Pith. sign in

REVIEW 3 major objections 4 minor 39 references

Represent, Then Generate: Multimodal-Conditioned Time-Series Generation under Irregular Missingness

T0 review · 3 major / 4 minor · reviewed 2026-08-16 · deepseek-v4-flash

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

Pith's one-line read ReCoGen decouples condition encoding from generation to make synthetic physiological signals outrank six generators on all sixteen benchmark settings.

desk verdict A genuinely new two-stage generator with an honest evaluation section, but the 'informative surrogates' claim outruns the protocol's train-on-real, test-on-synthetic design. read the letter →

arxiv 2608.12592 v1 pith:5VCWECKV submitted 2026-08-12 cs.LG

Haochen Zhang , Jiaheng Guo , Yu-Chao Huang , Nicholas Knoz , Tianlong Chen This is my paper · ORCID
classification cs.LG
keywords conditionaltime-seriesgenerationmultimodalconditioningirregularmissingnessmaskedautoencodersflowmatchingphysiologicalsignalscontinuousglucosemonitoringarterialbloodpressure
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 sets out to show that conditional generation of a hard-to-measure physiological signal from co-recorded, irregularly missing multimodal signals fails when raw streams are fed straight into the generator, and that the fix is to give condition representation its own trained stage. ReCoGen does this by training one masked autoencoder per time-variant modality, forcing each to impute from context and output a compact token sequence, then freezing those encoders and training a flow-matching generator on top. On three clinical benchmarks the method beats six existing conditional generators on all sixteen dataset-task-metric settings, and on thirteen of those settings a classifier trained on real signals reads the generated signal as carrying at least as much label-relevant information as the real signal. The paper is explicit that this measures information transfer, not pointwise waveform fidelity, and treats the real-signal value as an approximate anchor rather than a ceiling. If the claim holds, routinely collected signals can stand in for invasive or unavailable ones without losing clinically relevant structure.

What carries the argument

The load-bearing mechanism is the conditioning path rather than the generative backbone. Each Stage-I encoder turns an irregular, partially observed stream into a latent sequence; a bank of learnable per-modality query tokens then cross-attends over that latent to produce a compact set of in-context condition tokens, and when an entire modality is missing a learnable missing-token sequence stands in. Static clinical descriptors are encoded into one or two tokens and injected twice, once as an in-context token and once as adaptive layer-norm (AdaLN) modulation, which scales and shifts normalized activations per condition, an arrangement the ablations show to be complementary. The target series is handled through an invertible delay embedding that maps a one-dimensional waveform to a square image, so the flow-matching denoiser, a vision-transformer backbone, can generate the signal as an image and read it back as a waveform; the training loss is applied only on valid non-padded pixels, and inference integrates the flow ODE from noise to data. This pipeline is what allows the generator to consume robust condition tokens instead of learning to encode raw streams and generate at the same time.

What would settle it

Take a held-out cohort where the real target waveform was recorded but withheld from the generator; compare generated and real waveforms pointwise (for example by correlation or mean absolute error) while also running the downstream probe. If downstream utility remains near the reported levels while pointwise fidelity is near chance, or if a downstream probe built from the conditioning covariates alone already matches ReCoGen's scores, then the utility numbers come from condition re-expression rather than target fidelity, and the surrogate-signal claim would collapse.

Watch

Extended reading notes

Core claim

ReCoGen's central claim is that decoupling multimodal condition representation from target generation solves the problem of conditional physiological time-series generation under irregular missingness. In Stage I, each time-variant conditioning modality is distilled by its own masked autoencoder, trained to reconstruct held-out observed steps from context, so the resulting token sequences tolerate sparse and irregular sampling. In Stage II, a flow-matching vision transformer generates the target series from frozen per-modality tokens plus static descriptors, with learnable cross-attention queries summarizing each modality and static features injected both as in-context tokens and as adaptive layer-norm modulation. The paper reports that this design achieves the best downstream utility on all sixteen evaluated settings across AI-READI, MIMIC-III and MIMIC-IV, beating six representative conditional generators and reaching or exceeding the utility measured on real signals in thirteen of those settings.

