pith. sign in

arxiv: 2504.09463 · v1 · submitted 2025-04-13 · 💻 cs.LG · cs.AI

Comorbidity-Informed Transfer Learning for Neuro-developmental Disorder Diagnosis

Xin Wen , Shijie Guo , Wenbo Ning , Rui Cao , Jie Xiang , Xiaobo Liu , Jintai Chen This is my paper

Pith reviewed 2026-05-22 20:04 UTC · model grok-4.3

classification 💻 cs.LG cs.AI
keywords transfer learningfMRIautism spectrum disorderADHDcomorbiditypseudo-labellingneuro-developmental disorderscomputer-aided diagnosis
0
0 comments X

The pith

Comorbidity-informed transfer learning cleans fMRI temporal patterns to raise ASD and ADHD diagnosis accuracy.

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

The paper introduces the Comorbidity-Informed Transfer Learning (CITL) framework for diagnosing neuro-developmental disorders from fMRI scans. A reinforced representation generation network first applies transfer learning and pseudo-labelling to strip away interfering temporal patterns, then uses an encoder-decoder to produce new representations that feed into a simple classification network. The method explicitly folds in comorbidity mechanisms between disorders and reports 76.32 percent accuracy on autism spectrum disorder and 73.15 percent on ADHD, exceeding prior transfer-learning baselines by 7.2 and 0.5 percentage points respectively.

Core claim

The CITL framework combines transfer learning with pseudo-labelling inside a reinforced representation generation network to remove interfering temporal patterns from fMRI, generates cleaned representations through an encoder-decoder architecture, and integrates comorbidity mechanisms of neuro-developmental disorders with semi-supervised learning to train an architecturally simple classification network, yielding the stated accuracies.

What carries the argument

The reinforced representation generation network, which merges transfer learning and pseudo-labelling to eliminate distracting temporal signals before an encoder-decoder produces new fMRI representations, while comorbidity mechanisms guide the overall pipeline.

If this is right

  • CAD models for neuro-developmental disorders become more accurate when temporal distractions are explicitly removed before classification.
  • Comorbidity information can be fused with semi-supervised and transfer techniques to produce new interdisciplinary diagnostic pipelines.
  • Healthcare systems gain relief from strained neuroimaging resources once cleaned representations support reliable automated diagnosis.
  • The encoder-decoder step demonstrates a concrete route for regenerating usable fMRI features from noisy spatio-temporal data.

Where Pith is reading between the lines

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

  • The same pattern-removal step could be tested on resting-state versus task-based fMRI to measure whether the cleaning benefit holds across acquisition types.
  • Extending the comorbidity integration to additional co-occurring conditions such as anxiety disorders would test the framework's scalability.
  • Replacing the current pseudo-labelling with other semi-supervised signals might further lift accuracy without changing the overall architecture.

Load-bearing premise

The reinforced representation generation network successfully removes interfering temporal patterns from fMRI through transfer learning plus pseudo-labelling, and comorbidity mechanisms improve the final classification.

What would settle it

Apply the full CITL pipeline to a fresh, independent set of fMRI scans for autism spectrum disorder and ADHD; if the resulting accuracies fall below 70 percent or fail to exceed the best prior transfer-learning baselines, the central claim is falsified.

Figures

Figures reproduced from arXiv: 2504.09463 by Jie Xiang, Jintai Chen, Rui Cao, Shijie Guo, Wenbo Ning, Xiaobo Liu, Xin Wen.

Figure 1
Figure 1. Figure 1: illustrates the proposed CITL framework, which consists of three main components: (a) Construction of [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: illustrates the two convolution modules of the Depthwise Separable Convolution Model: (a) represents the [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: The Conversion Engine removes the set with fewer samples and then optimizes the remaining dFCs through [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: The Feature Optimization AE model. The variables from the hidden layer are transformed into a 6670- [PITH_FULL_IMAGE:figures/full_fig_p006_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: The ablation experiments of different modules. The result is shown in mean [PITH_FULL_IMAGE:figures/full_fig_p008_5.png] view at source ↗
read the original abstract

