Pith. sign in

REVIEW 4 major objections 7 minor 24 references

Predicting Survival of Hemodialysis Patients using Federated Learning

T0 review · 4 major / 7 minor · reviewed 2026-08-11 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read Federated learning matches or beats local survival models for most dialysis zones.

desk verdict A real-world FL survival application with an unverifiable core result: Table III is missing and FedSurf's tree-selection step may leak the test split. read the letter →

arxiv 2412.10919 v1 pith:JU57P3NE submitted 2024-12-14 cs.LG cs.AI

Abhiram Raju , Praneeth Vepakomma This is my paper · ORCID
classification cs.LGcs.AI
keywords federatedlearningsurvivalanalysishemodialysisconcordanceindexrandomforestsprivacy-preservingmachineIndiadialysisnetwork
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

The paper tries to show that federated learning (training a shared model by exchanging model updates, not patient data) can predict survival for hemodialysis patients as accurately as, and sometimes better than, models trained on each region's data alone. Using records from 24,052 patients across six zones of a nationwide Indian dialysis network, it compares four survival-analysis models in local and federated forms. The federated version of random survival forests wins the concordance-index comparison in four of the six zones, and every zone has at least one federated model that beats its local counterpart. The authors claim this is the first application of federated learning to a large real-life hemodialysis dataset. If the result holds, dialysis networks can improve or maintain predictive accuracy while keeping sensitive patient data within each center.

What carries the argument

The paper's central mechanism is the pairing of four survival-analysis models with straightforward federation rules. For Cox proportional hazards, the global model is built by averaging local beta coefficients weighted by each client's sample size. For the deep networks DeepSurv and Cox-nnet, Federated Averaging aggregates model weights across clients. For random survival forests, the global model is assembled by importance-based sampling of trees from each client's local ensemble, forming a federated survival forest. The evaluation metric is Harrell's concordance index, which measures how well predicted risk ranks actual survival times.

What would settle it

A re-analysis of the full 183,063-patient records with a pre-specified, documented exclusion rule that either eliminates the federated model's advantage or reverses the zone-level winners would overturn the central claim; an independent held-out validation cohort on which FedSurf's concordance index no longer beats local random survival forests would also falsify the claimed benefit.

Watch

Extended reading notes

Core claim

On the paper's own terms, the central discovery is that a federated survival model can be a strict performance improvement for most clients: across the six NephroPlus zones, the federated survival forest (FedSurf) produced the highest concordance index in the North, East, West, and Andhra Pradesh zones, while a locally trained random survival forest won in the South and a locally trained Cox model won in Bihar. Additionally, every one of the six clients had at least one federated model outperform its local counterpart. The paper interprets this as evidence that regional dialysis clients have an incentive to join a federated framework not only for privacy and exposure to more diverse data but also for improved accuracy.

Load-bearing premise

The patient inclusion criteria that shrank the cohort from 183,063 to 24,052 are never described; the study assumes they were fixed in advance and unrelated to outcomes.

Editorial extensions

If this is right

  • Regional dialysis authorities can adopt federated survival forests and expect concordance indices that match or exceed zone-local models for most regions, while patient records never leave the center.
  • Because four of six zones had a federated model as their single best scorer, accuracy incentives align with privacy incentives for a majority of clients.
  • Every zone had at least one federated model beat its local counterpart, so even clients whose best model is local can find a federated alternative with comparable or better ranking performance.
  • The client-level feature-importance differences reported for random survival forests imply that federated models absorb regional heterogeneity rather than forcing one national pattern on every center.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • A direct consequence the paper does not test: a centralized model trained on the pooled 24,052 records would quantify exactly how much accuracy is lost or gained for the privacy protection FL provides; without that baseline, 'comparable to local' and 'comparable to centralized' are different claims.
  • If the federated-survival advantage generalizes, the same zone-based setup could be used for live waitlist prioritization, where survival predictions influence transplant wait times; a prospective study would be needed to confirm.
  • The uneven zone sizes suggest the FedSurf advantage may depend on client heterogeneity; testing with balanced zone partitions or adding differential privacy noise would reveal whether the result survives stronger privacy guarantees.
  • Because the paper compares FL only against local models, not against a centralized model, the first federated-application claim is best read as 'privacy-compatible accuracy is viable,' not as 'FL exceeds central training.'
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

4 major / 7 minor

Summary. The paper applies federated learning (FL) to survival analysis for hemodialysis patients using data from NephroPlus, India's largest private dialysis network, with six geographic zones treated as clients. Four survival models (CoxPH, DeepSurv, Cox-nnet, Random Survival Forests) are trained locally and in federated variants (coefficient averaging, FedAvg, and a Federated Survival Forest, FedSurf), and compared by Harrell's concordance index on an 80/20 train/test split. The central claim is that federated survival models achieve comparable or better performance than locally trained models, with FedSurf being the best-performing model in four of six zones, and that this is the first application of FL to a large real-life hemodialysis dataset.