Load-bearing premise

The load-bearing premise is that a probe trained on real signals and scored on generated signals measures whether the synthetic waveform preserves label-relevant physiology; the paper itself notes that a high score could instead mean the generator re-expresses conditioning covariates in a form the probe reads more easily than the real waveform.

Editorial extensions

If this is right

  • Synthetic ABP waveforms generated from heart rate, respiratory rate, and SpO2 plus static clinical data can be read by a real-trained classifier as carrying sepsis, heart-failure, and mortality-relevant information at or above the level of real ABP on most tested settings.
  • Generated CGM traces from wearable modalities and static features preserve enough glycemic structure to reproduce the ordinal study-group separation seen in real signals, supporting cohort analysis when CGM was not collected.
  • The ablations attribute most of the gain to the conditioning path, so the design principle of training missingness-aware encoders separately and then freezing them is what future conditional generators would need to retain to match this result.
  • Because the pipeline emits a regular 24-hour, 5-minute target from irregularly sampled conditions, it directly addresses the sampling patterns of ICU charting and wearable sensors rather than assuming densely aligned covariates.

Reading between the lines

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

  • A direct stress test the paper does not run: encode the conditioning covariates directly into the probe input (no generated waveform) and measure the same downstream AUROC. If that already recovers most of ReCoGen's reported utility, the scores are condition re-expression rather than target fidelity, the ambiguity the paper itself flags.
  • The decoupling recipe is domain-agnostic: any setting with heterogeneous, irregularly missing co-observed context and a target that is expensive to measure, such as environmental sensor networks or financial tick streams, could use the same represent-then-generate split.
  • Because Stage-I encoders are frozen and per-modality, a deployment-time extension would be to add a new condition modality by training only its masked autoencoder and attaching a query-token bank, leaving the generator untouched; the paper does not test incremental-modality addition.
  • The thirteen-of-sixteen at-or-above-real results suggest generated signals could serve as training augmentation for downstream classifiers, but only for labels that are not leaked into the conditioning set; the paper already removes the outcome label from all generators, so this is a natural continuation.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

3 major / 4 minor

Summary. The paper proposes ReCoGen, a two-stage conditional physiological time-series generator. Stage I trains one masked autoencoder per conditioning modality to distill irregularly sampled, partially observed streams into missingness-tolerant token sequences; Stage II freezes these encoders and trains a flow-matching DiT-style generator that mixes the token sequences with static clinical features via cross-attention and AdaLN. The target signals are ABP on MIMIC-III/IV and CGM on AI-READI. Evaluation is by downstream utility: a 1D-CNN classifier trained on real signals is applied to each method's generated signals, with AUROC/AUPRC reported for sepsis, heart failure, mortality, and study_group classification. ReCoGen reports the best score on all sixteen (dataset, task, metric) settings versus six conditional baselines, and matches or exceeds the real-signal reference on thirteen of them. The paper includes ablations of the conditioning injection path, static-feature encoding, and Stage-I mask ratio, and a statistical significance appendix with Holm-corrected t-tests and a Wilcoxon signed-rank test.

Significance. If the central claim is taken as 'ReCoGen produces signals that carry label-relevant information recoverable by a real-trained probe,' the result is well supported and the design contribution is solid: the decoupling of condition representation from generation is clearly motivated and the ablations show it is the source of the gains. The paper is also exemplary in transparency, explicitly acknowledging the fidelity/re-expression conflation and the anchor-not-ceiling interpretation, and in shipping statistical rigor (Holm corrections, Wilcoxon tests, five-seed means). The weakness is that the abstract and conclusion state a stronger claim—'informative surrogates'—which implies the generated signal can replace the real one in downstream use. The reported protocol does not test that implication, and the paper's own caveats admit that condition re-expression can explain the scores. The missing train-on-synthetic/test-on-real direction is directly testable, so the gap is fixable within the manuscript's scope.