Neuro-developmental disorders are manifested as dysfunctions in cognition, communication, behaviour and adaptability, and deep learning-based computer-aided diagnosis (CAD) can alleviate the increasingly strained healthcare resources on neuroimaging. However, neuroimaging such as fMRI contains complex spatio-temporal features, which makes the corresponding representations susceptible to a variety of distractions, thus leading to less effective in CAD. For the first time, we present a Comorbidity-Informed Transfer Learning(CITL) framework for diagnosing neuro-developmental disorders using fMRI. In CITL, a new reinforced representation generation network is proposed, which first combines transfer learning with pseudo-labelling to remove interfering patterns from the temporal domain of fMRI and generates new representations using encoder-decoder architecture. The new representations are then trained in an architecturally simple classification network to obtain CAD model. In particular, the framework fully considers the comorbidity mechanisms of neuro-developmental disorders and effectively integrates them with semi-supervised learning and transfer learning, providing new perspectives on interdisciplinary. Experimental results demonstrate that CITL achieves competitive accuracies of 76.32% and 73.15% for detecting autism spectrum disorder and attention deficit hyperactivity disorder, respectively, which outperforms existing related transfer learning work for 7.2% and 0.5% respectively.

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

2 major / 0 minor

Summary. The manuscript proposes a Comorbidity-Informed Transfer Learning (CITL) framework for fMRI-based diagnosis of autism spectrum disorder (ASD) and attention deficit hyperactivity disorder (ADHD). It introduces a reinforced representation generation network that combines transfer learning with pseudo-labelling to remove interfering temporal patterns, employs an encoder-decoder to produce new representations, and integrates comorbidity mechanisms with semi-supervised and transfer learning. A simple classification network is then trained on these representations. Experimental results claim accuracies of 76.32% for ASD and 73.15% for ADHD, outperforming existing transfer learning methods by 7.2% and 0.5%, respectively.

Significance. If the performance claims can be substantiated through detailed experimental protocols and component analyses, the work could contribute an interdisciplinary perspective on leveraging comorbidity information in semi-supervised transfer learning for neuroimaging-based CAD. The approach addresses a relevant clinical need, but the current lack of supporting validation details prevents a stronger assessment of its potential impact.

major comments (2)
  1. [Abstract] Abstract: The headline accuracies (76.32% ASD, 73.15% ADHD) and claimed improvements (7.2%, 0.5%) are stated without any accompanying information on dataset sizes, number of subjects, cross-validation folds, statistical significance testing, baseline re-implementations, or controls for confounds such as site effects. These omissions make it impossible to assess the reliability or generalizability of the central empirical claims.
  2. [Experimental Results] Experimental Results: No ablation studies, component analyses, or quantitative isolation of contributions are provided to demonstrate that the reported gains arise specifically from comorbidity integration, pseudo-labelling, or the reinforced representation generation network rather than from dataset-specific preprocessing, architecture choices, or baseline differences. This directly undermines the novelty argument that ties performance to the proposed mechanisms.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the detailed and constructive feedback on our manuscript. We have carefully reviewed the major comments and will revise the paper to address the concerns about experimental transparency and component validation, which we agree will strengthen the work.

read point-by-point responses

  1. Referee: [Abstract] Abstract: The headline accuracies (76.32% ASD, 73.15% ADHD) and claimed improvements (7.2%, 0.5%) are stated without any accompanying information on dataset sizes, number of subjects, cross-validation folds, statistical significance testing, baseline re-implementations, or controls for confounds such as site effects. These omissions make it impossible to assess the reliability or generalizability of the central empirical claims.

    Authors: We acknowledge that the abstract, due to its brevity, does not include these supporting details. In the revised manuscript we will expand the Experimental Results section with a new 'Experimental Setup' subsection that explicitly reports dataset sizes and subject counts, the cross-validation procedure, statistical significance testing (e.g., paired tests against baselines), how the transfer-learning baselines were re-implemented, and the harmonization steps taken to mitigate site effects. We will also insert a concise reference to these elements in the abstract itself where space permits. revision: yes

  2. Referee: [Experimental Results] Experimental Results: No ablation studies, component analyses, or quantitative isolation of contributions are provided to demonstrate that the reported gains arise specifically from comorbidity integration, pseudo-labelling, or the reinforced representation generation network rather than from dataset-specific preprocessing, architecture choices, or baseline differences. This directly undermines the novelty argument that ties performance to the proposed mechanisms.