Significance. If the results are fully supported, the paper would provide a useful empirical demonstration that privacy-preserving federated training can match or exceed local survival models on a large, real-world, multi-center clinical dataset. The use of 183,063 patient records from 244 centers and the realistic client structure (six regional zones) are strengths, as is the comparison of four distinct model families with standard federated aggregation approaches. The study is directly relevant to applied ML for healthcare and to the operational question of whether dialysis networks should adopt FL. However, the significance is currently limited because the quantitative evidence for the central claim is incomplete: the actual C-index results are not visible in the manuscript, no uncertainty quantification or significance testing is reported, and a key methodological detail in the FedSurf tree-selection procedure could permit test-set leakage if not clarified.

major comments (4)
  1. [III.A] The inclusion criteria that reduced the cohort from 183,063 patients to 24,052 patients are never described. This is load-bearing because every subsequent client-level comparison depends on the representativeness and pre-specification of this subsample. The authors should state the exact eligibility criteria, the number of patients excluded at each step, and whether the criteria were defined before any outcome analysis. If the criteria were chosen post hoc or are correlated with survival, the entire comparison between local and federated models could be invalidated.
  2. [IV, Table III] The actual concordance index values are not present in the version under review: Table III is declared but its contents are missing. Since the paper's central claim—that federated models, especially FedSurf, match or outperform local models—rests entirely on this table, the quantitative basis for the claim cannot be assessed. In addition, even when the values are supplied, the authors should report error bars, confidence intervals, or significance tests (e.g., permutation or bootstrap tests on the C-index), because without such measures the observed differences may be within noise.
  3. [III.C.c] The FedSurf tree-selection step is not specified with respect to which data are used for the 'performance' sorting. Section III.C.c says each client's decision trees were 'sorted based on performance' and the top-performing trees were taken, but it does not state whether this performance is evaluated on the local training split or on the 20% held-out test split. If the test split is used at any point in selecting or weighting trees, the reported C-index is circular and the 4/6-zone advantage of FedSurf is unsupported. The authors must clarify that only training data are used for tree selection and provide the exact importance-weighted selection procedure, including how the number of trees per client is determined.
  4. [IV] The claim that 'each client has at least 1 Federated Learning model outperforming its corresponding local model' and the subsequent statement that FedSurf performs best in North, East, Andhra Pradesh, and West are not accompanied by the underlying per-client, per-model C-index values. Moreover, Table IV labels Andhra Pradesh as 'Federated & Local' while the text says FedSurf was the best model for Andhra Pradesh; these statements should be reconciled. The reader currently cannot verify any of the aggregate claims without the missing results table.
minor comments (7)
  1. [Abstract] The abstract contains informal language ('don't perform as well') that should be corrected to formal academic style.
  2. [Author block] The affiliation 'Massachusets Institute of Technology' is misspelled; it should be 'Massachusetts Institute of Technology'.
  3. [III.C.b] The text says weights were aggregated using 'equation (1)', but the federated averaging formula is equation (2). The equation reference should be corrected.
  4. [II.B] Equation (2) has a notational issue: 'N ∑ n_k' is awkwardly typeset and should be written as N = Σ_k n_k, with the aggregation formula clearly defined.
  5. [III.B] The phrase 'the baseline hazard λ0(t) has been parametrized by the cox-proportional hazards model' is inaccurate; in the standard Cox model the baseline hazard is left unspecified and only the log-risk function is parameterized. The text should be reworded.
  6. [IV] Figure 2 is referenced but not described in the text; the authors should explain what the bar charts show (e.g., exact C-index values) and why RSF/FedSurf were selected for this display.
  7. [V] The conclusion states that 'feature importances generated by RSF models differed from client-to-client' and that DeepSurv's weaker federated performance 'was likely due to the fact that the data was very heterogeneous', but no supporting results or analyses are provided; these statements should either be removed or supported with evidence.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity found; the federated-vs-local comparison is empirical and self-contained.

