Pith. sign in

REVIEW 3 major objections 5 minor 37 references

A Multimodal In Vitro Diagnostic Method for Parkinson's Disease Combining Facial Expressions and Behavioral Gait Data

T0 review · 3 major / 5 minor · reviewed 2026-08-15 · deepseek-v4-flash

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

Pith's one-line read A multimodal fusion of facial-expression and gait-video features is claimed to achieve perfect (1.0) accuracy in diagnosing Parkinson's disease on the authors' PDMM test set.

desk verdict A real dataset and plausible multimodal pipeline undercut by a likely identity-leakage flaw in the synthetic facial-expression training. read the letter →

arxiv 2506.17596 v1 pith:P5FCGPNV submitted 2025-06-21 cs.CV

Wei Huang , Yinxuan Xu , Yintao Zhou , Zhengyu Li , Jing Huang , Meng Pang This is my paper · ORCID
classification cs.CV
keywords Parkinson'sdiseasediagnosismultimodalfusionfacialexpressionsynthesisgaitanalysisStyleGANhybridPDMMdatasetdeeplearning
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

This paper argues that a multimodal diagnostic method that fuses facial-expression features with behavioral gait features can diagnose Parkinson's disease more accurately than either modality alone. The authors build the largest known multimodal PD dataset, PDMM, with 95 patients, and extract gait features from HRNet keypoints with STGCN++ and facial features with MobileNetV3 trained on StyleGAN-generated pre-morbid expressions. A 'hybrid fusion' layer, which stacks modality scores with feature vectors, yields a reported test accuracy of 1.0 on a test set of 19 PD patients and 47 controls, outperforming unimodal and prior methods. If this holds, it would make low-cost, camera-based PD screening feasible in settings without specialized equipment.

What carries the argument

The central object is the 'hybrid fusion' layer, a two-stage stacking design: each modality's feature vector passes through a fully connected layer to produce a score; that score is concatenated with its feature vector; the two enriched vectors go through another fully connected layer; and the two outputs are summed for the final diagnosis. The gait path relies on HRNet for skeletal keypoints and STGCN++, a spatio-temporal graph convolutional network, for gait features. The facial path relies on StyleGAN with an inversion encoder (Image2StyleGAN) to compute expression direction vectors via logistic regression in latent space, so that neutral photos can be re-rendered in six emotions while keeping identity. This machinery does the work of generating a 'pre-morbid' control group for each patient and of letting the fusion layer balance the two signals.

What would settle it

Re-run the entire pipeline with an explicit identity-disjoint split, generating synthetic expressions only from training-fold neutral photos and testing only on never-synthesized patients; then compare the resulting accuracy to 1.0.

Watch

Extended reading notes

Core claim

The central claim is that fusing the two modalities is not merely additive but qualitatively better: a diagnostic accuracy of exactly 1.0 on the PDMM test set, against 0.9755 for the best previous facial-expression method (FEPD) and 0.9692 for each unimodal branch of the proposed pipeline. The authors attribute the gain to two ingredients: StyleGAN synthesis of six basic expressions from a single neutral photo, which approximates the patient's pre-morbid face and supplies enough training signal for a lightweight facial feature extractor, and the hybrid fusion layer, which mimics ensemble stacking by appending per-modality scores to feature vectors before a final classification layer. Their experiments also show that the synthesized expressions preserve identity and quality better than StarGAN and CycleGAN, and that the gait-only and facial-only branches both fall short of the fused result.

Load-bearing premise

The paper does not confirm that the synthetic expressions used for training were made only from the patients who were also in the training group; if synthetic faces of test patients were used during training, the perfect test score is not a fair measurement.

Editorial extensions

If this is right

  • If the accuracy claim is reproducible, PD screening could run on ordinary smartphone cameras, since the feature extractors are lightweight (MobileNetV3 is 5.93 MB) and the pipeline needs no specialized sensors.
  • The StyleGAN direction-vector approach to synthesizing a pre-morbid baseline could transfer to other motor symptoms (e.g., hypomimia in other neurodegenerative diseases), as a general data-augmentation trick for small clinical cohorts.
  • The PDMM dataset, claimed as the largest multimodal PD dataset, becomes a common benchmark where future unimodal and multimodal in-vitro methods can be compared on the same patients.
  • The fusion architecture suggests that appending soft scores to feature vectors—a cheap form of stacking—can improve multimodal medical diagnosis even when each branch is moderately strong.