major comments (3)
  1. [Section 3.2 and Conclusion/Abstract] The evaluation protocol, a real-trained probe scored on synthetic signals, supports the claim that generated signals contain label-relevant information readable by a real-trained model, but it does not support the abstract's and conclusion's 'informative surrogates' claim. A surrogate for augmentation requires the converse direction: a predictor trained on synthetic data should transfer to real data, or at least a train-on-synthetic-plus-real/test-on-real experiment. The paper itself acknowledges in Section 3.2 that a high score can arise from re-expressing the conditions, and Table 2 shows that the static conditions alone already reach AUROC 0.656 on MIMIC-IV sepsis versus ReCoGen's 0.692, leaving room for most of the reported margin to be re-expression. As written, the headline claim is not settled by the evidence; either add the train-on-synthetic/test-on-real evaluation or restrict the claims of Section 5 and the abstract to what the protocol actually measures.
  2. [Appendix A (AI-READI static features)] The static descriptor on AI-READI is described as containing '6 numeric CGM-enhanced features,' but these features are never defined. If any of them are summaries derived from the CGM trace itself (mean glucose, time-in-range, glycemic variability, etc.), then the CGM-only rows of Table 1 are substantially explained by the generator re-expressing the target's own information through the static condition, and the study_group AUROC of 0.771 could be achievable without generating a novel CGM waveform. Please define these features explicitly in Appendix A, and if any are CGM-derived, remove them from the conditioning set or quantify how much of the AI-READI result is attributable to re-expression rather than to the synthesized signal.
  3. [Section 3.2 ('no clean target')] The paper justifies omitting all pointwise fidelity metrics by saying 'the setting offers no clean target,' but this is incomplete. The held-out test subjects do have observed target steps; these steps are simply not used as ground truth in the main protocol. On the union of observed target steps, paired fidelity measures (e.g., correlation, absolute error, or a distributional distance between generated and real windows conditioned on the same inputs) can be computed and would partially separate target fidelity from condition re-expression. Without any such metric, the 'necessary but not sufficient' caveat of Section 2.1 remains unquantified, and the thirteen-of-sixteen 'reaches or exceeds the real signal' result is left open to the re-expression interpretation that the authors themselves flag.
minor comments (4)
  1. [Section 3.1 (Fairness of the comparison)] The statement 'every method faces the same probe' is contradicted by the supplementary protocol (Appendix F, item iv), which reports that on AI-READI ReCoGen was scored by its own five-seed probe instance rather than the baseline probe instance; the two probe draws agree within 0.005 AUROC, but the main-text sentence should be qualified to reflect that AI-READI comparisons are across probe draws.
  2. [Appendix C (Bridge prototype)] Bridge is described as drawing a 'same-label real target window as its prototype.' This means Bridge receives label information through the exemplar, which is a form of label leakage even though the categorical label is disabled elsewhere. The paper acknowledges Bridge 'receives strictly more,' but it should also state explicitly that this prototype gives Bridge access to the downstream label at generation time, and discuss the implication for the fairness of the comparison.
  3. [Appendix F (limitations list)] The limitations in Appendix F—AUPRC not tested, probe seeds shared yet unpaired tests used, rounded statistics, AI-READI probe mismatch, and the untested Real-Valid* anchor—are commendably transparent, but several of them (especially the first two) should be summarized in the main text near Table 1 so that readers of the main paper are not misled about the strength of the significance claims.
  4. [Figure 5 caption] The caption notes the ρ=0.3 point need not reproduce the main table exactly because the sweep is a single end-to-end re-run; it would help to state whether the error bars reflect probe-seed variation only or also training-seed variation, since the latter is not otherwise reported.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity found: ReCoGen's reported downstream utilities are externally probed, label-conditioning is disabled, and the acknowledged re-expression caveat is an interpretation gap, not a constructional equivalence.

