Yang Merik Liu

All Publications

• Decoupling neural representation development from trial evaluation: an operational framework for target validation and discovery in older adults. Ageing research reviews Liu, Y. M., Turnbull, A., Ai, M., Zhang, Y., Vankee-Lin, F. 2026: 103365

  Abstract

  Randomized controlled trials (RCTs) targeting cognitive health in older adults increasingly incorporate resting-state functional MRI (rs-fMRI) to evaluate intervention-related neural change and its association with cognitive outcomes. However, these trials are typically small, and rs-fMRI data are high-dimensional and sensitive to measurement variability, making neural representations developed and evaluated within the same trial vulnerable to instability and overfitting. Neural representations learned from large observational datasets can reduce dependence on within-trial model development, but strong predictive performance after transfer does not establish prospective validity in an intervention setting. Here, we propose an operational framework for using externally developed neural representations in RCTs, separating representation development, optional adaptation, and trial use. The framework distinguishes target validation, in which the target and evaluation procedure are specified before outcomes are examined, from target discovery, in which outcomes may inform downstream fitting, selection, or interpretation, with the resulting patterns treated as candidate targets requiring independent confirmation. Using two RCTs as worked examples, we illustrate validation-oriented evaluation of an externally developed connectivity representation and discovery-oriented use of an adapted rs-fMRI foundation-model representation. This framework provides a practical basis for accumulating evidence across trials and distinguishing neural features that are consistently intervention-responsive from those that are context dependent.

  View details for DOI 10.1016/j.arr.2026.103365

  View details for PubMedID 42716443

• BRAIN-DISC: a novel analytical approach integrating data-driven brain pattern discovery with cohort studies for evaluating findings in brain modulation interventions. GeroScience Ai, M., Turnbull, A., Zhao, K., Liu, Y., Zhang, Y., Vankee-Lin, F. 2026

  Abstract

  Brain modulation interventions (BMIs) targeting cognitive and neuropsychiatric symptoms have shown substantial heterogeneity in response, limiting their clinical utility. Resting-state functional connectivity (rsFC) may capture BMI-induced neuroplasticity and support patient stratification. However, examining these biomarkers within small, heterogeneous intervention samples remains challenging. The objective of this study is to present BRAIN-DISC, an analytic framework that links large-scale cohort-derived rsFC patterns with evaluation in targeted BMI trials. Three demonstrations were conducted. In the CogTE trial (n = 74), the Alzheimer's-resilient connectome (ARC), derived from cohort contrasts of Superagers and Alzheimer's disease, was evaluated as a response biomarker for cognitive training in mild cognitive impairment (MCI). In the BEEM trial (n = 26), a brain-derived neuropsychiatric phenotyping (BNP) subtype was used to stratify response to transcranial direct current stimulation combined with training. In a third demonstration, we conducted an end-to-end implementation by discovering rsFC biotypes jointly informed by autonomic nervous system (ANS) and cognitive function (rsFC-AC) in the MIDUS cohort (n = 208), and evaluating it as a predictive biomarker in BREATHE trial (n = 56). In CogTE, greater shifts toward the ARC pattern were associated with improvements in executive function and episodic memory. In BEEM, individuals with the affective dysregulation subtype showed greater improvement in corresponding neuropsychiatric domains. In the third demo, three rsFC-AC biotypes were discovered in MIDUS cohort; the high-ANS subtype showed greater improvement in episodic memory in BREATHE trial. BRAIN-DISC provides a scalable framework for translating cohort-derived rsFC signatures into intervention settings to support both response monitoring and patient stratification.

  View details for DOI 10.1007/s11357-026-02471-w

  View details for PubMedID 42625098

  View details for PubMedCentralID 7355168

• Cross-Species Behavioral Analysis via Multidimensional Knowledge Transfer IEEE INTELLIGENT SYSTEMS Liu, Y., Shaw, L., Liu, H., Wang, K., Vankee-Lin, F., Zhao, G. 2026; 41 (4): 98-107

  View details for DOI 10.1109/MIS.2026.3692032

  View details for Web of Science ID 001853451800008

• Cross-Species Behavioral Analysis via Multidimensional Knowledge Transfer. IEEE intelligent systems Liu, Y., Shaw, L., Liu, H., Wang, K. H., Vankee-Lin, F., Zhao, G. 2026; 41 (4): 98-107

  Abstract

  Computer vision tools can provide objective, reproducible, and scalable measurements for understanding animal behaviors, which is a fundamental challenge in cross-species research. Existing approaches primarily emphasize body detection and tracking, with limited advancement in integrative behavioral pattern analysis due to sparse annotations and substantial variability in species' physical and cognitive traits. In this article, we conduct two pilot studies for cross-species behavioral analysis within primates using multidimensional transfer learning. First, prior knowledge of physiological similarities between humans and monkeys is introduced to adjust the representation learning for macaque facial expression classification under limited labeled data. Second, we improve our approach by leveraging the foundation model and self-supervised strategy to mitigate appearance variations across primate species for downstream marmoset engagement level estimation. Experiments on public and private cross-primates datasets demonstrate the effectiveness of our framework, highlighting the potential of knowledge transfer to bridge the gap between species-specific behavioral analysis and generalizable models.

