Accurate diagnosis of Alzheimer's disease (AD) requires handling tabular biomarker data, yet such data are often small and incomplete, where deep learning models frequently fail to outperform classical methods. Pretrained large language models (LLMs) offer few-shot generalization, structured reasoning, and interpretable outputs, providing a powerful paradigm shift for clinical prediction. We propose TAP-GPT Tabular Alzheimer's Prediction GPT, a domain-adapted tabular LLM framework built on TableGPT2 and fine-tuned for few-shot AD classification using tabular prompts rather than plain texts. We evaluate TAP-GPT across four ADNI-derived datasets, including QT-PAD biomarkers and region-level structural MRI, amyloid PET, and tau PET for binary AD classification. Across multimodal and unimodal settings, TAP-GPT improves upon its backbone models and outperforms traditional machine learning baselines in the few-shot setting while remaining competitive with state-of-the-art general-purpose LLMs. We show that feature selection mitigates degradation in high-dimensional inputs and that TAP-GPT maintains stable performance under simulated and real-world missingness without imputation. Additionally, TAP-GPT produces structured, modality-aware reasoning aligned with established AD biology and shows greater stability under self-reflection, supporting its use in iterative multi-agent systems. To our knowledge, this is the first systematic application of a tabular-specialized LLM to multimodal biomarker-based AD prediction, demonstrating that such pretrained models can effectively address structured clinical prediction tasks and laying the foundation for tabular LLM-driven multi-agent clinical decision-support systems. The source code is publicly available on GitHub: https://github.com/sophie-kearney/TAP-GPT.

Single-cell spatial transcriptomics has advanced spatial resolution from several cells per spot to hundreds of transcripts per cell, enabling a more comprehensive understanding of cellular interaction and local tissue organization. However, such high-resolution imaging introduces significant computational challenges, particularly in accurately segmenting cellular boundaries. Existing segmentation methods typically rely on a single modality, such as cellular imaging or transcript profiling, and thus fail to leverage the complementary information between modalities. Here we propose MSCA-Net, a Multi-Scale Convolutional Attention U-Net framework that integrates H&E staining images with selected transcriptomic features to achieve accurate cell boundary extraction. We evaluate MSCA-Net on dorsal root ganglia (DRG) neurons and demonstrate that it consistently outperforms state-of-the-art competing methods. Our study also shows that the reconstructed spatial transcriptomic slice can reproduce the downstream analysis consistent with prior knowledge, providing reliable and valuable insights for biological discovery.

Alzheimer's disease (AD) patients suffer from consequential diagnostic delay due to the lack of accessible biomarkers. They also show different responses to treatments due to disease heterogeneity and progression. Here, we developed a novel framework to identify disease progression and subtypes by using geometric brain signatures derived from multiple neuroimaging modalities, including [ 18 F]-Florbetapir (AV45) Positron Emission Tomography (PET), [ 18 F]-Fludeoxyglucose (FDG) PET, and structural Magnetic Resonance Imaging (MRI). These signatures were derived by decomposing corresponding maps of amyloid-beta levels, metabolic activity, and cortical thickness in terms of the fundamental, resonant modes--eigenmodes--of cortical geometry, each tied to a specific spatial resolution scale. Our results showed that geometric eigenmode-based features identified trajectories of disease progression, quantified as pseudotime, in distinct subtypes. The disease progression trajectories and subtypes are identified with high stability and are highly related to biological and cognitive measures. These performances are superior to those obtained using conventional localised features and remain robust across datasets, indicating that geometric signatures of brain structure and function can be used to uncover new markers of AD diagnosis and prognosis that are missed by conventional localisation approaches.