Reading between the lines

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

  • I would not take 1.0 as a trustworthy generalization estimate: the paper never states that the StyleGAN synthesis of a patient's six emotions was restricted to training folds, so synthetic versions of test identities may have been seen by the feature extractor; a clean identity-disjoint split would likely lower the number.
  • The gait-only accuracy (0.9692) is close to the multimodal result, so the fusion gain (about 0.03) may be small in absolute terms; an ablation that varies the quality of one modality while holding the other fixed would separate the fusion benefit from the strength of the gait branch.
  • The method's reliance on a single neutral photo plus direction vectors assumes the neutral photo itself carries PD-relevant identity information; this assumption could be tested by repeating the diagnosis with synthetic expressions generated from a different neutral pose or lighting condition.
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 / 5 minor

Summary. The paper proposes a multimodal in vitro diagnostic method for Parkinson's disease that fuses facial-expression features and behavioral gait features. The authors introduce a new PDMM dataset of 95 PD patients, extract gait features with STGCN++ from HRNet skeleton keypoints, and generate six synthetic emotion images from each patient's neutral photo using a StyleGAN-based pipeline; these synthetic images are used as a control group to train a MobileNetV3 facial-expression feature extractor. A hybrid fusion layer then combines the two feature streams for PD diagnosis. The paper reports a diagnostic accuracy of 1.0 on a test set of 19 held-out PD patients and 47 healthy controls assembled from Tsinghua-FED and CASIA-B-HRNet, and reports unimodal ablations with accuracies around 0.96.

Significance. If the reported results were trustworthy, the claimed performance would be a substantial advance for non-invasive, video-based PD screening, and the PDMM dataset could be a useful community resource. The paper also has positive aspects: it collects a new hospital-collaboration dataset, evaluates several lightweight backbones, and proposes an explicit fusion scheme. However, the empirical evaluation as reported does not support the central claims: the synthetic-data generation is not shown to respect the train/test identity split, the healthy controls are constructed by pairing faces and gait from two unrelated public datasets, and the 'pre-morbid' control group is generated by the method itself without validation. These issues are load-bearing for the reported accuracy.

major comments (3)
  1. [Feature Extraction From Facial Expressions, Step 3; PD Diagnosis Evaluation] The paper does not establish that the synthetic facial-expression training data were generated only from the 76 training-fold PD patients. Stage 2 states that six emotion images are generated from a single neutral facial image of a PD patient and that these images augment the training dataset, and the PD Diagnosis Evaluation then tests on the 19 patients in the held-out fold. If synthetic images of those 19 identities were included in the training of MobileNetV3, the test-time features already encode those identities, so the perfect accuracy in Table 2 would reflect identity memorization rather than generalized PD diagnosis. No fold assignment, code, or data are provided to rule this out.
  2. [PD Diagnosis Evaluation] The 47 non-PD controls are formed by pairing facial expression images of 47 subjects from Tsinghua-FED with 47 gait clips from CASIA-B-HRNet. Because these are independent public datasets, each control 'subject' has a face from one person and a gait sequence from another person; the multimodal fusion model is therefore evaluated on mismatched cross-subject pairs for the control class. Additionally, the acquisition conditions differ between the PDMM hospital recordings and both public datasets (camera, lighting, age distribution), so the classifiers in Tables 2 and 3 may separate datasets rather than disease status. This makes the reported diagnostic accuracies uninterpretable as clinical performance.
  3. [Feature Extraction From Facial Expressions, Step 3; PD Diagnosis Evaluation] The synthetic six-emotion images are described as forming a 'control group' for PD facial-expression training, but they are generated by the method's own StyleGAN pipeline from current neutral photos of the same patients, not measured pre-morbid data. The paper does not specify how negative labels are assigned during multimodal fusion training; if the negative class consists of these synthetic pre-morbid images of the 76 training patients, the fusion layer is trained to separate a patient's current expression from a synthetic version of that same patient, which is circular. Even if the split issue were fixed, the claim that the model learns pre-morbid expression differences would remain unsupported because there is no validation that the synthetic expressions resemble true pre-morbid expressions.
minor comments (5)
  1. [PD Diagnosis Evaluation] There is a typo in 'testinig set' in the paragraph describing the test set composition.
  2. [Synthetic Facial Expression Evaluation] The comparison of the proposed synthesis with StarGAN and CycleGAN in Figure 4 is qualitative; no quantitative metrics (e.g., FID, identity preservation, emotion classification accuracy on generated images) are reported.
  3. [Facial Expression Feature Extraction Evaluation] The description of how generated PD images are combined with Tsinghua-FED for the expression-classification experiment is ambiguous; the paper should state the exact composition of the training and test partitions.
  4. [PD Diagnosis Evaluation] Only a single accuracy number is reported in Table 2 with no confidence interval or significance test; with 66 test subjects, the precision of the 1.0 estimate should be quantified.
  5. [Abstract; Experimental Setup] The term 'in vitro diagnosis' is used to mean non-invasive video-based biomarker analysis; this is not standard medical usage and may confuse readers.