  View details for DOI 10.1109/mis.2026.3692032

  View details for PubMedID 42663010

  View details for PubMedCentralID PMC13521619

• Resting-state fMRI foundation models enable robust and generalizable latent neural target discovery in cognitive aging interventions. bioRxiv : the preprint server for biology Zhou, X., Ai, M., Adeli, E., Zhang, Y., Liu, Y. M., Vankee-Lin, F. 2026

  Abstract

  The benefits of interventions targeting cognitive aging vary substantially across individuals, largely owing to heterogeneity in aging-related comorbidities. It is necessary to robustly identify neural patterns underlying intervention response and test their generalizability across heterogeneous cohorts. Resting-state functional MRI (rsfMRI) offers a potential pathway, but relying on predefined summary features with conventional methods has limited capacity to capture both within-individual longitudinal variation and between-individual differences, particularly in small and heterogeneous studies. Recent rsfMRI foundation models pretrained on large observational cohorts present a promising alternative by learning transferable spatiotemporal representations from time-series signals. Yet their validity and generalizability in local intervention settings remain unclear. Here, we systematically evaluated rsfMRI foundation models using data from two independent randomized controlled trials of older adults with mild cognitive impairment, testing whether these models can robustly extract longitudinal brain representations that predict post-intervention changes in episodic memory across trials. Foundation models outperformed conventional machine learning and deep learning approaches across both trials. Clinically informed adaptation using an external Alzheimer's disease cohort further improved performance and robustness to confounders (i.e., head motion, site, and intervention arm), with accuracy up to 82%. Multivariate decomposition of foundation model embeddings identified latent neural patterns associated with episodic memory change with cross-study consistency at baseline that became more spatially distributed at post-intervention. These findings show that rsfMRI foundation models can enable robust and generalizable identification of latent neural patterns linking longitudinal brain dynamics to individual intervention response, laying the foundation for precision-driven neural target discovery in cognitive aging research.

  View details for DOI 10.64898/2025.12.30.697042

  View details for PubMedID 42039513

  View details for PubMedCentralID PMC13105019

• APGFusion : Adaptive PoolFormer and CNN medical image fusion network based on convolutional gated linear units EXPERT SYSTEMS WITH APPLICATIONS Hu, X., Lu, D., Liu, Y., Wu, B., Yang, F. 2026; 307

  View details for DOI 10.1016/j.eswa.2025.130867

  View details for Web of Science ID 001660774100004

• SORT-LFR: Revisiting SORT for Multi-Object Tracking in Low-Frame-Rate Videos IEEE TRANSACTIONS ON MULTIMEDIA Zeng, L., Huang, Y., Lin, Y., Zhu, Z., Zhang, X., Liu, Y., Li, Y., Zheng, Y. 2026; 28: 3364-3379

  View details for DOI 10.1109/TMM.2026.3651130

  View details for Web of Science ID 001754908800006

• Edge Harmony Attention Network for Semi-Supervised Medical Image Segmentation IEEE TRANSACTIONS ON INSTRUMENTATION AND MEASUREMENT Lu, M., Liu, Y., Chen, Y., Yang, F. 2026; 75

  View details for DOI 10.1109/TIM.2026.3664601

  View details for Web of Science ID 001714477700012

• A multi-dimensional transfer learning framework for studying reward-guided behaviors across species NATURE MENTAL HEALTH Liu, Y., Turnbull, A., Adeli, E., Zhao, G., Wang, K., Vankee-Lin, F. 2025

  View details for DOI 10.1038/s44220-025-00547-8

  View details for Web of Science ID 001626268600001

• RAPOO: An Efficient Privacy-Preserving Facial Expression Recognition via Mobile Crowdsensing IEEE TRANSACTIONS ON MOBILE COMPUTING Tian, B., Zhao, B., Xiao, Y., Liu, Y., Pei, Q., Shen, Y. 2025; 24 (11): 11568-11581

