Corey W. Arnold

BACKGROUND: Microvascular hemodynamic deterioration is a proposed mechanism underlying the no-reflow phenomenon in acute ischemic stroke. Experimental studies suggested that relative heterogeneity of capillary transit time (RTH) is a potential imaging biomarker of microvascular impairment. We investigated the relationship between peri-infarct RTH and infarct growth, measured on immediate post-thrombectomy (PT) perfusion magnetic resonance imaging (MRI), and functional outcomes in patients with acute ischemic stroke with successful reperfusion. METHODS: In this retrospective case-control study from the University of California, Los Angeles, patients with acute ischemic stroke from 2015 to 2020 were included if they had excellent reperfusion (modified Thrombolysis in Cerebral Infarction score ≥2c) and diffusion- and perfusion-weighted MRIs at baseline, immediately PT, and 24 to 48 hours after thrombectomy. Peri-infarct regions were defined as time to maximum ≥6-second regions on baseline MRI that were not infarcted on immediate PT MRI. Following image coregistration, quantitative values of perfusion parameters, including RTH, were obtained from peri-infarct regions on immediate PT MRI. Two primary outcomes were assessed: (1) infarct growth ≥10 mL and (2) poor functional outcome (90-day modified Rankin Scale score >2). Associations between clinical and imaging parameters and outcomes were evaluated using univariate analysis. Age, sex, and covariates that were significant in univariate analysis were included in logistic regression analyses. RESULTS: Among 47 patients included in the analysis (62% women; mean age, 72±13 years; median National Institutes of Health Stroke Scale score, 14), 20 (43%) patients had substantial infarct growth and poor functional outcome. Peri-infarct RTH was significantly ( P =0.005) higher in patients with infarct growth >10 mL than those without and remained as an independent predictor (odds ratio, 1.4 per 10% increase [95% CI, 1.0–2.0]; P =0.04). Similarly, the peri-infarct RTH was significantly higher ( P P =0.03). CONCLUSIONS: Increased peri-infarct RTH on immediate PT MRI predicts substantial infarct growth and poor functional outcomes in successfully reperfused patients with acute ischemic stroke. Results highlight the role of microvascular dysfunction depicted by elevated RTH as a potential marker of no-reflow phenomenon despite excellent macrovascular reperfusion.

Following successful large-vessel recanalization via endovascular thrombectomy (EVT) for acute ischemic stroke (AIS), some patients experience a complication known as no-reflow, defined by persistent microvascular hypoperfusion that undermines tissue recovery and worsens clinical outcomes. Although prompt identification is crucial, standard clinical practice relies on perfusion magnetic resonance imaging (MRI) within 24 hours post-procedure, delaying intervention. In this work, we introduce the first-ever machine learning (ML) framework to predict no-reflow immediately after EVT by leveraging previously unexplored intra-procedural digital subtraction angiography (DSA) sequences and clinical variables. Our retrospective analysis included AIS patients treated at UCLA Medical Center (2011–2024) who achieved favorable mTICI scores (2c or 3) and underwent pre- and post-procedure MRI. No-reflow was defined as a > 15% reduction in relative cerebral blood volume or flow within the infarct core compared to the contralateral hemisphere. From DSA sequences (anteroposterior and lateral views), we extracted statistical and temporal perfusion features from the target downstream territory to train ML classifiers for predicting no-reflow. Our preliminary results demonstrate that this novel method out-performed a clinical-features baseline (AUROC: 0.9330 vs. 0.7768 (p = 0.006)), suggesting that real-time DSA perfusion dynamics may encode clinically relevant information related to microvascular integrity. This approach establishes a preliminary foundation for immediate, accurate no-reflow prediction, enabling clinicians to proactively manage high-risk patients without reliance on delayed imaging, though it warrants validation in larger, independent cohorts.

