Moreno-Blasco , Anusha Ihalapathirana , Pekka Siirtola , Miguel Fernandez-de-Retana This is my paper

Pith reviewed 2026-06-26 08:46 UTC · model grok-4.3

classification 💻 cs.LG q-bio.QMstat.MEstat.ML
keywords federated learningsurvival analysisbreast cancerheterogeneous datahealthcare privacyCox proportional hazardsrandom survival forestDeepSurv
0
0 comments X

The pith

Federated learning lets survival models for breast cancer match or exceed centralized performance while keeping patient data local.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper tests whether federated learning can produce reliable time-to-event predictions when privacy rules block pooling patient records from multiple hospitals. It runs a head-to-head comparison of the Cox Proportional Hazards model, DeepSurv, and Random Survival Forest under centralized training, purely local training, and three federated strategies on a real multi-institution breast cancer dataset whose clients differ in distribution. Federated runs beat local training across the board and reach or surpass centralized results in several cases, with Random Survival Forest delivering the strongest combination of discrimination, calibration, and stability. Performance also varies with how diverse the client data are, and two of the federated optimizers prove more reliable than the third. The authors translate the findings into concrete guidelines that link data characteristics, privacy constraints, and resource limits to model and training choices.

Core claim

On a cross-institutional breast cancer cohort with naturally heterogeneous clients, federated learning with FedAvg or FedProx consistently outperforms local training for Cox, DeepSurv, and RSF models and approaches or exceeds centralized performance. RSF provides the best balance of discrimination, calibration, and robustness, while performance depends on client diversity and FedAdam is less stable.

What carries the argument

The systematic comparison of Cox Proportional Hazards, DeepSurv, and Random Survival Forest models across centralized, local, and federated (FedAvg, FedProx, FedAdam) training paradigms on heterogeneous distributed clients.

If this is right

  • Random Survival Forest offers the best overall balance when robustness across heterogeneous clients matters most.
  • FedAvg and FedProx are more stable and effective federated optimizers than FedAdam for the gradient-based survival models.
  • Model and training-paradigm selection should be guided by measured client diversity, privacy requirements, and the need for interpretability.
  • Federated survival analysis can produce usable clinical predictions from distributed data that would otherwise remain siloed.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The same evaluation design could be applied to other time-to-event outcomes such as disease progression in different cancer types to check whether RSF and FedAvg retain their advantage.
  • Hospitals facing high client heterogeneity might begin with local RSF and add federated steps only when performance gains justify the added coordination.
  • The observed dependence on client diversity suggests that future federated survival work may need client-specific weighting or clustering to maintain performance in more extreme distribution shifts.
  • The decision-oriented guidelines could serve as a starting point for institutional policies on when to adopt federated versus centralized survival modeling.

Load-bearing premise

The cross-institutional breast cancer cohort possesses naturally heterogeneous distributed clients whose variation is representative of real-world clinical data distributions.

What would settle it

Replication on another set of institutions where federated training fails to outperform local training across the same three models would falsify the central performance claim.

Figures

Figures reproduced from arXiv: 2606.23871 by Anusha Ihalapathirana, Miguel Fernandez-de-Retana, Natalia Moreno-Blasco, Pekka Siirtola.

Figure 1
Figure 1. Figure 1: Client-level performance variability across training paradigms for CoxPH under the five-client configuration. Results are reported as mean ± standard deviation across 10 independent runs for each client: (a) C-Index, (b) AUC, and (c) IBS. TABLE III MEAN DEEPSURV PERFORMANCE ACROSS TRAINING PARADIGMS AND NUMBER OF CLIENTS Training Paradigm Clients C-Index (↑) AUC (↑) IBS (↓) Local 5 0.618∗ 0.620 0.159 4 0.5… view at source ↗
Figure 2
Figure 2. Figure 2: Client-level performance variability across training paradigms for DeepSurv under the five-client configuration. Results are reported as mean ± standard deviation across 10 independent runs for each client: (a) C-Index, (b) AUC, and (c) IBS [PITH_FULL_IMAGE:figures/full_fig_p008_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Client-level performance variability across training paradigms for RSF under the five-client configuration (100 trees). Results are reported as mean ± standard deviation across 10 independent runs for each client: (a) C-Index, (b) AUC, and (c) IBS. work. By comparing the models in this scenario, it is possible to assess their efficiency in a real environment, as well as their robustness with respect to dat… view at source ↗
Figure 4
Figure 4. Figure 4: Client-level performance comparison of RSF, DeepSurv, and CoxPH under federated learning for the five-client configuration. Results are reported as mean ± standard deviation across 10 independent runs for each client: (a) C-Index, (b) AUC, and (c) IBS. TABLE V AVERAGE FEDERATED PERFORMANCE ACROSS MODELS AND NUMBER OF CLIENTS Model Clients C-Index (↑) AUC (↑) IBS (↓) CoxPH 5 0.611 0.600 0.155 4 0.594 0.581 … view at source ↗
read the original abstract