full rationale

The study's central claim is an empirical comparison of standard survival models (CoxPH, DeepSurv, Cox-nnet, RSF) in local and federated forms, evaluated by concordance index on a held-out 20% test split described in Section IV ('The testing data was obtained locally for each client through an 80/20 train/test split'). The federated CoxPH aggregation (Eq. 7) is a weighted average of local coefficients; the neural-network models use FedAvg; the RSF federation follows an external method [24]. None of these steps defines the target result in terms of itself or fits a parameter to the same data used for evaluation. The only step that could hide circularity is the FedSurf tree-selection sentence in Section III.C.c: 'Each client's decision trees were sorted based on performance. A central server determined the number of trees it required from each client based on the importance assigned to each of them. The top-performing trees were taken from each client's forest.' The paper does not name the dataset used for that 'performance' ranking. If that ranking used the same held-out test split later used for Table III, the subsequent C-index would be optimistically biased. However, the paper does not state this, and a circularity finding requires exhibiting the specific reduction rather than speculating about an ambiguous detail. Similarly, the unreported inclusion criteria in Section III.A ('24052 patients fulfilled our criteria') are a reproducibility and selection-bias concern, not a circularity. There are no load-bearing self-citations; all cited methods are external and standard. The C-index wins of FedSurf in 4/6 zones are empirical outputs, not consequences of model definitions or fitted inputs. Therefore the derivation chain is self-contained and no circular step is demonstrated.

Assumptions & free parameters 0 free parameters · 5 assumptions · 0 invented entities

The central claim relies on standard survival analysis assumptions (proportional hazards, non-informative censoring), a domain assumption that administrative zones represent realistic FL clients, an adopted heuristic for Cox model averaging, and an ad hoc assumption that the unreported patient selection criteria are unbiased. No invented entities, particles, or new theoretical constructs are introduced.

assumptions (5)
  • domain assumption Proportional hazards assumption for CoxPH, DeepSurv, and Cox-nnet.
    Stated in Section III.B; these models use the Cox partial likelihood, which requires the hazard ratio to be constant over time.
  • domain assumption Non-informative censoring.
    Implicit in the survival analysis setup; censored patients are assumed to have the same future risk as uncensored patients with the same covariates, which is required for unbiased C-index estimation.
  • domain assumption The six administrative zones simulate realistic federated learning clients.
    Section III.A partitions patients into North, South, East, West, Andhra Pradesh, and Bihar zones; the validity of the federated comparison depends on these partitions resembling real multi-center data distributions.
  • domain assumption Naive averaging of Cox regression coefficients produces a valid global model.
    Adopted from [23] in Section III.C; the paper provides no proof that averaged beta coefficients yield a valid proportional hazards model on heterogeneous data.
  • ad hoc to paper The unreported patient inclusion criteria are pre-specified and unbiased.
    Section III.A says 24,052 of 183,063 patients 'fulfilled our criteria' but does not define the criteria; if the selection was post hoc or outcome-dependent, the comparison is invalid.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Predicting Survival of Hemodialysis Patients using Federated Learning." pith.science (2026). https://pith.science/paper/JU57P3NE

@misc{pith2026241210919,
  author       = {Pith},
  title        = {Pith review of: Predicting Survival of Hemodialysis Patients using Federated Learning},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/JU57P3NE}},
  note         = {Machine review of arXiv:2412.10919}
}
read the original abstract

Hemodialysis patients who are on donor lists for kidney transplant may get misidentified, delaying their wait time. Thus, predicting their survival time is crucial for optimizing waiting lists and personalizing treatment plans. Predicting survival times for patients often requires large quantities of high quality but sensitive data. This data is siloed and since individual datasets are smaller and less diverse, locally trained survival models do not perform as well as centralized ones. Hence, we propose the use of Federated Learning in the context of predicting survival for hemodialysis patients. Federated Learning or FL can have comparatively better performances than local models while not sharing data between centers. However, despite the increased use of such technologies, the application of FL in survival and even more, dialysis patients remains sparse. This paper studies the performance of FL for data of hemodialysis patients from NephroPlus, the largest private network of dialysis centers in India.