  View details for DOI 10.1109/TMC.2025.3581687

  View details for Web of Science ID 001587233000020

• Fast highway abandoned object detection via block-based multi-group foreground extraction SCIENTIFIC REPORTS Liu, D., Wang, H., Zhang, X., Zhang, X., Liu, Y. 2025; 15 (1): 37389

  Abstract

  Abandoned objects on highways pose a significant risk of causing severe traffic accidents. Current abandoned object detection technologies are limited by hardware constraints and real-time processing requirements in complex highway environments. To address these challenges, we propose a Universal foreground extraction framework consisting of Block-based frame selection and Multi-Group foreground Detection, called UBMG, which effectively alleviates false alarms caused by either illumination changes on the highway, traffic targets, or road markings. Specifically, two modules of block preprocessing and adaptive size filtering are first designed, followed by a static target matching module, to extract candidate targets. To accurately distinguish abandoned objects from other entities, a candidate verification strategy is proposed involving traffic target elimination, road noise elimination, and trajectory discrimination. This framework can seamlessly integrate with existing pixel-wise foreground detection algorithms and demonstrate high efficiency in practical applications. In addition, we have established a comprehensive video dataset of highway abandoned objects, named HAO, under various conditions (e.g.,lighting conditions, camera movements, and object occlusions) for thorough evaluation. Extensive experiments on HAO and public ABODA datasets demonstrate that the UBMG framework can perform robust and real-time detection of abandoned objects, especially outperforming the state-of-the-art methods in the HAO dataset and showing good performance on ABODA.

  View details for DOI 10.1038/s41598-025-20331-z

  View details for Web of Science ID 001603728600006

  View details for PubMedID 41145526

  View details for PubMedCentralID PMC12559225

• 3-D Face De-Identification With Preserving Multi-Facial Attributes: A Benchmark IEEE TRANSACTIONS ON BIOMETRICS, BEHAVIOR, AND IDENTITY SCIENCE Liu, Y., Cheng, K. H. M., Savic, M., Chen, H., Yu, Z., Zhao, G. 2025; 7 (4): 681-694

  View details for DOI 10.1109/TBIOM.2025.3570599

  View details for Web of Science ID 001585336700013

• Personalized cognitive enhancement for older adults: An aging-friendly closed-loop human-machine interface framework. Ageing research reviews Zhou, S., Liu, Y., Turnbull, A., Tapparello, C., Adeli, E., Lin, F. V. 2025: 102877

  Abstract

  Emerging digitally delivered non-pharmacological interventions (dNPIs) offer scalable, low-risk solutions for enhancing cognitive function in older adults, yet their effectiveness remains inconsistent due to a lack of personalization and precise mechanisms of action. Generic, population-based designs often fail to predict individual gains, underscoring the need for more tailored approaches. To address this, we propose a closed-loop human-machine interface (HMI) framework for personalizing dNPIs by optimizing the engagement of neurocognitive resources for cognitive enhancement. Our framework tackles three major challenges: (1) comprehensive and effective neurobehavioral representations for cognitive decoding, (2) tailoring interventions for domain-specific cognitive processes, and (3) ensuring aging-friendly design on usability, validity, and reliability for long-term adherence. We provide reviews and perspectives to guide the development of closed-loop HMIs by outlining the operational details of three key components-sensor, controller, and external actuator-that monitor, analyze, and modulate neurobehavioral activities through real-time adaptive interventions. Centering on neurobehavioral characteristics of older adults, we propose to advance closed-loop HMIs toward (1) deploying multimodal sensor network that captures activities from both central and peripheral nervous systems, (2) artificial intelligence (AI)-powered cognitive decoding and modulation that integrates multi-modal easy-to-acquire neurobehavioral signals and predicts the cross-modal harder-to-acquire signals, and (3) targeting neurobehavioral processes via internal and/or external regulation. We envision that the proposed closed-loop HMI framework could provide personalized dNPI with enhanced effectiveness and scalability for cognitive enhancement in older adults, promoting brain resilience and healthy longevity in the aging population.

  View details for DOI 10.1016/j.arr.2025.102877

  View details for PubMedID 40850344

• Bridging Gaps in Sundown Syndrome Research: a Scoping Review and Roadmap for Future Multimodal Approaches. Archives of clinical neuropsychology : the official journal of the National Academy of Neuropsychologists Xu, Q., Lin, F. V., Liu, Y., Zhao, G. 2025