Survival analysis is central to clinical decision-making, yet reliable time-to-event models require large, diverse cohorts that are rarely available at a single institution, while privacy regulations restrict the centralization of patient data. Federated learning (FL) offers a privacy-preserving alternative by training shared models without exchanging raw data, but its effectiveness for survival modeling under realistic, heterogeneous conditions remains insufficiently understood. This paper presents a systematic, multi-model evaluation of federated survival analysis on a cross-institutional breast cancer cohort with naturally heterogeneous distributed clients. Three representative survival models, the Cox Proportional Hazards model, DeepSurv, and Random Survival Forest (RSF), are compared across centralized, local, and federated training, and three federated optimization strategies (FedAvg, FedProx, and FedAdam) are assessed for the gradient-based models. Results show that FL consistently outperforms local training and approaches, and occasionally exceeds, centralized performance, while RSF offers the best overall balance of discrimination, calibration, and robustness across heterogeneous clients. We further find that performance depends on the diversity of client distributions, and that FedAvg and FedProx are stronger and more stable than FedAdam. Based on these findings, we derive practical, decision-oriented guidelines mapping data, privacy, interpretability, and resource constraints to recommended model and training-paradigm choices for federated survival modeling in healthcare.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

1 major / 1 minor

Summary. The paper evaluates three survival models (Cox PH, DeepSurv, RSF) under centralized, local, and federated training (FedAvg, FedProx, FedAdam) on a cross-institutional breast cancer cohort. It claims FL consistently outperforms local training and approaches or exceeds centralized performance, RSF provides the best balance of discrimination/calibration/robustness, performance depends on client diversity, FedAvg/FedProx are more stable than FedAdam, and derives practical guidelines mapping constraints to model/training choices.

Significance. If the empirical comparisons hold under representative heterogeneity, the work supplies actionable evidence for privacy-preserving survival modeling in healthcare, clarifying when federated approaches are viable and identifying RSF as robust. It fills a gap in multi-model FL survival evaluations on real distributed clinical data.

major comments (1)
  1. [Abstract and §3] Abstract and §3 (Data/Methods): The headline claims that FL outperforms local training, approaches/exceeds centralized performance, and that RSF is most robust rest on the assertion of 'naturally heterogeneous distributed clients' representative of real-world clinical variation. No quantitative client-level diagnostics (per-client covariate means/variances, censoring fractions, event-time distributions, or inter-client divergence metrics) are reported to substantiate this representativeness. Without them, the stability findings may be specific to this split rather than general.
minor comments (1)
  1. [§4] Table captions and §4 (Results) could explicitly list the number of clients, total samples, and per-client event rates to allow readers to assess heterogeneity scale directly.

Simulated Author's Rebuttal

1 responses · 0 unresolved

We thank the referee for the constructive feedback on substantiating client heterogeneity. We address the major comment below and will incorporate the suggested diagnostics to strengthen the manuscript.

read point-by-point responses

  1. Referee: [Abstract and §3] Abstract and §3 (Data/Methods): The headline claims that FL outperforms local training, approaches/exceeds centralized performance, and that RSF is most robust rest on the assertion of 'naturally heterogeneous distributed clients' representative of real-world clinical variation. No quantitative client-level diagnostics (per-client covariate means/variances, censoring fractions, event-time distributions, or inter-client divergence metrics) are reported to substantiate this representativeness. Without them, the stability findings may be specific to this split rather than general.