full rationale

ReCoGen's central empirical claim (best downstream utility on all sixteen settings, Table 1) is not circular: the generator is trained without the downstream labels (Appendix C: 'the categorical label is removed from every method... the true label never reaches the network'), and the probe is trained on real signals, so the reported utilities are not fit to the evaluation targets. The method reuses standard building blocks (masked autoencoding, flow matching, DiT/SiT backbones, delay embedding) cited to independent prior work rather than to the authors' own results; no uniqueness theorem or self-citation chain carries the argument. The main caveat, stated in Section 3.2, is that downstream utility conflates target fidelity with condition re-expression: 'a high score can arise in two ways: the synthesized target carries the same label-relevant physiology as the real one, or the generator re-expresses information already in the conditions in a form the probe reads easily.' Table 2 supports the seriousness of this concern (static conditions alone give AUROC 0.656 for MIMIC-IV sepsis; on AI-READI the static-only AUROC is 0.750), and Appendix A lists '6 numeric CGM-enhanced features' without defining them, leaving open the possibility that target-derived summaries enter the conditioning set. However, this is an evaluation-interpretation gap, not a constructional equivalence: the paper's numbers are exactly what they claim to be (train-on-real, test-on-synthetic utilities), the missing train-on-synthetic/test-on-real direction would further test the 'surrogate' framing, and no fitted parameter is renamed as a prediction and no equation reduces to its own input. The paper explicitly flags rather than conceals the confound, and the statistical tests in Appendix F are conducted against real-signal references and baselines, not against the paper's own fitted quantities.

Assumptions & free parameters 3 free parameters · 5 assumptions · 0 invented entities

The central claim depends on a handful of tuned hyperparameters (Pmean, mask ratio) and on several domain assumptions that the paper mostly states or acknowledges. No invented entities. The most consequential assumptions are that the downstream utility protocol measures useful label-relevant transfer, and that independent per-modality encoding is sufficient.

free parameters (3)
  • Flow-matching time schedule Pmean = 0.5 (AI-READI), 0.0 (MIMIC)
    Selected on validation utility in a per-cohort sweep (Appendix C). Controls the training-time distribution of flow-matching time steps and thus generation quality.
  • Stage-I mask ratio rho = 0.3
    Default mask ratio for the per-modality masked autoencoders; swept over {0, 0.1, 0.3, 0.5, 0.7} in Figure 5 and chosen as a consistently strong setting.
  • Per-modality query tokens n = 4
    Number of learnable query tokens per modality for cross-attention pooling (Table 5). Architecture choice, not swept in the paper.
assumptions (5)
  • standard math Flow matching with the x-prediction v-loss (Lipman et al., 2023; Liu et al., 2022; Li and He, 2026) is a valid generative framework for this setting.
    Stage II uses this formulation without modification; it is background theory from the generative modeling literature.
  • domain assumption The delay embedding D is invertible on the valid region, so generating in image space and mapping back loses no information.
    Section 2.4 states D is invertible on the valid region and the flow-matching loss is computed only on non-padded pixels, underpinning the image-domain generation.
  • domain assumption A probe classifier trained on real signals, applied to generated signals, measures label-relevant information transfer that is a meaningful proxy for generation utility.
    Section 3.1 and Equation 2.1 define generation quality via downstream risk; Section 3.2 acknowledges the conflation with condition re-expression.
  • domain assumption The conditioning covariates (co-recorded vitals and static features) carry sufficient information about the target signal and the downstream labels.
    Used throughout; the ablations in Table 2 show both condition groups contribute, but the method's premise is that the target can be generated from conditions.
  • ad hoc to paper Independent per-modality encoding is sufficient; cross-modal dependencies can be left to the generator.
    The paper states in the Conclusion that 'encoding each modality independently leaves cross-modal dependencies to the generator', explicitly a limitation and an assumption of the design.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Represent, Then Generate: Multimodal-Conditioned Time-Series Generation under Irregular Missingness." pith.science (2026). https://pith.science/paper/5VCWECKV