Circularity Check

2 steps flagged · score 8.0 of 10

Reported 1.0 PD accuracy is not an independent test: synthetic facial-expression training includes held-out PD test identities, so the evaluation can measure identity memorization rather than diagnosis.

  1. fitted input called prediction [Feature Extraction From Facial Expressions (Step 3) and PD Diagnosis Evaluation]
    "Using the aforementioned facial expression synthesis method, we generate images of six basic emotions (anger, disgust, fear, sadness, happiness, surprise) from a single neutral facial image of a PD patient, capturing pre-disease expressions. These images form a control group and augment our training dataset for Parkinson's disease facial expressions. ... we follow the evaluation protocol in (W. Huang et al., 2023; Zhou et al., 2024), and divide the PDMM dataset into 5 folds, with 4 folds (76 PD patients) utilized for training and the remaining 1 fold (19 PD patients) used for testing."

    The synthetic six-emotion images are generated from every PD patient's neutral photo in Stage 2, before any diagnostic split is described, and the text never limits generation to the four training folds. The PD Diagnosis Evaluation then uses the remaining 19 patients as the positive test fold. Because the frozen facial feature extractor (MobileNetV3) was trained with synthetic versions of those same identities, the held-out patients are not unseen. The fusion layer sees facial features that already encode test identities, so the reported 1.0 accuracy can be achieved by identity memorization rather than by generalized PD diagnosis. The prediction is thus fitted to the test identities by construction.

  2. self definitional [Introduction and Feature Extraction From Facial Expressions (Step 3)]
    "we generate facial expression images depicting other six basic emotions from a single neutral facial expression image of PD patients, thereby simulating their pre-morbid facial expression state as a reference group to train a discriminative model"

    The 'pre-morbid' control group is not an independent reference: it is defined as StyleGAN transformations of the patient's own current neutral face. The discriminative model is therefore trained to separate expression classes using features that are derived from, and entangled with, the patient's identity. 'Pre-morbid expression' is defined as a function of the patient's diseased-state image, so the diagnostic feature is not an independent premorbid measurement. The claim that the method detects PD from premorbid facial behavior reduces to the fact that the model has seen the patient's face, or synthetic variants of it, during training.

full rationale

The central claim of Table 2 (accuracy 1.0 on 66 subjects) is not independently established because the facial feature extractor is trained on synthetic images generated from the same patients later used as the positive test fold. The paper's own description places the StyleGAN generation (Stage 2) before and outside the 5-fold PDMM split described in PD Diagnosis Evaluation, and no sentence restricts generation to the four training folds. The frozen MobileNetV3 features then carry identity information about the 19 test patients into the fusion layer, so the 'prediction' is not a prediction about unseen identities. The self-citations to prior work by the same group (W. Huang et al. 2023; Zhou et al. 2024) for the evaluation protocol and baselines are not the source of the problem; the problem is the by-construction overlap between synthetic training data and test identities. Score 8: the central accuracy claim is invalidated by this leakage; it is not a fully definitional derivation (score 10) because gait features and the fusion training on the training folds add some independent content, but the facial modality alone collapses the evaluation.

Assumptions & free parameters 5 free parameters · 5 assumptions · 1 invented entities

The ledger shows that the method relies on a stack of pretrained models and on the assumption that StyleGAN-generated images can stand in for real pre-morbid expressions. The main free hyperparameters (lambda_mse, lambda_j, k, expression strength) are set by hand or by citing earlier work, and the direction vector is fitted. No independent evidence supports the synthetic pre-morbid images as a faithful control group.