    Authors: We agree that explicit quantitative client-level diagnostics are needed to support claims of representativeness and to allow readers to evaluate generalizability. The current manuscript describes the data sources and notes natural heterogeneity arising from distinct institutions but does not include the requested per-client statistics. In the revision we will add a dedicated subsection to §3 reporting: (i) per-client means and standard deviations for key covariates, (ii) censoring fractions, (iii) summary statistics of observed event times, and (iv) pairwise inter-client divergence measures (e.g., Wasserstein distance on normalized feature distributions and KL divergence on binned survival times). These additions will directly address the concern that stability results may be split-specific. revision: yes

Circularity Check

0 steps flagged

No circularity: empirical model comparison with no derivations

full rationale

The paper performs an empirical evaluation of existing survival models (Cox PH, DeepSurv, RSF) and federated optimization strategies (FedAvg, FedProx, FedAdam) on a cross-institutional dataset. No mathematical derivations, predictions from fitted parameters, or uniqueness theorems are claimed. All reported results are direct experimental outcomes (discrimination, calibration, robustness metrics) rather than quantities defined by construction from the paper's own inputs. Self-citations, if present, are not load-bearing for any central claim. The work is self-contained as a standard benchmark study.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

This is an empirical evaluation study that applies existing models and optimization strategies; no new free parameters, axioms, or invented entities are introduced.

pith-pipeline@v0.9.1-grok · 5808 in / 1062 out tokens · 18217 ms · 2026-06-26T08:46:45.050787+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

43 extracted references · 4 linked inside Pith

  1. [1]

    High-performance medicine: the convergence of human and artificial intelligence,

    E. J. Topol, “High-performance medicine: the convergence of human and artificial intelligence,”Nature medicine, vol. 25, pp. 44–56, 2019

    2019

  2. [2]

    Artificial intelligence index report 2025,

    N. Maslej, L. Fattorini, R. Perrault, Y . Gil, V . Parli, N. Kariuki, E. Capstick, A. Reuel, E. Brynjolfsson, J. Etchemendyet al., “Artificial intelligence index report 2025,”arXiv preprint arXiv:2504.07139, 2025

    arXiv 2025

  3. [3]

    Key challenges for delivering clinical impact with artificial intelligence,

    C. J. Kelly, A. Karthikesalingam, M. Suleyman, G. Corrado, and D. King, “Key challenges for delivering clinical impact with artificial intelligence,”BMC medicine, vol. 17, no. 1, p. 195, 2019. N. MORENO-BLASCOet al.: EVALUATION OF FEDERATED SURVIVAL ANAL YSIS ON HETEROGENEOUS BREAST CANCER DATA 13 TABLE IX PRACTICALGUIDELINES FORFEDERATEDSURVIVALMODELING ...

    2019

  4. [4]

    The future of digital health with federated learning,

    N. Rieke, J. Hancox, W. Li, F. Milletari, H. R. Roth, S. Albarqouni, S. Bakas, M. N. Galtier, B. A. Landman, K. Maier-Heinet al., “The future of digital health with federated learning,”NPJ digital medicine, vol. 3, no. 1, p. 119, 2020

    2020

  5. [5]

    [Online]

    European Commission, “Regulation (EU) 2016/679 of the European Parliament and of the Council of 27 April 2016 on the protection of natural persons with regard to the processing of personal data and on the free movement of such data, and repealing Directive 95/46/EC (General Data Protection Regulation),” 2016. [Online]. Available: https://gdpr-info.eu/

    2016

  6. [6]

    The Health Insurance Portability and Accountability Act of 1996 (HIPAA),

    U. D. of Health and H. Services, “The Health Insurance Portability and Accountability Act of 1996 (HIPAA),” 1996. [Online]. Available: https://www.hhs.gov/hipaa/index.html

    1996

  7. [7]

    Ethical machine learning in healthcare,

    I. Y . Chen, E. Pierson, S. Rose, S. Joshi, K. Ferryman, and M. Ghassemi, “Ethical machine learning in healthcare,”Annual review of biomedical data science, vol. 4, no. 1, pp. 123–144, 2021

    2021

  8. [8]

    Communication-efficient learning of deep networks from decentralized data,

    B. McMahan, E. Moore, D. Ramage, S. Hampson, and B. A. y Arcas, “Communication-efficient learning of deep networks from decentralized data,” inArtificial intelligence and statistics. Pmlr, 2017, pp. 1273– 1282