    Authors: The referee is correct that the current version lacks explicit ablation or component-isolation experiments. We will add a dedicated ablation study subsection in the revised Experimental Results. This will include controlled variants that successively disable comorbidity integration, pseudo-labelling, and the reinforced representation generation network, reporting the resulting accuracy drops on both ASD and ADHD tasks. These quantitative results will directly link the observed improvements to the proposed mechanisms rather than to preprocessing or architectural differences. revision: yes

Circularity Check

0 steps flagged

Empirical accuracies on held-out data; no reduction to inputs by construction

full rationale

The paper introduces the CITL framework, which combines transfer learning, pseudo-labelling, an encoder-decoder for reinforced representations, and comorbidity integration to address temporal interference in fMRI. The central claims are the reported accuracies of 76.32% (ASD) and 73.15% (ADHD) with outperformance margins of 7.2% and 0.5% over prior transfer learning methods. These metrics are presented as results from experimental evaluation rather than quantities defined directly from the same fitted parameters, self-citations, or ansatzes. No equations or framework steps in the provided description reduce the performance numbers to the inputs by construction, and the derivation relies on independent held-out testing against external benchmarks. The chain is therefore self-contained.

Axiom & Free-Parameter Ledger

1 free parameters · 2 axioms · 0 invented entities

The framework rests on standard deep-learning assumptions plus two domain-specific premises about comorbidity integration and temporal-pattern removal; no new physical entities are postulated and free parameters are the usual neural-network hyperparameters whose values are not reported.

free parameters (1)
  • network hyperparameters and pseudo-label thresholds
    Typical learned or chosen values in the encoder-decoder and classification stages; exact values and selection procedure not stated in abstract.
axioms (2)
  • domain assumption Transfer learning combined with pseudo-labelling can remove interfering temporal patterns from fMRI
    Invoked in the description of the reinforced representation generation network.
  • domain assumption Comorbidity mechanisms of neuro-developmental disorders can be effectively fused with semi-supervised and transfer learning
    Stated as a core design principle of the CITL framework.

pith-pipeline@v0.9.0 · 5767 in / 1418 out tokens · 95777 ms · 2026-05-22T20:04:33.398221+00:00 · methodology

discussion (0)

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

Lean theorems connected to this paper

Citations machine-checked in the Pith Canon. Every link opens the source theorem in the public Lean library.

What do these tags mean?
matches
The paper's claim is directly supported by a theorem in the formal canon.
supports
The theorem supports part of the paper's argument, but the paper may add assumptions or extra steps.
extends
The paper goes beyond the formal theorem; the theorem is a base layer rather than the whole result.
uses
The paper appears to rely on the theorem as machinery.
contradicts
The paper's claim conflicts with a theorem or certificate in the canon.
unclear
Pith found a possible connection, but the passage is too broad, indirect, or ambiguous to say the theorem truly supports the claim.

Reference graph

Works this paper leans on

40 extracted references · 40 canonical work pages

  1. [1]

    Neurodevelopmental disorders

    Anita Thapar, Miriam Cooper, and Michael Rutter. Neurodevelopmental disorders. The Lancet Psychiatry , 4(4):339–346, 2017

    work page 2017

  2. [2]

    Autism spectrum disorders

    Marlene Briciet Lauritsen. Autism spectrum disorders. European child & adolescent psychiatry , 22:37–42, 2013

    work page 2013

  3. [3]

    Neurobiology of adhd

    Gail Tripp and Jeffery R Wickens. Neurobiology of adhd. Neuropharmacology, 57(7-8):579–589, 2009

    work page 2009

  4. [4]

    Neurological disorders: public health challenges

    World Health Organization. Neurological disorders: public health challenges . World Health Organization, 2006

    work page 2006

  5. [5]

    Computer-aided diagnosis in medical imaging: historical review, current status and future potential

    Kunio Doi. Computer-aided diagnosis in medical imaging: historical review, current status and future potential. Computerized medical imaging and graphics , 31(4-5):198–211, 2007

    work page 2007

  6. [6]

    A review of resting-state fmri and its use to examine psychiatric disorders

    Edgar Canario, Donna Chen, and Bharat Biswal. A review of resting-state fmri and its use to examine psychiatric disorders. Psychoradiology, 1(1):42–53, 2021

    work page 2021

  7. [7]