free parameters (5)
  • lambda_mse = 1
    MSE loss weight in Eq. 1, set empirically following Abdal et al. 2019.
  • lambda_j = 1
    Perceptual-layer weight in Eq. 2, set empirically.
  • k = 4
    Number of VGG-16 layers used in perceptual loss, set empirically.
  • lambda (expression strength) = not reported
    Controls the degree of expression change in Eq. 3; no value is given, so results depend on an unspecified choice.
  • direction vector n_AB = fitted via logistic regression on StyleGAN latent codes
    The expression transition vector is learned from latent vectors; it is a fitted quantity, not derived from first principles.
assumptions (5)
  • domain assumption A pre-trained StyleGAN generator can encode facial identity and expression into a latent space where linear arithmetic works.
    Used in Steps 1-3; no evidence is provided that this holds for PD patients' faces.
  • domain assumption The logistic-regression direction vector n_AB linearly interpolates between facial expressions while preserving identity.
    Algorithm 1 and Eq. 3; the generated images are not validated against real pre-morbid photos.
  • ad hoc to paper Six synthetic emotions generated from a neutral PD photo represent the patient's true pre-morbid expressions.
    The entire facial feature training relies on this simulation, which has no independent ground truth.
  • domain assumption Tsinghua-FED and CASIA-B-HRNet healthy subjects are comparable to PDMM patients for PD diagnosis.
    Controls are borrowed from different datasets with different cameras, ages, and settings; the paper does not match acquisition conditions.
  • domain assumption HRNet keypoints and STGCN++ features capture gait abnormalities specific to PD rather than aging.
    Used in gait feature extraction; no control analysis separates PD gait from age-related gait changes.
invented entities (1)
  • Synthetic pre-morbid facial expression images
    purpose: Serve as a healthy/reference training set for the facial-expression feature extractor.
    These images are generated by the paper's own StyleGAN pipeline and are not validated against any real pre-morbid photographs of PD patients, so they provide no independent falsifiable evidence.

how reviews work

0 comments
Cite this review

Pith. "Pith review of A Multimodal In Vitro Diagnostic Method for Parkinson's Disease Combining Facial Expressions and Behavioral Gait Data." pith.science (2026). https://pith.science/paper/P5FCGPNV

@misc{pith2026250617596,
  author       = {Pith},
  title        = {Pith review of: A Multimodal In Vitro Diagnostic Method for Parkinson's Disease Combining Facial Expressions and Behavioral Gait Data},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/P5FCGPNV}},
  note         = {Machine review of arXiv:2506.17596}
}
read the original abstract

Parkinson's disease (PD), characterized by its incurable nature, rapid progression, and severe disability, poses significant challenges to the lives of patients and their families. Given the aging population, the need for early detection of PD is increasing. In vitro diagnosis has garnered attention due to its non-invasive nature and low cost. However, existing methods present several challenges: 1) limited training data for facial expression diagnosis; 2) specialized equipment and acquisition environments required for gait diagnosis, resulting in poor generalizability; 3) the risk of misdiagnosis or missed diagnosis when relying on a single modality. To address these issues, we propose a novel multimodal in vitro diagnostic method for PD, leveraging facial expressions and behavioral gait. Our method employs a lightweight deep learning model for feature extraction and fusion, aimed at improving diagnostic accuracy and facilitating deployment on mobile devices. Furthermore, we have established the largest multimodal PD dataset in collaboration with a hospital and conducted extensive experiments to validate the effectiveness of our proposed method.

Figures

Figures reproduced from arXiv: 2506.17596 by the authors.

Figure 1
Figure 1. A comparison between the masked faces of PD pa [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. Overview of the proposed multimodal in vitro diagnostic method for PD. [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. The direction vector ˆnAB governs the transition from expression A to expression B. GAN. Subsequently, by performing a linear calculation be￾tween the latent vector and the specific direction vector that governs facial expressions, we can utilize StyleGAN’s gener￾ator G to synthesize a new image with the desired facial ex￾pression while preserving the identity information. The pro￾cess of facial expression generatio… view at source ↗
Figures from the paper (1 more)
Figure 4
Figure 4. Figure 4: The neutral image of a PD patient, along with the [PITH_FULL_IMAGE:figures/full_fig_p005_4.png]

Review history (2 revisions) →

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

37 extracted references · 36 canonical work pages

  1. [1]

    Abdal, R., Qin, Y ., & Wonka, P. (2019). Image2stylegan: How to embed images into the stylegan latent space? In ICCV(pp. 4432–4441)

    work page 2019

  2. [2]

    Blin, O., Ferrandez, A.-M., & Serratrice, G. (1990). Quanti- tative analysis of gait in parkinson patients: increased vari- ability of stride length.J. Neurol. Sci.,98(1), 91–97

    work page 1990

  3. [3]