    2017

  9. [9]

    Federated learning for predicting clinical outcomes in patients with covid-19,

    I. Dayan, H. R. Roth, A. Zhong, A. Harouni, A. Gentili, A. Z. Abidin, A. Liu, A. B. Costa, B. J. Wood, C.-S. Tsaiet al., “Federated learning for predicting clinical outcomes in patients with covid-19,”Nature medicine, vol. 27, no. 10, pp. 1735–1743, 2021

    2021

  10. [10]

    Privacy-preserving feder- ated brain tumour segmentation,

    W. Li, F. Milletar `ı, D. Xu, N. Rieke, J. Hancox, W. Zhu, M. Baust, Y . Cheng, S. Ourselin, M. J. Cardosoet al., “Privacy-preserving feder- ated brain tumour segmentation,” inInternational workshop on machine learning in medical imaging. Springer, 2019, pp. 133–141

    2019

  11. [11]

    Advances and open problems in federated learning,

    P. Kairouz and H. B. McMahan, “Advances and open problems in federated learning,”Foundations and trends in machine learning, vol. 14, no. 1-2, pp. 1–210, 2021

    2021

  12. [12]

    On the Convergence of FedAvg on Non-IID Data,

    X. Li, K. Huang, W. Yang, S. Wang, and Z. Zhang, “On the Convergence of FedAvg on Non-IID Data,” inInternational Conference on Learning Representations (ICLR), 2020

    2020

  13. [13]

    Inverting gradients-how easy is it to break privacy in federated learning?

    J. Geiping, H. Bauermeister, H. Dr ¨oge, and M. Moeller, “Inverting gradients-how easy is it to break privacy in federated learning?”Ad- vances in neural information processing systems, vol. 33, pp. 16 937– 16 947, 2020

    2020

  14. [14]

    Deep leakage from gradients,

    L. Zhu, Z. Liu, and S. Han, “Deep leakage from gradients,”Advances in neural information processing systems, vol. 32, 2019

    2019

  15. [15]

    Differential privacy: Gradient leakage attacks in federated learning environments,

    M. Fernandez-de Retana, U. Zulaika, R. S ´anchez-Corcuera, and A. Almeida, “Differential privacy: Gradient leakage attacks in federated learning environments,”arXiv preprint arXiv:2510.23931, 2025

    arXiv 2025

  16. [16]

    Survival analysis part i: basic concepts and first analyses,

    T. G. Clark, M. J. Bradburn, S. B. Love, and D. G. Altman, “Survival analysis part i: basic concepts and first analyses,”British journal of cancer, vol. 89, no. 2, pp. 232–238, 2003

    2003

  17. [17]

    D. G. Kleinbaum and M. Klein,Survival analysis a self-learning text. Springer, 1996

    1996

  18. [18]

    Global cancer statistics 2020: Globocan estimates of incidence and mortality worldwide for 36 cancers in 185 countries,

    H. Sung, J. Ferlay, R. L. Siegel, M. Laversanne, I. Soerjomataram, A. Je- mal, and F. Bray, “Global cancer statistics 2020: Globocan estimates of incidence and mortality worldwide for 36 cancers in 185 countries,”CA: a cancer journal for clinicians, vol. 71, no. 3, pp. 209–249, 2021

    2020

  19. [19]

    Scaling survival analysis in healthcare with federated survival forests: A comparative study on heart failure and breast cancer genomics,

    A. Archetti, F. Ieva, and M. Matteucci, “Scaling survival analysis in healthcare with federated survival forests: A comparative study on heart failure and breast cancer genomics,”Future Generation Computer Systems, vol. 149, pp. 343–358, 2023

    2023

  20. [20]

    Federated survival analysis with discrete-time cox models,

    M. Andreux, A. Manoel, R. Menuet, C. Saillard, and C. Simpson, “Federated survival analysis with discrete-time cox models,”arXiv preprint arXiv:2006.08997, 2020

    arXiv 2006

  21. [21]

    Developing federated time-to-event scores using heterogeneous real-world survival data,