Figures

Figures reproduced from arXiv: 2412.10919 by the authors.

Figure 1
Figure 1. Federated Learning Setup IV. RESULTS Each of these global federated model types (Federated Survival Forests, Federated DeepSurv etc.) was then evaluated for local client data to compare the performance of federated learning models against typical local model performance to determine whether or not it is beneficial to join the FL framework. The testing data was obtained locally for each client through an 80/20 train/… view at source ↗

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

24 extracted references · 14 canonical work pages

  1. [1]

    Bharati, J., & Jha, V. (2020). Global Dialysis Perspective: India. Kidney360, 1(10), 1143–1147. https://doi.org/10.34067/KID.0003982020

    work page doi:10.34067/kid.0003982020 2020

  2. [2]

    Ministry of Health & Family Welfare. (n.d.). Introduction of Pradhan Mantri National Dialysis Program (PMNDP). Retrieved August 4, 2024, from https://pmndp.mohfw.gov.in/en

    work page 2024

  3. [3]

    Survival analysis of patients on maintenance hemodialysis

    Chandrashekar, A.; Ramakrishnan, S.1; Rangarajan, D.1. Survival analysis of patients on maintenance hemodialysis. Indian Journal of Nephrology 24(4):p 206-213, Jul–Aug 2014. | DOI: 10.4103/0971-4065.132985

    work page arXiv 2014

  4. [4]

    S., & Belmonte, A

    Díez-Sanmartín, C., Cabezuelo, A. S., & Belmonte, A. A. (2023). A new approach to predicting mortality in dialysis patients using sociodemographic features based on artificial intelligence. Artificial Intelligence in Medicine, 136, 102478. https://doi.org/10.1016/j.artmed.2022.102478

    arXiv 2023

  5. [5]

    Data -driven, two -stage machine learning algorithm -based prediction scheme for assessing 1-year and 3-year mortality risk in chronic hemodialysis patients

    Lee (2023). Data -driven, two -stage machine learning algorithm -based prediction scheme for assessing 1-year and 3-year mortality risk in chronic hemodialysis patients. Scientific Reports, 13(1). https://doi.org/10.1038/s41598-023-48905-9

    work page doi:10.1038/s41598-023-48905-9 2023

  6. [6]

    M., Griffith, J

    Wagner, M., Ansell, D., Kent, D. M., Griffith, J. L., Naimark, D., Wanner, C., & Tangri, N. (2011). Predicting mortality in incident dialysis patients: An analysis of the United Kingdom Renal Registry. American Journal of Kidney Diseases, 57(6), 894 –902. https://doi.org/10.1053/j.ajkd.2010.12.023

    work page doi:10.1053/j.ajkd.2010.12.023 2011

  7. [7]

    Hockham, C., Ghosh, A., Agarwal, A., Shah, K., Woodward, M., & Jha, V. (2024). Centre -level variation in the survival of patients receiving haemodialysis in India: findings from a nationwide private haemodialysis network. The Lancet Regional Health - Southeast Asia, 100383. https://doi.org/10.1016/j.lansea.2024.10038

    arXiv 2024

  8. [8]

    B., Moore, E., Ramage, D., Hampson, S., & Arcas, B

    McMahan, H. B., Moore, E., Ramage, D., Hampson, S., & Arcas, B. a. Y. (2016, February 17). Communication -Efficient Learning of Deep Networks from Decentralized Data. arXiv.org. https://arxiv.org/abs/1602.05629

    arXiv 2016

Show all 24 references
  1. [9]

    Li, T., Sahu, A., Talwalkar, A., & Smith, V. (2019). Federated Learning: challenges, methods, and future directions. ResearchGate. https://www.researchgate.net/publication/335319008_Federated_Learnin g_Challenges_Methods_and_Future_Directions

    2019

  2. [10]

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

    2020 arXiv

  3. [11]

    Federated Cox Proportional Hazards Model with multicentric privacy -preserving LASSO feature selection for survival analysis from the perspective of personalized medicine,

    C. Masciocchi, "Federated Cox Proportional Hazards Model with multicentric privacy -preserving LASSO feature selection for survival analysis from the perspective of personalized medicine," 2022 IEEE 35th International Symposium on Computer-Based Medical Systems (CBMS), Shenzen...