@misc{pith2026260812592,
  author       = {Pith},
  title        = {Pith review of: Represent, Then Generate: Multimodal-Conditioned Time-Series Generation under Irregular Missingness},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/5VCWECKV}},
  note         = {Machine review of arXiv:2608.12592}
}
read the original abstract

Continuous physiological time series underpin modern clinical monitoring, yet many of the most informative signals are invasive, expensive, or simply unavailable for a given patient. Conditional generation offers a remedy: an absent signal can be synthesized from co-recorded signals and routine clinical variables. Existing generators, however, are built around a single conditioning modality and degrade when forced to handle the heterogeneous, irregularly missing mix of time-variant signals and static covariates seen in practice. We propose ReCoGen (Represent Conditions, then Generate), a two-stage framework that decouples multimodal condition representation from target generation. Stage I trains one masked autoencoder per modality, distilling each time-variant condition into a compact and missingness-tolerant token sequence. Stage II trains a flow-matching generator that fuses these tokens with static conditions to synthesize the target signal. Across three physiological benchmarks, including continuous glucose monitoring on AI-READI and arterial blood pressure generation on MIMIC-III and MIMIC-IV, ReCoGen attains the best downstream utility on all sixteen (dataset, task, metric) settings, surpassing six representative conditional generators; on thirteen of them its utility also reaches or exceeds the utility measured on the real signal, a reference we read as an approximate anchor rather than a ceiling. Ablations trace the gains to the conditioning path: learnable cross-attention over the frozen per-modality encoders, and a dual token-plus-AdaLN route for the static conditions. ReCoGen thus turns routinely collected signals into informative surrogates for invasive or unavailable ones, a step toward less invasive, lower-cost continuous clinical monitoring.

Figures

Figures reproduced from arXiv: 2608.12592 by the authors.

Figure 1
Figure 1. Motivation and overview. (A) Existing conditional generators ingest multimodal physiolog￾ical conditions directly and fail on heterogeneous modalities with irregular missingness. (B) ReCo￾Gen first represents each condition modality with missingness-aware encoders, then generates the tar￾get with a conditional flow-matching model. Conditional time series generation methods have been studied for handling a wide range… view at source ↗
Figure 2
Figure 2. Overview of ReCoGen. Stage I trains one masked autoencoder per time-variant conditioning [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. Generated CGM reproduces the class structure of the real signal on AI-READI. Each curve is the 24-hour trace averaged over the held-out par￾ticipants of one study_group class; the vertical axis is mean glucose in mg/dL, shared across panels, and “spread” is the gap between the highest and lowest class mean. Bridge is the strongest baseline on this dataset. Dataset Task Full −TS −static Real MIMIC-III Sepsis 0.677 0.… view at source ↗
Figures from the paper (2 more)
Figure 4
Figure 4. Figure 4: Time-series conditioning mechanism: per-task downstream AUROC (bars: 5-seed mean; error [PITH_FULL_IMAGE:figures/full_fig_p008_4.png]
Figure 5
Figure 5. Figure 5: Stage-I mask-ratio sensitivity. Down￾stream AUROC against the stage-I mask ratio; each line is the mean over a dataset’s tasks (AI-READI: study_group; MIMIC-III/-IV: sepsis, HF, mortal￾ity), error bars the pooled 5-seed probe std. The sweep is a single end-to-end re-ru…

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

39 extracted references · 20 canonical work pages

  1. [1]

    Scientific Data , volume=

    Multitask learning and benchmarking with clinical time series data , author=. Scientific Data , volume=. 2019 , publisher=

    work page 2019

  2. [2]