    S. Li, Z. Wang, Y . Shang, Q. Wu, C. Hong, Y . Ning, D. Miao, M. E. H. Ong, B. Chakraborty, and N. Liu, “Developing federated time-to-event scores using heterogeneous real-world survival data,”Computers in Biology and Medicine, vol. 197, p. 111084, 2025

    2025

  22. [22]

    Fedpseudo: Pseudo value-based deep learning models for federated survival analysis,

    M. M. Rahman and S. Purushotham, “Fedpseudo: Pseudo value-based deep learning models for federated survival analysis,”arXiv preprint arXiv:2207.05247, 2022

    arXiv 2022

  23. [23]

    Flicu: a federated learning workflow for intensive care unit mortality prediction,

    L. Mondrejevski, I. Miliou, A. Montanino, D. Pitts, J. Hollm ´en, and P. Papapetrou, “Flicu: a federated learning workflow for intensive care unit mortality prediction,” in2022 IEEE 35th International Symposium on Computer-Based Medical Systems (CBMS). IEEE, 2022, pp. 32–37

    2022

  24. [24]

    Collaborative federated learning for healthcare: Multi-modal covid-19 diagnosis at the edge,

    A. Qayyum, K. Ahmad, M. A. Ahsan, A. Al-Fuqaha, and J. Qadir, “Collaborative federated learning for healthcare: Multi-modal covid-19 diagnosis at the edge,”IEEE Open Journal of the Computer Society, vol. 3, pp. 172–184, 2022

    2022

  25. [25]

    Hierarchical federated learning based anomaly detection using digital twins for smart healthcare,

    D. Gupta, O. Kayode, S. Bhatt, M. Gupta, and A. S. Tosun, “Hierarchical federated learning based anomaly detection using digital twins for smart healthcare,” in2021 IEEE 7th international conference on collaboration and internet computing (CIC). IEEE, 2021, pp. 16–25

    2021

  26. [26]

    Metafed: Feder- ated learning among federations with cyclic knowledge distillation for personalized healthcare,

    Y . Chen, W. Lu, X. Qin, J. Wang, and X. Xie, “Metafed: Feder- ated learning among federations with cyclic knowledge distillation for personalized healthcare,”IEEE Transactions on Neural Networks and Learning Systems, vol. 35, no. 11, pp. 16 671–16 682, 2023

    2023

  27. [27]

    Flamby: Datasets and benchmarks for cross-silo federated learning in realistic healthcare settings,

    J. Ogier du Terrail, S.-S. Ayed, E. Cyffers, F. Grimberg, C. He, R. Loeb, P. Mangold, T. Marchand, O. Marfoq, E. Mushtaqet al., “Flamby: Datasets and benchmarks for cross-silo federated learning in realistic healthcare settings,”Advances in Neural Information Processing Systems, vol. 35, pp. 5315–5334, 2022

    2022

  28. [28]

    The cancer genome atlas pan-cancer analysis project,

    J. N. Weinstein, E. A. Collisson, G. B. Mills, K. R. Shaw, B. A. Ozenberger, K. Ellrott, I. Shmulevich, C. Sander, and J. M. Stuart, “The cancer genome atlas pan-cancer analysis project,”Nature genetics, vol. 45, no. 10, pp. 1113–1120, 2013

    2013

  29. [29]

    The nci genomic data commons,

    A. P. Heath, V . Ferretti, S. Agrawal, M. An, J. C. Angelakos, R. Arya, 14 A PREPRINT TABLE XI FEDERATEDPER-CLIENTPERFORMANCEACROSSDIFFERENTNUMBERS OFCLIENTS Client CoxPH (FedA vg) DeepSurv (FedA vg) RSF C-Index(↑)AUC(↑)IBS(↓)C-Index(↑)AUC(↑)IBS(↓)C-Index(↑)AUC(↑)IBS(↓) 5 Clients C0 0.712±0.139 0.669±0.137 0.169±0.023 0.804±0.055 0.763±0.054 0.162±0.011 0...