Large Language Models (LLMs) have shown significant promise for clinical applications, yet their application to triage remains underexplored. In this study, we systematically investigate the capabilities of LLMs in emergency department triage through two key dimensions: (1) robustness to distribution shifts and missing data, and (2) intersectional biases across sex and race. We assess multiple LLM-based approaches, ranging from continued pre-training to in-context learning, as well as conventional machine learning (ML) approaches. First, we demonstrate that LLMs exhibit superior robustness compared to traditional ML, which is promising due to their ability to provide explanatory rationales. Second, we show that the most effective LLM-based methods are those that select similar examples from prior patient cases, whereas reasoning capabilities in LLMs offer little benefit for triage. Lastly, we identify critical gaps in LLM preferences that emerge at the intersections of sex and race. LLMs exhibit sex-based differences, and they are more pronounced in certain racial groups, suggesting that LLMs encode preferences that emerge in specific clinical contexts and combinations of characteristics. We perform this audit through counterfactual analysis, providing a systematic way to identify such biases before real-world integration.

Accurately predicting molecular properties requires effective integration of structural information from both 2D molecular graphs and their corresponding equilibrium conformer ensembles. In this work, we propose FACET, a scalable Structure-Aware Graph Transformer that efficiently aggregates features from multiple 3D conformers while incorporating fragment-level information from 2D graphs. Unlike prior methods that rely on static geometric solvers or rigid fusion strategies, our approach utilizes a differentiable graph transformer to theoretically approximate the computationally expensive Fused Gromov–Wasserstein (FGW), enabling dynamic and scalable fusion of 2D and 3D structural information. We further enhance this mechanism by injecting fragment-specific structural priors into the attention layers, enabling the model to capture fine-grained molecular details. This unified design scales to large datasets, handling up to 75,000 molecules and hundreds of thousands of conformers, and provides over a 6× speedup compared to geometry-aware FGW-based baselines. Our method also achieves state-of-the-art results in molecular property prediction, Boltzmann-weighted ensemble modeling, and reaction-level tasks, and is particularly effective on chemically diverse compounds, including organocatalysts and transition-metal complexes.

Understanding the pathogenesis of amyloid- β pathology in Alzheimer's Disease (AD) proves to be a challenge. In this work, we expand upon the application of network diffusion models (NDM) to study pathophysiological spread of amyloid- β throughout white matter structural brain networks. We found that the NDM successfully recaptures subpopulation-level spatial patterns (Pearson's R = 0.45-0.48, PFDR < 0.01) of amyloid- β deposition in the Alzheimer's Disease Neuroimaging Cohort at a regional level, but with drawbacks in mechanism interpretability. We then moved to an extended NDM framework (eNDM), including a protein synthesis term to better reflect the role of amyloid- β metabolism, as well as including regional vulnerability using spatial transcriptomics from the Allen Human Brain Atlas to modulate the region-level rate parameters of the synthesis term. The novel gene eNDMs exhibited significant performance increases in Pearson's correlation (Steiger's Z, PFDR < 0.10) over baseline NDM performance in mild cognitive impairment and AD groups using APOE, SORL1, and FGL2 for gene modulation. The results were robust and replicable when testing on an external cohort of the Alzheimer's Disease Sequencing Project. The study thus demonstrates the importance of regional genetic vulnerability, in conjunction with network diffusion mechanisms, in improving the modelling and prediction of amyloid- β pathophysiological spread.

Accurate and early diagnosis of Alzheimer's disease (AD) is critical for effective intervention and requires integrating complementary information from multimodal neuroimaging data. However, conventional fusion approaches often rely on simple concatenation of features, which cannot adaptively balance the contributions of biomarkers such as amyloid PET and MRI across brain regions. In this work, we propose MREF-AD, a Multimodal Regional Expert Fusion model for AD diagnosis. It is a Mixture-of-Experts (MoE) framework that models mesoscopic brain regions within each modality as independent experts and employs a gating network to learn subject-specific fusion weights. Utilizing tabular neuroimaging and demographic information from the Alzheimer's Disease Neuroimaging Initiative (ADNI), MREF-AD achieves competitive performance over strong classic and deep baselines while providing interpretable, modality- and region-level insight into how structural and molecular imaging jointly contribute to AD diagnosis. The source code is available at https://github.com/PennShenLab/mref-ad.