    Critical Care Medicine , volume=

    An interpretable machine learning model for accurate prediction of sepsis in the ICU , author=. Critical Care Medicine , volume=. 2018 , publisher=

    work page 2018

  3. [3]

    Nature Medicine , volume=

    Early prediction of circulatory failure in the intensive care unit using machine learning , author=. Nature Medicine , volume=. 2020 , publisher=

    work page 2020

  4. [4]

    Anesthesiology , volume=

    Machine-learning algorithm to predict hypotension based on high-fidelity arterial pressure waveform analysis , author=. Anesthesiology , volume=

    work page

  5. [5]

    PLoS Biology , volume=

    Glucotypes reveal new patterns of glucose dysregulation , author=. PLoS Biology , volume=. 2018 , publisher=

    work page 2018

  6. [6]

    National Science Review , volume=

    A pretrained transformer model for decoding individual glucose dynamics from continuous glucose monitoring data , author=. National Science Review , volume=. 2025 , publisher=

    work page 2025

  7. [7]

    ICASSP 2023-2023 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP) , pages=

    Gluformer: Transformer-based personalized glucose forecasting with uncertainty quantification , author=. ICASSP 2023-2023 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP) , pages=. 2023 , organization=

    2023

  8. [8]

    2025 47th Annual International Conference of the IEEE Engineering in Medicine and Biology Society (EMBC) , pages=

    Attengluco: Multimodal transformer-based blood glucose forecasting on ai-readi dataset , author=. 2025 47th Annual International Conference of the IEEE Engineering in Medicine and Biology Society (EMBC) , pages=. 2025 , organization=

    2025

Show all 39 references
  1. [9]

    New England Journal of Medicine , volume=

    Circulatory shock , author=. New England Journal of Medicine , volume=. 2013 , publisher=

    2013

  2. [10]

    Intensive Care Medicine , volume=

    The relationship between ICU hypotension and in-hospital mortality and morbidity in septic patients , author=. Intensive Care Medicine , volume=. 2018 , publisher=

    2018

  3. [11]

    JMIR Medical Informatics , volume=

    Predicting intensive care unit length of stay and mortality using patient vital signs: machine learning model development and validation , author=. JMIR Medical Informatics , volume=. 2021 , publisher=

    2021

  4. [12]

    arXiv Preprint arXiv:2206.13676 , year=

    Tts-cgan: A transformer time-series conditional gan for biosignal data augmentation , author=. arXiv Preprint arXiv:2206.13676 , year=

    arXiv

  5. [13]

    arXiv Preprint arXiv:2403.02682 , year=

    Time weaver: A conditional time series generation model , author=. arXiv Preprint arXiv:2403.02682 , year=

  6. [14]

    arXiv Preprint arXiv:2303.04743 , year=

    Vector quantized time series generation with a bidirectional prior model , author=. arXiv Preprint arXiv:2303.04743 , year=

    arXiv

  7. [15]

    Advances in Neural Information Processing Systems , volume=

    Towards editing time series , author=. Advances in Neural Information Processing Systems , volume=

  8. [16]

    Proceedings of the ACM on Management of Data , volume=

    Wavestitch: Flexible and fast conditional time series generation with diffusion models , author=. Proceedings of the ACM on Management of Data , volume=. 2025 , publisher=

    2025

  9. [17]

    Patterns , volume=

    Diffusets: 12-lead ecg generation conditioned on clinical text reports and patient-specific information , author=. Patterns , volume=. 2025 , publisher=

    2025

  10. [18]

    Proceedings of the AAAI Conference on Artificial Intelligence , volume=

    Cardiogan: Attentive generative adversarial network with dual discriminators for synthesis of ecg from ppg , author=. Proceedings of the AAAI Conference on Artificial Intelligence , volume=

  11. [19]

    Forty-second International Conference on Machine Learning , year=

    VerbalTS: Generating Time Series from Texts , author=. Forty-second International Conference on Machine Learning , year=