    2021

  30. [30]

    Regression models and life-tables,

    D. R. Cox, “Regression models and life-tables,”Journal of the royal statistical society: Series B (methodological), vol. 34, no. 2, pp. 187– 202, 1972

    1972

  31. [31]

    Deepsurv: personalized treatment recommender system using a cox proportional hazards deep neural network,

    J. L. Katzman, U. Shaham, A. Cloninger, J. Bates, T. Jiang, and Y . Kluger, “Deepsurv: personalized treatment recommender system using a cox proportional hazards deep neural network,”BMC medical research methodology, vol. 18, no. 1, p. 24, 2018

    2018

  32. [32]

    Deephit: A deep learning approach to survival analysis with competing risks,

    C. Lee, W. Zame, J. Yoon, and M. Van Der Schaar, “Deephit: A deep learning approach to survival analysis with competing risks,” in Proceedings of the AAAI conference on artificial intelligence, vol. 32, no. 1, 2018

    2018

  33. [33]

    Improved survival analysis by learning shared genomic information from pan-cancer data,

    S. Kim, K. Kim, J. Choe, I. Lee, and J. Kang, “Improved survival analysis by learning shared genomic information from pan-cancer data,” Bioinformatics, vol. 36, no. Supplement 1, pp. i389–i398, 2020

    2020

  34. [34]

    Random survival forests,

    H. Ishwaran, U. Kogalur, E. Blackstone, and M. Lauer, “Random survival forests,”Ann. Appl. Statist., vol. 2, no. 3, pp. 841–860, 2008. [Online]. Available: https://arXiv.org/abs/0811.1645v1

    Pith/arXiv arXiv 2008

  35. [35]

    Random forests,

    L. Breiman, “Random forests,”Machine learning, vol. 45, no. 1, pp. 5–32, 2001

    2001

  36. [36]

    Random survival forest algorithm for risk stratification and survival prediction in gastric neuroendocrine neoplasms,

    T. Liao, T. Su, Y . Lu, L. Huang, W.-Y . Wei, and L.-H. Feng, “Random survival forest algorithm for risk stratification and survival prediction in gastric neuroendocrine neoplasms,”Scientific Reports, vol. 14, no. 1, p. 26969, 2024

    2024

  37. [37]

    Federated optimization in heterogeneous networks,

    T. Li, A. K. Sahu, M. Zaheer, M. Sanjabi, A. Talwalkar, and V . Smith, “Federated optimization in heterogeneous networks,”Proceedings of Machine learning and systems, vol. 2, pp. 429–450, 2020

    2020

  38. [38]

    Adaptive federated optimization,

    S. Reddi, Z. Charles, M. Zaheer, Z. Garrett, K. Rush, J. Kone ˇcn`y, S. Kumar, and H. B. McMahan, “Adaptive federated optimization,”arXiv preprint arXiv:2003.00295, 2020

    Pith/arXiv arXiv 2003

  39. [39]

    Adam: A method for stochastic optimization,

    D. P. Kingma and J. Ba, “Adam: A method for stochastic optimization,” arXiv preprint arXiv:1412.6980, 2014

    Pith/arXiv arXiv 2014

  40. [40]

    Flower: A friendly federated learning research framework,

    D. J. Beutel, T. Topal, A. Mathur, X. Qiu, J. Fernandez-Marques, Y . Gao, L. Sani, K. H. Li, T. Parcollet, P. P. B. de Gusm ˜AG ¸ oet al., “Flower: A friendly federated learning research framework,”arXiv preprint arXiv:2007.14390, 2020

    Pith/arXiv arXiv 2007

  41. [41]

    Evaluating the Yield of Medical Tests,

    F. E. Harrell Jr., R. M. Califf, D. B. Pryor, K. L. Lee, and R. A. Rosati, “Evaluating the Yield of Medical Tests,”JAMA, vol. 247, no. 18, pp. 2543–2546, 1982

    1982

  42. [42]

    Time-dependent ROC curves for censored survival data and a diagnostic marker,

    P. J. Heagerty, T. Lumley, and M. S. Pepe, “Time-dependent ROC curves for censored survival data and a diagnostic marker,”Biometrics, vol. 56, no. 2, pp. 337–344, 2000

    2000

  43. [43]

    Assessment and comparison of prognostic classification schemes for survival data,

    E. Graf, C. Schmoor, W. Sauerbrei, and M. Schumacher, “Assessment and comparison of prognostic classification schemes for survival data,” Statistics in Medicine, vol. 18, no. 17-18, pp. 2529–2545, 1999

    1999