Machine learning for medical image analysis has received unprecedented attention and success in the recent years. Yet, ensuring the reliability of discovered features for neuroimaging-based classification remains a challenge. In this study, we demonstrate the knockoff framework, a robust statistical approach that guarantees a theoretical bound on the False Discovery Rate (FDR), even under complex feature dependencies and model mis-specification. Unlike the Benjamini-Hochberg (BH) procedure, which assumes feature independence, knockoffs provide precise and provable FDR control for any feature correlation structure. We evaluate multiple knockoff construction methods on synthetic data for binary classification, observing near-perfect detection of true features and valid FDR control. We then apply the knockoff filter to fMRI data from the Human Connectome Project (HCP) to identify brain regions that contribute to an imaging-based classification model. The regions identified by the knockoff filter are found to be stable and reproducible across test-retest scans, indicating that they capture consistent task-related neural activation.

Accurate and early diagnosis of Alzheimer’s disease (AD) is critical for effective intervention and requires integrating complementary information from multimodal neuroimaging data. However, conventional fusion approaches often rely on simple concatenation of features, which cannot adaptively balance the contributions of biomarkers such as amyloid PET and MRI across brain regions. In this work, we propose MREF-AD, a Multimodal Regional Expert Fusion model for AD diagnosis. It is a Mixture-of-Experts (MoE) framework that models mesoscopic brain regions within each modality as independent experts and employs a gating network to learn subject-specific fusion weights. Utilizing tabular neuroimaging and demographic information from the Alzheimer’s Disease Neuroimaging Initiative (ADNI), MREF-AD achieves competitive performance over strong classic and deep baselines while providing interpretable, modality- and region-level insight into how structural and molecular imaging jointly contribute to AD diagnosis. The source code is available at https://github.com/PennShenLab/mref-ad.

Background and Objectives:Alzheimer's disease (AD) and AD-related dementias (ADRD) are expected to affect over 100 million people by 2050, placing a significant strain on public health systems. Social determinants of health (SDoH), which include factors such as socioeconomic conditions and environment, play a crucial role in AD risk. Despite growing evidence, the understanding of SDoH's impact on AD remains limited. Research Design and Methods:This study leverages large language models and knowledge graphs (KGs) to extract AD-related SDoH knowledge from literature and electronic health records (EHR). We integrate this knowledge into biological research on AD through KG construction and graph deep learning, performing KG-link predictions validated by multimodal biological data from single-cell RNA-seq and proteomics. Results:We generated an SDoH knowledge graph with around 92k triplets, integrating literature and EHR data. In various link prediction experiments, we observed higher accuracy when integrating SDoH into knowledge graphs. Additionally, exploratory predictions uncovered potential SDoH-gene interactions, many of which were validated through differential expression analysis using proteomics and RNA-seq data. Discussion and Implications:This novel KG-based analysis enhances link prediction in AD-related biomedical networks by integrating SDoH and biological knowledge. Our findings highlight the potential interaction between social determinants and biological factors in AD, offering insights into more personalized and socially aware healthcare interventions.

This paper studies the data-driven control of unknown linear-threshold network dynamics to stabilize the state to a reference value. We consider two types of controllers: (i) a state feedback controller with feed-forward reference input and (ii) an augmented feedback controller with error integration. The first controller features a simpler structure and is easier to design, while the second offers improved performance in the presence of system parameter changes and disturbances. Our design strategy employs state-input datasets to construct data-based representations of the closed-loop dynamics. Since these representations involve linear threshold functions, we rewrite them as switched linear systems, and formulate the design problem as that of finding a common controller for all the resulting modes. This gives rise to a set of linear matrix inequalities (LMIs) whose solutions corresponds to the controller gain matrices. We analyze the computational complexity of solving the LMIs and propose a simplified, sufficient set of conditions that scales linearly with the system state. Simulations on two case studies involving regulation of firing rate dynamics in rodent brains and of arousal level dynamics in humans demonstrate the effectiveness of the controller designs.

Recent advances in cytometry have enabled high-throughput data collection with multiple single-cell protein expression measurements. The significant biological and technical variance in cytometry has posed a formidable challenge during the gating process, especially for the initial pre-gates which deal with unpredictable events, such as debris and technical artifacts. To mitigate the labor-intensive manual gating process, we propose UNITO, a framework to rigorously identify the hierarchical cytometric subpopulations. UNITO transforms a cell-level classification task into an image-based segmentation problem. The framework is validated on three independent cohorts (two mass cytometry and one flow cytometry datasets). We compare its results with previous automated methods using the consensus of at least four experienced immunologists. UNITO outperforms existing methods and deviates from human consensus by no more than any individual does. UNITO can reproduce a similar contour compared to manual gating for post-hoc inspection, and it also allows parallelization of samples for faster processing.