  Abstract

  Sundown syndrome (SS), or sundowning, is a neuropsychiatric phenomenon marked by the worsening of symptoms in the late afternoon or evening, primarily in individuals with dementia. By systematically examining previous studies, this scoping review aims to (1) bridge traditional questionnaire-based assessment methods with advanced sensor-based tools and (2) propose a multimodal framework to guide future research in enhancing risk identification, diagnosis, monitoring, and treatment across key symptom categories.We conducted a comprehensive review of Web of Science, PubMed, Medline, APA PsycInfo, and IEEE Xplore to identify studies on SS. Following established scoping review guidelines, 13 review papers and 41 empirical studies were selected and analyzed based on traditional questionnaire-based observation and/or sensor-based measurement methods.We identified key limitations in traditional assessment methods and classified SS symptoms into five domains: psychomotor symptoms, cognitive and perceptual disturbances, mood and affective symptoms, psychosis, and disruptions in activities of daily living and instrumental activities of daily living. Building on these insights, we proposed a multimodal platform integrating sensor technologies to enhance risk identification, diagnosis, continuous monitoring, and treatment.This study advances the understanding of SS by synthesizing prior research, refining symptom domains, and proposing a roadmap for future investigation and intervention. The integration of multimodal sensor technologies holds the potential to reduce caregiver burden, enhance patient care, and enable more effective management of SS and other behavioral disturbances in older adults.

  View details for DOI 10.1093/arclin/acaf062

  View details for PubMedID 40608498

• Dynamic class-balanced threshold Federated Semi-Supervised Learning by exploring diffusion model and all unlabeled data FUTURE GENERATION COMPUTER SYSTEMS-THE INTERNATIONAL JOURNAL OF ESCIENCE Wang, Z., Liu, Y., Liang, G., Zhong, C., Yang, F. 2025; 169

  View details for DOI 10.1016/j.future.2025.107820

  View details for Web of Science ID 001457481600001

• Bridging Cultures: A Framework for Facial Expression and Empathy Gong, Z., Liu, Y., Shi, H., Georgiev, G. V., Hakkila, J., IEEE IEEE. 2025

  View details for DOI 10.1109/ICMEW68306.2025.11152165

  View details for Web of Science ID 001588623300049

• Gender Fairness of Machine Learning Algorithms for Pain Detection Green, D., Shang, Y., Cheong, J., Liu, Y., Gunes, H., IEEE IEEE. 2025

  View details for DOI 10.1109/FG61629.2025.11099137

  View details for Web of Science ID 001554171100013

• Diffusion Model and Class-Balanced Adaptive Threshold for Federated Semi-supervised Non-IID Image Classification Liang, G., Liu, Y., Wang, Z., Zhao, Z., Yang, F. edited by Huang, D. S., Li, B., Chen, H., Zhang, C. SPRINGER-VERLAG SINGAPORE PTE LTD. 2025: 232-241

  View details for DOI 10.1007/978-981-96-9891-2_20

  View details for Web of Science ID 001569107100020

• Editorial: Towards Emotion AI to next generation healthcare and education. Frontiers in psychology Liu, Y., Kauttonen, J., Zhao, B., Li, X., Peng, W. 2024; 15: 1533053

  View details for DOI 10.3389/fpsyg.2024.1533053

  View details for PubMedID 39749281

  View details for PubMedCentralID PMC11694222

• Multi-consistency for semi-supervised medical image segmentation via diffusion models PATTERN RECOGNITION Chen, Y., Liu, Y., Lu, M., Fu, L., Yang, F. 2025; 161

  View details for DOI 10.1016/j.patcog.2024.111216

  View details for Web of Science ID 001373362000001

• Interactions for Socially Shared Regulation in Collaborative Learning: An Interdisciplinary Multimodal Dataset ACM TRANSACTIONS ON INTERACTIVE INTELLIGENT SYSTEMS Li, Y., Liu, Y., Nguyen, A., Shi, H., Vuorenmaa, E., Jarvela, S., Zhao, G. 2024; 14 (3)

  View details for DOI 10.1145/3658376

  View details for Web of Science ID 001325864200001

• A diffusion model multi-scale feature fusion network for imbalanced medical image classification research COMPUTER METHODS AND PROGRAMS IN BIOMEDICINE Zhu, Z., Liu, Y., Yuan, C., Qin, X., Yang, F. 2024; 256: 108384