    Functional connectivity-based prediction of autism on site harmonized abide dataset

    Madhura Ingalhalikar, Sumeet Shinde, Arnav Karmarkar, Archith Rajan, D Rangaprakash, and Gopikrishna Desh- pande. Functional connectivity-based prediction of autism on site harmonized abide dataset. IEEE Transactions on Biomedical Engineering, 68(12):3628–3637, 2021

    work page 2021

  8. [8]

    Multi-task learning for efficient diagnosis of asd and adhd using resting-state fmri data

    Zhi-An Huang, Rui Liu, and Kay Chen Tan. Multi-task learning for efficient diagnosis of asd and adhd using resting-state fmri data. In 2020 International Joint Conference on Neural Networks (IJCNN) , pages 1–7. IEEE, 2020

    work page 2020

  9. [9]

    A survey of brain functional network extraction methods using fmri data

    Yuhui Du, Songke Fang, Xingyu He, and Vince D Calhoun. A survey of brain functional network extraction methods using fmri data. Trends in Neurosciences, 2024

    work page 2024

  10. [10]

    Source free semi-supervised transfer learning for diagnosis of mental disorders on fmri scans

    Yao Hu, Zhi-An Huang, Rui Liu, Xiaoming Xue, Xiaoyan Sun, Linqi Song, and Kay Chen Tan. Source free semi-supervised transfer learning for diagnosis of mental disorders on fmri scans. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2023

    work page 2023

  11. [11]

    Brainsck: Brain structure and cognition alignment via knowledge injection and reactivation for diagnosing brain disorders

    Lilong Wang, Mianxin Liu, Shaoting Zhang, and Xiaosong Wang. Brainsck: Brain structure and cognition alignment via knowledge injection and reactivation for diagnosing brain disorders. In International Conference on Medical Image Computing and Computer-Assisted Intervention , pages 45–55. Springer, 2024

    work page 2024

  12. [12]

    A brief review of domain adaptation

    Abolfazl Farahani, Sahar V oghoei, Khaled Rasheed, and Hamid R Arabnia. A brief review of domain adaptation. Advances in data science and information engineering: proceedings from ICDATA 2020 and IKE 2020 , pages 877–894, 2021

    work page 2020

  13. [13]

    Domain adaptation using a three-way decision improves the identification of autism patients from multisite fmri data

    Chunlei Shi, Xianwei Xin, and Jiacai Zhang. Domain adaptation using a three-way decision improves the identification of autism patients from multisite fmri data. Brain Sciences, 11(5):603, 2021

    work page 2021

  14. [14]

    Characterizing psychiatric comorbidity in children with autism spectrum disorder receiving publicly funded mental health services

    Lauren Brookman-Frazee, Nicole Stadnick, Colby Chlebowski, Mary Baker-Ericzén, and William Ganger. Characterizing psychiatric comorbidity in children with autism spectrum disorder receiving publicly funded mental health services. Autism, 22(8):938–952, 2018

    work page 2018

  15. [15]

    Autism spectrum disorders and adhd: Overlapping phenomenology, diagnostic issues, and treatment considerations

    Kevin M Antshel and Natalie Russo. Autism spectrum disorders and adhd: Overlapping phenomenology, diagnostic issues, and treatment considerations. Current psychiatry reports, 21:1–11, 2019

    work page 2019

  16. [16]

    A semi-supervised autoencoder for autism disease diagnosis

    Wutao Yin, Longhai Li, and Fang-Xiang Wu. A semi-supervised autoencoder for autism disease diagnosis. Neurocomputing, 483:140–147, 2022

    work page 2022

  17. [17]

    The co-occurrence of autism spectrum disorder in children with adhd

    Benjamin Zablotsky, Matthew D Bramlett, and Stephen J Blumberg. The co-occurrence of autism spectrum disorder in children with adhd. Journal of attention disorders, 24(1):94–103, 2020

    work page 2020

  18. [18]

    Braintgl: A dynamic graph representation learning model for brain network analysis

    Lingwen Liu, Guangqi Wen, Peng Cao, Tianshun Hong, Jinzhu Yang, Xizhe Zhang, and Osmar R Zaiane. Braintgl: A dynamic graph representation learning model for brain network analysis. Computers in Biology and Medicine , 153:106521, 2023

    work page 2023

  19. [19]