The accurate identification of individuals from functional connectomes (FCs) is central to individualized neuro/psychiatric assessment. Traditional metrics (Pearson and Euclidean) fail to capture the non-Euclidean geometry of FCs, and geodesic metrics (affine-invariant and Log-Euclidean) require task- and scale-specific regularization and degrade in high-dimensional settings. To address these challenges, we propose the Alpha-Z Bures-Wasserstein divergence, a geometry-aware divergence for FC comparison that operates effectively without meticulous parameter tuning. Across Human Connectome Project tasks, scan lengths, and spatial resolutions, we benchmark Alpha-Z against classical and state-of-the-art manifold-based distances and quantify how varying regularization influences geodesic performance. Alpha-Z yields consistently higher identification rates, with pronounced advantages in rank-deficient regimes, and preserves performance across parcellations and conditions. We further verify generalization across resting-state and task fMRI under multiple parcellation schemes. These results position Alpha-Z as a reliable, robust, and scalable framework for functional connectivity analysis, improving sensitivity to cognitive and behavioral patterns and offering strong potential for individualized clinical neuroscience.

Neuroimaging studies have demonstrated that Alzheimer's disease (AD) is closely related to changes in neuroanatomy in the form of damage to both grey matter and white matter. However, the exact nature of AD's relationship with white matter anatomical deterioration is not fully understood at a systemic level. To investigate this knowledge gap, we constructed structural brain networks from ADNI-GO/2 diffusion tensor imaging (DTI) images with brain regions of interest (ROIs) as nodes and white matter connections as edges weighted by fiber density. The cohort consists of healthy control (HC), mild cognitive impairment (MCI), and clinically diagnosed AD subjects. By optimizing consensus modularity of structural brain networks at a subpopulation level to investigate community structure throughout a range of resolution parameters (γ), we observed a split of the reward-based decision-making module in the AD group at γ = 1.3, thus finding a 7th consensus community in the AD consensus brain network partition that was not present in that of MCI or HC populations. Upon further investigation, we found that thalamic and caudal regions were involved in the increased segregation of AD brain networks. These regions are implicated in regulation of decision-making processes, and their segregation from other decision-making regions is a novel finding in white matter biomarker studies of AD. Our study presents novel evidence that AD may be a disconnection syndrome at the mesoscopic structural level, with potential new avenues of exploration into the role of the thalamus and caudate that may reveal neural correlates of cognitive deficits in clinically diagnosed AD.

IntroductionUnderstanding the role of various brain regions of interest (ROIs) in various cognitive functions or tasks, across healthy or neurodegenerative conditions and multiple degrees of separation, remains a key challenge in neuroscience. Conventional network measures can only capture localized or quasi-localized features of brain ROIs. Topological data analysis (TDA), particularly persistent homology, provides a threshold-free, mathematically rigorous framework for identifying topologically salient features in complex networks. In this paper, we introduce a new metric, the Homological Vertex Importance Profile (H-VIP), designed to assess the relevance of vertices that participate in persistent topological structures (e.g., connected components, cycles or cavities) in brain networks. The H-VIP quantifies the topological features of the network at the ROI (node) level by compressing its higher-order connectivity profile using homological constructs.MethodsLeveraging homological constructs of brain connectomes, we extend two of our previously defined network-level measures—average persistence and persistence entropy—to an ROI-level measure, i.e., the H-VIP. We then applied the H-VIP to two independent datasets: structural connectomes from the Human Connectome Project and functional connectomes from the Alzheimer’s Disease Neuroimaging Initiative. Persistent homology was computed for each network, and H-VIP scores were derived to evaluate vertex-level contributions. Finally, H-VIP scores were used for the prediction of multiple cognitive measures.ResultsIn both anatomical and functional brain networks, H-VIP values demonstrate predictive power for various cognitive measures. Notably, the connectivity of the frontal lobe exhibited stronger correlations with cognitive performance than the whole-brain network.DiscussionH-VIP offers a robust and interpretable means to locate, quantify, and visualize region-specific contributions to network’s topological, higher-order landscape. Its ability to detect potentially impaired connectivity at the individual level suggests possible applications in personalized medicine for neurological diseases and disorders. Beyond brain connectomics, the H-VIP can be used for other types of complex networks where topological features are of importance, such as financial, social, or ecological networks.