    L., Palmeri, R., De Salvo, S., Berenati, M., Greco, A., Ciurleo, R.,

    Buono, V. L., Palmeri, R., De Salvo, S., Berenati, M., Greco, A., Ciurleo, R., . . . others (2021). Anxiety, depression, and quality of life in parkinson’s disease: the implications of multidisciplinary treatment.Neural Regener . Res.,16(3), 587–590

    work page 2021

  4. [4]

    Choi, Y ., Choi, M., Kim, M., Ha, J.-W., Kim, S., & Choo, J. (2018). Stargan: Unified generative adversarial networks for multi-domain image-to-image translation. InCVPR (pp. 8789–8797)

    work page 2018

  5. [5]

    Duan, H., Wang, J., Chen, K., & Lin, D. (2022). Pyskl: Towards good practices for skeleton action recognition. In ACM MM(pp. 7351–7354)

    work page 2022

  6. [6]

    Fu, Y ., Meng, S., Hou, S., Hu, X., & Huang, Y . (2023). Gp- gait: Generalized pose-based gait recognition. InICCV (pp. 19595–19604)

    work page 2023

  7. [7]

    others (2022)

    Gray, R., Patel, S., Ives, N., Rick, C., Woolley, R., Muzerengi, S., . . . others (2022). Long-term effectiveness of adjuvant treatment with catechol-o-methyltransferase or monoamine oxidase b inhibitors compared with dopamine agonists among patients with parkinson disease uncontrolled by lev- odopa therapy: the pd med randomized clinical trial.JAMA Neuro...

    work page 2022

  8. [8]

    Guo, R., Shao, X., Zhang, C., & Qian, X. (2021). Multi- scale sparse graph convolutional network for the assess- ment of parkinsonian gait.IEEE Trans. Multimedia,24, 1583–1594

    work page 2021

Show all 37 references
  1. [9]

    He, K., Zhang, X., Ren, S., & Sun, J. (2016). Deep residual learning for image recognition. InCVPR(pp. 770–778)

    2016

  2. [10]

    Hou, X., Qin, S., & Su, J. (2022). Visual detection of parkin- son’s disease via facial features recognition. InCIAC(pp. 249–257)

    2022

  3. [11]

    others (2019)

    Tan, M., . . . others (2019). Searching for mobilenetv3. In ICCV(pp. 1314–1324)

    2019

  4. [12]

    Huang, G., Liu, Z., Van Der Maaten, L., & Weinberger, K. Q. (2017). Densely connected convolutional networks. In CVPR(pp. 4700–4708)

    2017

  5. [13]

    Huang, W., Xu, W., Wan, R., Zhang, P., Zha, Y ., & Pang, M. (2024). Auto diagnosis of parkinson’s disease via a deep learning model based on mixed emotional facial expres- sions.IEEE journal of biomedical and health informatics

    2024

  6. [14]

    Huang, W., Zhou, Y ., Cheung, Y.-m., Zhang, P., Zha, Y ., & Pang, M. (2023). Facial expression guided diagnosis of parkinson’s disease via high-quality data augmentation. IEEE Trans. Multimedia,25, 7037–7050

    2023

  7. [15]

    Jin, B., Qu, Y ., Zhang, L., & Gao, Z. (2020). Diagnosing parkinson disease through facial expression recognition: video analysis.J. Med. Internet Res.,22(7), e18697

    2020

  8. [16]

    Karras, T., Laine, S., Aittala, M., Hellsten, J., Lehtinen, J., & Aila, T. (2020). Analyzing and improving the image quality of stylegan. InCVPR(pp. 8110–8119)

    2020

  9. [17]

    Kauw-A-Tjoe, R., Thalen, J., Marin-Perianu, M., & Havinga, P. (2007). Sensorshoe: Mobile gait analysis for parkinson’s disease patients. InProc. ubicomp 2007 workshop proc. (pp. 187–191)

    2007

  10. [18]

    Li, Q., Wang, Y ., Sharf, A., Cao, Y ., Tu, C., Chen, B., & Yu, S. (2018). Classification of gait anomalies from kinect. Visual Comput.,34, 229–241

    2018

  11. [19]

    Zitnick, C

    Lin, T.-Y ., Maire, M., Belongie, S., Hays, J., Perona, P., Ra- manan, D., . . . Zitnick, C. L. (2014). Microsoft COCO: Common objects in context. InECCV(pp. 740–755)

    2014

  12. [20]

    Liu, P., Yu, N., Yang, Y ., Yu, Y ., Sun, X., Yu, H., . . . Wu, J. (2022). Quantitative assessment of gait characteristics in patients with parkinson’s disease using 2d video.PARKIN- SONISM RELAT D,101, 49–56