    Spatio- temporal graph convolution for resting-state fmri analysis

    Soham Gadgil, Qingyu Zhao, Adolf Pfefferbaum, Edith V Sullivan, Ehsan Adeli, and Kilian M Pohl. Spatio- temporal graph convolution for resting-state fmri analysis. In Medical Image Computing and Computer Assisted Intervention–MICCAI 2020: 23rd International Conference, Lima, Peru, October 4–8, 2020, Proceedings, Part VII 23, pages 528–538. Springer, 2020

    work page 2020

  20. [20]

    Transient brain activity disentangles fmri resting-state dynamics in terms of spatially and temporally overlapping networks

    Fikret I¸ sik Karahano˘glu and Dimitri Van de Ville. Transient brain activity disentangles fmri resting-state dynamics in terms of spatially and temporally overlapping networks. Nature Communications, 6, 2015

    work page 2015

  21. [21]

    Detecting the information of functional connectivity networks in normal aging using deep learning from a big data perspective

    Xin Wen, Li Dong, Junjie Chen, Jie Xiang, Jie Yang, Hechun Li, Xiaobo Liu, Cheng Luo, and Dezhong Yao. Detecting the information of functional connectivity networks in normal aging using deep learning from a big data perspective. Frontiers in Neuroscience, 13, 2020. 11 PRIME AI paper

    work page 2020

  22. [22]

    Spatial–temporal co-attention learning for diagnosis of mental disorders from resting-state fmri data

    Rui Liu, Zhi-An Huang, Yao Hu, Zexuan Zhu, Ka-Chun Wong, and Kay Chen Tan. Spatial–temporal co-attention learning for diagnosis of mental disorders from resting-state fmri data. IEEE Transactions on Neural Networks and Learning Systems, 35(8):10591–10605, 2024

    work page 2024

  23. [23]

    Modeling dynamic characteristics of brain functional connectivity networks using resting-state functional mri

    Mingliang Wang, Jiashuang Huang, Mingxia Liu, and Daoqiang Zhang. Modeling dynamic characteristics of brain functional connectivity networks using resting-state functional mri. Medical image analysis, 71:102063, 2021

    work page 2021

  24. [24]

    Ssmda: Semi-supervised multi-source domain adaptive autism prediction model using neuroimaging

    Mehak Mengi and Deepti Malhotra. Ssmda: Semi-supervised multi-source domain adaptive autism prediction model using neuroimaging. Biomedical Signal Processing and Control, 95:106337, 2024

    work page 2024

  25. [25]

    Semi-supervised clustering for neuro-subtyping of autism spectrum disorder

    Zihao Li, Yaping Wang, and Xiujuan Geng. Semi-supervised clustering for neuro-subtyping of autism spectrum disorder. In Medical Imaging 2023: Image Processing , volume 12464, pages 644–651. SPIE, 2023

    work page 2023

  26. [26]

    A dual-stage pseudo-labeling method for the diagnosis of mental disorder on mri scans

    Yao Hu, Zhi-An Huang, Rui Liu, Xiaoming Xue, Linqi Song, and Kay Chen Tan. A dual-stage pseudo-labeling method for the diagnosis of mental disorder on mri scans. In 2022 International Joint Conference on Neural Networks (IJCNN), pages 1–8. IEEE, 2022

    work page 2022

  27. [27]

    A semi-supervised federated learning approach leveraging pseudo-labeling for Knee Osteoarthritis severity detection

    Rakib Hossain Rifat. A semi-supervised federated learning approach leveraging pseudo-labeling for Knee Osteoarthritis severity detection. PhD thesis, Brac University, 2024

    work page 2024

  28. [28]

    Federated learning in computer vision

    Donald Shenaj, Giulia Rizzoli, and Pietro Zanuttigh. Federated learning in computer vision. IEEE Access, 2023

    work page 2023

  29. [29]

    Can sliding-window correlations reveal dynamic functional connectivity in resting-state fmri? Neuroimage, 127:242–256, 2016

    Rikkert Hindriks, Mohit H Adhikari, Yusuke Murayama, Marco Ganzetti, Dante Mantini, Nikos K Logothetis, and Gustavo Deco. Can sliding-window correlations reveal dynamic functional connectivity in resting-state fmri? Neuroimage, 127:242–256, 2016

    work page 2016

  30. [30]