The accurate identification of individuals from functional connectomes (FCs) is critical for advancing individualized assessments in neuropsychiatric research. Traditional methods, such as Pearson's correlation, have limitations in capturing the complex, non-Euclidean geometry of FC data, leading to suboptimal performance in identification performance. Recent developments have introduced geodesic distance as a more robust metric; however, its performance is highly sensitive to regularization choices, which vary by spatial scale and task condition. To address these challenges, we propose a novel divergence-based distance metric, the Alpha-Z Bures-Wasserstein divergence, which provides a more flexible and geometry-aware framework for FC comparison. Unlike prior methods, our approach does not require meticulous parameter tuning and maintains strong identification performance across multiple task conditions and spatial resolutions. We evaluate our approach against both traditional (e.g., Euclidean, Pearson) and state-of-the-art manifold-based distances (e.g., affine-invariant, log-Euclidean, Bures-Wasserstein), and systematically investigate how varying regularization strengths affect geodesic distance performance on the Human Connectome Project dataset. Our results show that the proposed method significantly improves identification rates over traditional and geodesic distances, particularly when optimized regularization is applied, and especially in high-dimensional settings where matrix rank deficiencies degrade existing metrics. We further validate its generalizability across resting-state and task-based fMRI, using multiple parcellation schemes. These findings suggest that the new divergence provides a more reliable and generalizable framework for functional connectivity analysis, offering enhanced sensitivity in linking FC patterns to cognitive and behavioral outcomes.

Brain imaging genomics has manifested considerable potential in illuminating the genetic determinants of human brain structure and function. This has propelled us to develop the GIANT (Genetically Informed brAiN aTlas) that accounts for genetic and neuroanatomical variations simultaneously. Integrating voxel-wise heritability and spatial proximity, GIANT clusters brain voxels into genetically informed regions, while retaining fundamental anatomical knowledge. Compared to conventional (non-genetics) brain atlases, GIANT exhibits smaller intra-region variations and larger inter-region variations in terms of voxel-wise heritability. As a result, GIANT yields increased regional SNP heritability, enhanced polygenicity, and its polygenic risk score explains more brain volumetric variation than traditional neuroanatomical brain atlases. We provide extensive validation to GIANT and demonstrate its neuroanatomical validity, confirming its generalizability across populations with diverse genetic ancestries and various brain conditions. Furthermore, we present a comprehensive genetic architecture of the GIANT regions, covering their functional annotation at the molecular levels, their associations with other complex traits/diseases, and the genetic and phenotypic correlations among GIANT-defined imaging endophenotypes. In summary, GIANT constitutes a brain atlas that captures the complexity of genetic and neuroanatomical heterogeneity, thereby enhancing the discovery power and applicability of imaging genomics investigations in biomedical science.