  Abstract

  Medicine image classification are important methods of traditional medical image analysis, but the trainable data in medical image classification is highly imbalanced and the accuracy of medical image classification models is low. In view of the above two common problems in medical image classification. This study aims to: (i) effectively solve the problem of poor training effect caused by the imbalance of class imbalanced data sets. (ii) propose a network framework suitable for improving medical image classification results, which needs to be superior to existing methods.In this paper, we put in the diffusion model multi-scale feature fusion network (DMSFF), which mainly uses the diffusion generation model to overcome imbalanced classes (DMOIC) on highly imbalanced medical image datasets. At the same time, it is processed according to the cropped image augmentation strategy through cropping (IASTC). Based on this, we use the new dataset to design a multi-scale feature fusion network (MSFF) that can fully utilize multiple hierarchical features. The DMSFF network can effectively solve the problems of small and imbalanced samples and low accuracy in medical image classification.We evaluated the performance of the DMSFF network on highly imbalanced medical image classification datasets APTOS2019 and ISIC2018. Compared with other classification models, our proposed DMSFF network achieved significant improvements in classification accuracy and F1 score on two datasets, reaching 0.872, 0.731, and 0.906, 0.836, respectively.Our newly proposed DMSFF architecture outperforms existing methods on two datasets, and verifies the effectiveness of generative model inverse balance for imbalance class datasets and feature enhancement by multi-scale feature fusion. Further, the method can be applied to other class imbalanced data sets where the results will be improved.

  View details for DOI 10.1016/j.cmpb.2024.108384

  View details for Web of Science ID 001303138800001

  View details for PubMedID 39205335

• Benchmarking deep Facial Expression Recognition: An extensive protocol with balanced dataset in the wild ENGINEERING APPLICATIONS OF ARTIFICIAL INTELLIGENCE Tutuianu, G., Liu, Y., Alamaki, A., Kauttonen, J. 2024; 136

  View details for DOI 10.1016/j.engappai.2024.108983

  View details for Web of Science ID 001275005900001

• Unified Video and Image Representation for Boosted Video Face Forgery Detection Liu, H., Pan, C., Liu, Y., Zhao, G., Li, X. edited by Endriss, u., Melo, F. S., Bach, K., Bugarin-Diz, A., Alonso-Moral, J. M., Barro, S., Heintz, F. IOS PRESS. 2024: 673-680

  View details for DOI 10.3233/FAIA240548

  View details for Web of Science ID 001593512300087

• Uncertain Facial Expression Recognition via Multi-Task Assisted Correction IEEE TRANSACTIONS ON MULTIMEDIA Liu, Y., Zhang, X., Kauttonen, J., Zhao, G. 2024; 26: 2531-2543

  View details for DOI 10.1109/TMM.2023.3301209

  View details for Web of Science ID 001168330100032

• PFCFuse: A Poolformer and CNN Fusion Network for Infrared-Visible Image Fusion IEEE TRANSACTIONS ON INSTRUMENTATION AND MEASUREMENT Hu, X., Liu, Y., Yang, F. 2024; 73

  View details for DOI 10.1109/TIM.2024.3450061

  View details for Web of Science ID 001308263200031

• Graph-Based Facial Affect Analysis: A Review IEEE TRANSACTIONS ON AFFECTIVE COMPUTING Liu, Y., Zhang, X., Li, Y., Zhou, J., Li, X., Zhao, G. 2023; 14 (4): 2657-2677

  View details for DOI 10.1109/TAFFC.2022.3215918

  View details for Web of Science ID 001124163900026

• EXPLORING EFFECTIVE KNOWLEDGE DISTILLATION FOR TINY OBJECT DETECTION Liu, H., Liu, Q., Liu, Y., Liang, Y., Zhao, G., IEEE IEEE. 2023: 770-774

  View details for DOI 10.1109/ICIP49359.2023.10222589

  View details for Web of Science ID 001106821000148

• Deep Learning for Micro-Expression Recognition: A Survey IEEE TRANSACTIONS ON AFFECTIVE COMPUTING Li, Y., Wei, J., Liu, Y., Kauttonen, J., Zhao, G. 2022; 13 (4): 2028-2046

  View details for DOI 10.1109/TAFFC.2022.3205170

  View details for Web of Science ID 000892948500026

• Uncertain Label Correction via Auxiliary Action Unit Graphs for Facial Expression Recognition Liu, Y., Zhang, X., Kauttonen, J., Zhao, G., IEEE IEEE. 2022: 777-783

  View details for DOI 10.1109/ICPR56361.2022.9956650

  View details for Web of Science ID 000897707600108

• Learning the Connectivity: Situational Graph Convolution Network for Facial Expression Recognition Zhou, J., Zhang, X., Liu, Y., IEEE IEEE. 2020: 230-234

  View details for DOI 10.1109/vcip49819.2020.9301773

  View details for Web of Science ID 000718911500057

• FACIAL EXPRESSION RECOGNITION USING SPATIAL-TEMPORAL SEMANTIC GRAPH NETWORK Zhou, J., Zhang, X., Liu, Y., Lan, X., IEEE IEEE. 2020: 1961-1965

  View details for Web of Science ID 000646178502014