    Automated anatomical labelling atlas 3

    Edmund T Rolls, Chu-Chung Huang, Ching-Po Lin, Jianfeng Feng, and Marc Joliot. Automated anatomical labelling atlas 3. Neuroimage, 206:116189, 2020

    work page 2020

  31. [31]

    Cross-entropy loss functions: Theoretical analysis and applications

    Anqi Mao, Mehryar Mohri, and Yutao Zhong. Cross-entropy loss functions: Theoretical analysis and applications. In International conference on Machine learning , pages 23803–23828. PMLR, 2023

    work page 2023

  32. [32]

    Sagar Imambi, Kolla Bhanu Prakash, and GR Kanagachidambaresan. Pytorch. Programming with TensorFlow: solution for edge computing applications , pages 87–104, 2021

    work page 2021

  33. [33]

    Automated deep learning using neural network intelligence

    Models Using Python and Ivan Gridin. Automated deep learning using neural network intelligence

    work page

  34. [34]

    The autism brain imaging data exchange: towards a large-scale evaluation of the intrinsic brain architecture in autism

    Adriana Di Martino, Chao-Gan Yan, Qingyang Li, Erin Denio, Francisco X Castellanos, Kaat Alaerts, Jeffrey S Anderson, Michal Assaf, Susan Y Bookheimer, Mirella Dapretto, et al. The autism brain imaging data exchange: towards a large-scale evaluation of the intrinsic brain architecture in autism. Molecular psychiatry, 19(6):659–667, 2014

    work page 2014

  35. [35]

    The adhd-200 consortium: a model to advance the translational potential of neuroimaging in clinical neuroscience

    ADHD-200 consortium. The adhd-200 consortium: a model to advance the translational potential of neuroimaging in clinical neuroscience. Frontiers in systems neuroscience, 6:62, 2012

    work page 2012

  36. [36]

    The neuro bureau preprocessing initiative: open sharing of preprocessed neuroimaging data and derivatives

    Cameron Craddock, Yassine Benhajali, Carlton Chu, Francois Chouinard, Alan Evans, András Jakab, Budhachan- dra Singh Khundrakpam, John David Lewis, Qingyang Li, Michael Milham, et al. The neuro bureau preprocessing initiative: open sharing of preprocessed neuroimaging data and derivatives. Frontiers in Neuroinformatics, 7(27):5, 2013

    work page 2013

  37. [37]

    The neuro bureau adhd-200 preprocessed repository

    Pierre Bellec, Carlton Chu, Francois Chouinard-Decorte, Yassine Benhajali, Daniel S Margulies, and R Cameron Craddock. The neuro bureau adhd-200 preprocessed repository. Neuroimage, 144:275–286, 2017

    work page 2017

  38. [38]

    Braingnn: Interpretable brain graph neural network for fmri analysis

    Xiaoxiao Li, Yuan Zhou, Nicha Dvornek, Muhan Zhang, Siyuan Gao, Juntang Zhuang, Dustin Scheinost, Lawrence H Staib, Pamela Ventola, and James S Duncan. Braingnn: Interpretable brain graph neural network for fmri analysis. Medical Image Analysis, 74:102233, 2021

    work page 2021

  39. [39]

    Brainnetcnn: Convolutional neural networks for brain networks; towards predicting neurodevelopment

    Jeremy Kawahara, Colin J Brown, Steven P Miller, Brian G Booth, Vann Chau, Ruth E Grunau, Jill G Zwicker, and Ghassan Hamarneh. Brainnetcnn: Convolutional neural networks for brain networks; towards predicting neurodevelopment. NeuroImage, 146:1038–1049, 2017

    work page 2017

  40. [40]

    Federated multi- task learning for joint diagnosis of multiple mental disorders on mri scans

    Zhi-An Huang, Yao Hu, Rui Liu, Xiaoming Xue, Zexuan Zhu, Linqi Song, and Kay Chen Tan. Federated multi- task learning for joint diagnosis of multiple mental disorders on mri scans. IEEE Transactions on Biomedical Engineering, 70(4):1137–1149, 2022. 12

    work page 2022