Objectives: Heart failure is a leading cause of mortality, necessitating identification of patients at increased risk needing intervention. In this study, we investigated if Fitbit data can reveal physiological trends associated with hospital visit risk. Materials and methods: Individuals with heart failure (n=249) were randomized into three arms for prospective 180-day monitoring. All arms received a Fit-bit and wireless weight scale. Arm 1 received devices only; Arm 2 received a mobile app with surveys; Arm 3 received the app plus financial incentives. Results: 51 participants had hospital visits during the study period. These individuals took fewer steps (p=.002) and reported increased symptom severity (p=.044). Resting heart rate increased three days prior to a visit (p=.022). Baseline steps revealed a higher visit probability for less active participants (p=.003). Discussion and conclusion: Passive physiological monitoring can effectively identify individuals at risk of health exacerbation, demonstrating the potential of wearable devices for timely clinical intervention.

Neuromodulation targeting thalamic nuclei is increasingly used to treat drug-resistant focal epilepsy, yet human intracranial EEG studies describing how thalamocortical interactions evolve across seizures remain limited. We aimed to define frequency-specific thalamocortical network dynamics from seizure onset to termination, compare thalamocortical and cortico-cortical network activation, and test whether thalamic EEG features can classify seizure state to inform closed-loop or adaptive thalamic stimulation strategies. We retrospectively analyzed chronic stereo-EEG recordings from 19 patients with pediatric-onset, drug-resistant focal epilepsy (6 females; age at thalamic recording 1.0–28.1 years, median 16.9) with cortical and thalamic sampling. Sixty-six focal seizures were included. Spectral power, imaginary coherence, and spectral Granger causality were computed in non-overlapping two-second windows across slow (1–12 Hz), beta (13–30 Hz), and gamma (30–70 Hz) bands and compared with an interictal baseline. Random forest classifiers were trained using thalamic spectral power and thalamocortical connectivity features to distinguish ictal from non-ictal states using leave-one-patient-out cross-validation, with Shapley additive explanations used for feature attribution. Visual analysis identified thalamic ictal involvement at seizure onset in 82/101 thalamic contacts (81.2%), increasing to near-universal involvement by seizure termination, with onset-to-termination patterns dominated by low-voltage fast activity at onset and rhythmic spike or rhythmic slow-wave patterns at termination. The thalamus and cortical seizure onset zone exhibited broadband power increases at seizure onset that attenuated toward termination, while slow- and beta-band thalamocortical connectivity increased throughout seizures and peaked around the end-of-seizure epoch. Directed connectivity demonstrated bidirectional thalamocortical coupling, with slow-frequency thalamus-to-seizure onset zone outflow exceeding propagation-zone-to-seizure onset zone cortico-cortical outflow during both ictal and end-of-seizure epochs (anterior nucleus: p = 9.06 × 10L ³ and p = 8.80 × 10L ³ ; centromedian nucleus: p = 3.30 × 10L ³ and p = 5.70 × 10L ³ ). Seizure state was classifiable from thalamic spectral power and thalamocortical network features, achieving an area under the receiver operating characteristic curve of 0.825 ± 0.163 (anterior nucleus model) and 0.839 ± 0.149 (centromedian nucleus model), with thalamic broadband power plus slow-frequency thalamus-to-cortex outflow and beta-frequency cortex-to-thalamus inflow among the most informative features. Leveraging human intracranial EEG data, we define coordinated, frequency- and direction-specific thalamocortical and cortico-cortical network dynamics that evolve from seizure onset to termination. These findings establish a mechanistic basis and identify actionable thalamocortical EEG targets—particularly slow- and beta-band interactions—to inform individualized, adaptive, closed-loop neuromodulation aimed at optimizing seizure outcomes.