    2022

  4. [12]

    VERTICOX: Vertically Distributed Cox Proportional Hazards Model Using the Alternating Direction Method of Multipliers,

    W. Dai, X. Jiang, L. Bonomi, Y. Li, H. Xiong, and L. OhnoMachado, “VERTICOX: Vertically Distributed Cox Proportional Hazards Model Using the Alternating Direction Method of Multipliers,” IEEE Transactions on Knowledge and Data Engineering, vol. 34, no. 2, pp. 996– 1010, Feb. 2...

    2022

  5. [13]

    R., Price, G., Field, M., Sarup, N., Zukauskaite, R., Johansen, J., Eriksen, J

    Hansen, C. R., Price, G., Field, M., Sarup, N., Zukauskaite, R., Johansen, J., Eriksen, J. G., Aly, F., McPartlin, A., Holloway, L., Thwaites, D., & Brink, C. (2022). Larynx cancer survival model developed through open- source federated learning. Radiothera py and Oncology, 17...

    2022 doi

  6. [14]

    M., & Purushotham, S

    Rahman, M. M., & Purushotham, S. (2022). Fedpseudo: Pseudo value - based deep learning models for federated survival analysis. arXiv preprint arXiv:2207.05247

    2022 arXiv

  7. [15]

    Federated Learning Approach for Breast Cancer Detection Based on DCNN,

    H. AlSalman, M. S. Al-Rakhami, T. Alfakih and M. M. Hassan, "Federated Learning Approach for Breast Cancer Detection Based on DCNN," in IEEE Access, vol. 12, pp. 40114 -40138, 2024, doi: 10.1109/ACCESS.2024.3374650

    2024

  8. [16]

    K., Akhtar, S

    Jha, V., Ur -Rashid, H., Agarwal, S. K., Akhtar, S. F., Kafle, R. K., & Sheriff, R. (2019). The state of nephrology in South Asia. Kidney international, 95(1), 31-37

    2019

  9. [17]

    M., & Grambsch, P

    Therneau, T. M., & Grambsch, P. M. (2000). The Cox model. In Statistics in the health sciences (pp. 39 –77). https://doi.org/10.1007/978-1-4757- 3294-8_3

    2000 doi

  10. [18]

    & General Electric Research and Development Center

    Nelson, W. & General Electric Research and Development Center. (1972). Theory and applications of hazard plotting for censored failure data. In TECHNOMETRICS (Vol. 14, Issue 4). https://www.stat.cmu.edu/technometrics/70-79/VOL-14-04/v1404945.pdf

    1972

  11. [19]

    Breslow, N.E. (1974). Covariance analysis of censored survival data. Biometrics, 30 1, 89-99

    1974

  12. [20]

    L., Shaham, U., Cloninger, A., Bates, J., Jiang, T., & Kluger, Y

    Katzman, J. L., Shaham, U., Cloninger, A., Bates, J., Jiang, T., & Kluger, Y. (2018). DeepSurv: personalized treatment recommender system using a Cox proportional hazards deep neural network. BMC Medical Research Methodology, 18(1). https://doi.org/10.1186/s12874-018-0482-1

    2018 doi

  13. [21]

    Wang, D., Jing, Z., He, K., & Garmire, L. X. (2021). Cox -nnet v2.0: improved neural -network-based survival prediction extended to large - scale EMR data. Bioinformatics (Oxford, England), 37(17), 2772 –2774. https://doi.org/10.1093/bioinformatics/btab046

    2021 doi

  14. [22]

    Random survival forests

    Hemant Ishwaran. Udaya B. Kogalur. Eugene H. Blackstone. Michael S. Lauer. "Random survival forests." Ann. Appl. Stat. 2 (3) 841 - 860, September 2008. https://doi.org/10.1214/08-AOAS169

    2008 doi

  15. [23]

    K., & Tripathy, A

    Seidi, N., Roy, S., Das, S. K., & Tripathy, A. (n.d.). Addressing data heterogeneity in federated learning of COX Proportional Hazards models. NASA/ADS. https://doi.org/10.48550/arXiv.2407.14960

    doi arXiv

  16. [24]

    Archetti, A., DEIB, Matteucci, M., & DEIB. (2023). Federated Survival Forests [Journal-article]. arXiv. https://arxiv.org/pdf/2302.02807

    2023 arXiv

Pith tools

Reviewed August 11, 2026 · model on record in the stance chip above.