  12. [20]

    arXiv Preprint arXiv:2505.02417 , year=

    T2s: High-resolution time series generation with text-to-series diffusion models , author=. arXiv Preprint arXiv:2505.02417 , year=

    arXiv

  13. [21]

    arXiv Preprint arXiv:2503.02445 , year=

    BRIDGE: Bootstrapping Text to Control Time-Series Generation via Multi-Agent Iterative Optimization and Diffusion Modeling , author=. arXiv Preprint arXiv:2503.02445 , year=

    arXiv

  14. [22]

    Advances in Neural Information Processing Systems , volume=

    Csdi: Conditional score-based diffusion models for probabilistic time series imputation , author=. Advances in Neural Information Processing Systems , volume=

  15. [23]

    arXiv Preprint arXiv:2403.01742 , year=

    Diffusion-ts: Interpretable diffusion for general time series generation , author=. arXiv Preprint arXiv:2403.01742 , year=

    arXiv

  16. [24]

    Advances in Neural Information Processing Systems , volume=

    Utilizing image transforms and diffusion models for generative modeling of short and long time series , author=. Advances in Neural Information Processing Systems , volume=

  17. [25]

    Nature Metabolism , volume=

    AI-READI: rethinking AI data collection, preparation and sharing in diabetes research and beyond , author=. Nature Metabolism , volume=. 2024 , publisher=

    2024

  18. [26]

    Scientific Data , volume=

    MIMIC-III, a freely accessible critical care database , author=. Scientific Data , volume=. 2016 , publisher=

    2016

  19. [27]

    Scientific Data , volume=

    MIMIC-IV, a freely accessible electronic health record dataset , author=. Scientific Data , volume=. 2023 , publisher=

    2023

  20. [28]

    Proceedings of the IEEE/CVF International Conference on Computer Vision , pages=

    Scalable diffusion models with transformers , author=. Proceedings of the IEEE/CVF International Conference on Computer Vision , pages=

  21. [29]

    Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pages=

    Back to basics: Let denoising generative models denoise , author=. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pages=

  22. [30]

    Nature Medicine , volume=

    High-performance medicine: the convergence of human and artificial intelligence , author=. Nature Medicine , volume=. 2019 , publisher=

    2019

  23. [31]

    Personalized Medicine , volume=

    Wearables and the medical revolution , author=. Personalized Medicine , volume=. 2018 , publisher=

    2018

  24. [32]

    Anesthesiology , volume=

    Surgical and patient risk factors for severe arterial line complications in adults , author=. Anesthesiology , volume=. 2016 , publisher=

    2016

  25. [33]

    Scientific Reports , volume=

    Recurrent neural networks for multivariate time series with missing values , author=. Scientific Reports , volume=. 2018 , publisher=

    2018

  26. [34]

    IEEE Transactions on Pattern Analysis and Machine Intelligence , volume=

    Multimodal machine learning: A survey and taxonomy , author=. IEEE Transactions on Pattern Analysis and Machine Intelligence , volume=. 2018 , publisher=

    2018

  27. [35]

    Nature Medicine , volume=

    Multimodal biomedical AI , author=. Nature Medicine , volume=. 2022 , publisher=

    2022

  28. [36]

    Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pages=

    Masked autoencoders are scalable vision learners , author=. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pages=

  29. [37]

    The eleventh international conference on learning representations , year=

    Flow matching for generative modeling , author=. The eleventh international conference on learning representations , year=

  30. [38]

    arXiv Preprint arXiv:2209.03003 , year=

    Flow straight and fast: Learning to generate and transfer data with rectified flow , author=. arXiv Preprint arXiv:2209.03003 , year=

    arXiv

  31. [39]

    arXiv Preprint arXiv:2603.04767 , year=

    ConTSG-Bench: A Unified Benchmark for Conditional Time Series Generation , author=. arXiv Preprint arXiv:2603.04767 , year=

Pith tools

Reviewed August 16, 2026 · model on record in the stance chip above.