Trait anxiety is an individual disposition marked by heightened anticipation of potential threats under uncertainty. It has been associated with allostatic load, the cumulative physiological cost of chronic stress, suggesting that enduring anxiety vulnerability emerges from brain and body interactions across multiple scales. Yet these domains have largely been examined separately and the temporal dynamics of brain activity remain underexplored. Using data from healthy young adults (LEMON cohort, N = 120), we implemented a graph attention framework integrating low frequency (slow4 and slow5) fMRI dynamics, structural connectivity and systemic biomarkers through cross modal attention to predict individual trait anxiety outcome (The State Trait Anxiety Inventory). Temporal modeling significantly enhanced prediction compared with static or amplitude based features, highlighting the importance of time resolved neural information. Model derived importance mapping identified the limbic and visual systems as core predictive networks. Dynamic functional connectivity revealed that higher trait anxiety was associated with longer occupancy of states marked by strong limbic default mode frontoparietal coupling and shorter occupancy of visually decoupled states. Metabolic and immune markers further contributed to prediction and transcriptomic enrichment linked these networks to neurodevelopmental and synaptic signaling pathways. Together, these findings delineate a temporally dynamic brain and body architecture underlying stable anxiety vulnerability. ### Competing Interest Statement The authors have declared no competing interest. ### Funding Statement This study did not receive any funding ### Author Declarations I confirm all relevant ethical guidelines have been followed, and any necessary IRB and/or ethics committee approvals have been obtained. Yes The details of the IRB/oversight body that provided approval or exemption for the research described are given below: The study used ONLY openly available human data that were originally located at : https://fcon\_1000.projects.nitrc.org/indi/retro/MPI\_LEMON.html I confirm that all necessary patient/participant consent has been obtained and the appropriate institutional forms have been archived, and that any patient/participant/sample identifiers included were not known to anyone (e.g., hospital staff, patients or participants themselves) outside the research group so cannot be used to identify individuals. Yes I understand that all clinical trials and any other prospective interventional studies must be registered with an ICMJE-approved registry, such as ClinicalTrials.gov. I confirm that any such study reported in the manuscript has been registered and the trial registration ID is provided (note: if posting a prospective study registered retrospectively, please provide a statement in the trial ID field explaining why the study was not registered in advance). Yes I have followed all appropriate research reporting guidelines, such as any relevant EQUATOR Network research reporting checklist(s) and other pertinent material, if applicable. Yes All data produced are available online at https://fcon\_1000.projects.nitrc.org/indi/retro/MPI\_LEMON.html [https://fcon\_1000.projects.nitrc.org/indi/retro/MPI\_LEMON.html][1] [1]: https://fcon_1000.projects.nitrc.org/indi/retro/MPI_LEMON.html

Early amyloid-β deposition is a hallmark of Alzheimer's disease (AD), though the exact nature of amyloid pathogenesis is not fully characterized. In this study, we designed a network diffusion model to simulate the spread of amyloid pathology through white matter brain networks of diagnostic subpopulations of healthy control (HC), mild cognitive impairment (MCI), and AD. Our network diffusion model was able to successfully model the spread of amyloid, recapturing regional distributions of amyloid observed in 18F-florbetapir positron emission tomography (r=0.44-0.46, P<0.01). When tuning the optimal parameters, we found that the optimal diffusion time (t) provided a notion of temporal progression, where the HC group had the lowest time (t = 107.22 ± 16.67), followed by MCI (t = 122.78 ± 19.63), and lastly AD (t =136.20 ± 24.47). The optimal starting seeds were the brainstem in all three diagnostic groups, followed by the lateral orbitofrontal lobes for HC and MCI and the lingual gyri in AD. Our findings corroborate evidence from amyloid staging studies where amyloid starts in the primary neocortex and associative cortex. The significance of the white matter structural network in the diffusion process provides evidence for the trans-synaptic spread hypothesis of amyloid in AD. In conclusion, our study provides novel insights into the pathogenesis of amyloid in AD and its subsequent propagation throughout the brain.

Predicting phenotypes with complex genetic bases based on a small, interpretable set of variant features remains a challenging task. Conventionally, data-driven approaches are used for this task, yet the high dimensional nature of genotype data makes the analysis and prediction difficult. Motivated by the biomedical knowledge encoded in pre-trained LLMs and the emerging applications for genetics, we set to examine the ability of LLMs in feature selection and engineering for tabular genotype data, with a novel knowledge-driven framework. We develop FREEFORM, Free-flow Reasoning and Ensembling for Enhanced Feature Output and Robust Modeling, designed with chain-of-thought and ensembling principles, to select and engineer features with the intrinsic knowledge of LLMs. Evaluated on two distinct genotype-phenotype datasets, genetic ancestry and hereditary hearing loss, we find this framework outperforms several data-driven methods, particularly on low-data regimes. FREEFORM is available as open-source framework at GitHub: https://github.com/PennShenLab/FREEFORM.