    2022

  13. [21]

    Xie, S. (2022). A convnet for the 2020s. InCVPR(pp. 11976–11986). O’Shea, S., Morris, M. E., & Iansek, R. (2002). Dual task interference during gait in people with parkinson disease: effects of motor versus cognitive secondary tasks.Phys Ther,82(9), 888–897

    2022

  14. [22]

    Huang, W., & Wen, B. (2024). Heterogeneous proto- type learning from contaminated faces across domains via disentangling latent factors.IEEE Transactions on Neural Networks and Learning Systems

    2024

  15. [23]

    P., Girshick, R., He, K., & Doll´ar, P

    Radosavovic, I., Kosaraju, R. P., Girshick, R., He, K., & Doll´ar, P. (2020). Designing network design spaces. In CVPR(pp. 10428–10436)

    2020

  16. [24]

    Bologna, M

    Ricciardi, L., De Angelis, A., Marsili, L., Faiman, I., Prad- han, P., Pereira, E., . . . Bologna, M. (2020). Hypomimia in parkinson’s disease: an axial sign responsive to levodopa. Eur . J. Neurol.,27(12), 2422–2429

    2020

  17. [25]

    D., Seeher, K

    Steinmetz, J. D., Seeher, K. M., Schiess, N., Nichols, E., Cao, B., Servili, C., . . . others (2024). Global, regional, and national burden of disorders affecting the nervous system, 1990–2021: a systematic analysis for the global burden of disease study 2021.Lancet Neurol.,23...

    2024

  18. [26]

    Sun, K., Xiao, B., Liu, D., & Wang, J. (2019). Deep high- resolution representation learning for human pose estima- tion. InCVPR(pp. 5693–5703)

    2019

  19. [27]

    Tan, M., & Le, Q. (2021). Efficientnetv2: Smaller models and faster training. InICML(pp. 10096–10106)

    2021

  20. [28]

    Terven, J., C´ordova-Esparza, D.-M., & Romero-Gonz´alez, J.- A. (2023). A comprehensive review of yolo architectures in computer vision: From yolov1 to yolov8 and yolo-nas. Mach. Learn. Knowl. Extr .,5(4), 1680–1716

    2023

  21. [29]

    W., & Poewe, W

    Tolosa, E., Garrido, A., Scholz, S. W., & Poewe, W. (2021). Challenges in the diagnosis of parkinson’s disease.Lancet Neurol.,20(5), 385–397

    2021

  22. [30]

    Touvron, H., Cord, M., Douze, M., Massa, F., Sablayrolles, A., & J ´egou, H. (2021). Training data-efficient image transformers & distillation through attention. InICML(pp. 10347–10357)

    2021

  23. [31]

    Weinshall, D. (2015). Quantifying hypomimia in parkinson patients using a depth camera. InMindCare(pp. 63–71)

    2015

  24. [32]

    N., & Chaurasia, A

    Wadekar, S. N., & Chaurasia, A. (2022). Mobilevitv3: Mobile-friendly vision transformer with simple and ef- fective fusion of local, global and input features.arXiv preprint arXiv:2209.15159

    2022 arXiv

  25. [33]

    Yan, S., Xiong, Y ., & Lin, D. (2018). Spatial temporal graph convolutional networks for skeleton-based action recogni- tion. InAAAI(V ol. 32)

    2018

  26. [34]

    others (2020)

    Yang, T., Yang, Z., Xu, G., Gao, D., Zhang, Z., Wang, H., . . . others (2020). Tsinghua facial expression database–a database of facial expressions in chinese young and older women and men: Development and validation.PloS one, 15(4), e0231304

    2020

  27. [35]

    J., Wang, Z., Contarino, M

    Yin, Z., Geraedts, V. J., Wang, Z., Contarino, M. F., Dibekli- oglu, H., & Van Gemert, J. (2021). Assessment of parkin- son’s disease severity from videos using deep architectures. IEEE J. Biomed. Health. Inf.,26(3), 1164–1176

    2021

  28. [36]

    Zhou, Y ., Pang, M., Huang, W., & Wang, B. (2024). Early diagnosing parkinson’s disease via a deep learning model based on augmented facial expression data. InICASSP(pp. 1621–1625)

    2024

  29. [37]

    Zhu, J.-Y ., Park, T., Isola, P., & Efros, A. A. (2017). Un- paired image-to-image translation using cycle-consistent adversarial networks. InICCV(pp. 2223–2232)

    2017

Pith tools

Reviewed August 15, 2026 · model on record in the stance chip above.