Current histopathological grading of prostate cancer relies primarily on glandular architecture, largely overlooking the tumor microenvironment. Here, we present PROTAS, a deep learning framework that quantifies reactive stroma (RS) in routine hematoxylin and eosin (H&E) slides and links stromal morphology to underlying biology. PROTAS-defined RS is characterized by nuclear enlargement, collagen disorganization, and transcriptomic enrichment of contractile pathways. PROTAS detects RS robustly in the external Prostate, Lung, Colorectal, and Ovarian (PLCO) dataset and, using domain-adversarial training, generalizes to diagnostic biopsies. In head-to-head comparisons, PROTAS outperforms pathologists for RS detection, and spatial RS features predict biochemical recurrence independently of established prognostic variables (c-index 0.80). By capturing subtle stromal phenotypes associated with tumor progression, PROTAS provides an interpretable, scalable biomarker to refine risk stratification.

Background: For acute ischemic stroke (AIS) patients, successful endovascular thrombectomy (EVT) may not recover functional independence. One potential mechanism driving this disparity is no-reflow, a lack of perfusion despite recanalization of the large vessel occlusion (LVO). None of the published methods for assessing no-reflow have gained acceptance, and debate persists about isolating no-reflow from upstream hemodynamic anomalies as well as the temporal dynamics no-reflow assessment. This work conducts an exploratory analysis of the effect of time since EVT on the performance of different methods for assessing no-reflow. Methods: We identified a cohort of UCLA Health patients with successful LVO recanalization (≥ mTICI 2b) of a thrombus located in M1 or ICA. Patients were stratified into ‘early’ and ‘follow-up’ cohorts of < 24 and < 48 hours since EVT. A set of published methods using perfusion imaging for no-reflow detection were adapted to run semi-automatically for both cohorts. Results: We compiled 83 imaging studies from 64 patients. No-reflow rates varied from 4 . 2 % to 51 . 4 % by method, time since EVT, and mTICI. Follow-up no-reflow rates more closely match the original literature across all methods. mTICI 2b patients likely inflated no-reflow, as rates dropped approximately 50 % for mTICI 2c/3 cases. Early cohort inter-method agreement shows kappa scores as low as − 0 . 017 , but improved up to 0 . 72 in the follow-up cohort. Conclusion: No-reflow exhibits substantial dependence on time across all no-reflow detection methods studied. Inter-method agreement is good for methods using the same perfusion parameters, but each method is subject to confounding influences, motivating an aggregate method for robust evaluation. The temporal dynamics of no-reflow detection raises questions about imaging timing post-EVT for detection within the intervention window.

Objectives: Longitudinal data in electronic health records (EHRs) represent an individual’s clinical history through a sequence of codified concepts, including diagnoses, procedures, medications, and laboratory tests. Generative pretrained transformers (GPT) can leverage this data to predict future events. While fine-tuning of these models can enhance task-specific performance, it becomes costly when applied to many clinical prediction tasks. In contrast, a pretrained foundation model can be used in zero-shot forecasting setting, offering a scalable alternative to fine-tuning separate models for each outcome. Materials and Methods: This study presents the first comprehensive analysis of zero-shot forecasting with GPT-based foundational models in EHRs, introducing a novel pipeline that formulates medical concept prediction as a generative modeling task. Unlike supervised approaches requiring extensive labeled data, our method enables the model to forecast the next medical event purely from a pretraining knowledge. We evaluate performance across multiple time horizons and clinical categories, demonstrating model’s ability to capture latent temporal dependencies and complex patient trajectories without task supervision. Results: The model’s performance in predicting the next medical concept was evaluated using precision and recall metrics, achieving an average top-1 precision of 0.614 and recall of 0.524. For 12 major diagnostic conditions, the model demonstrated strong zero-shot performance, achieving high true positive rates while maintaining low false positives. Discussion: We demonstrate the power of a foundational EHR GPT model in capturing diverse phenotypes and enabling robust, zero-shot forecasting of clinical outcomes. This capability highlights both its versatility across conditions like liver cancer and SLE, and its limitations in more ambiguous settings such as depression, while also revealing meaningful latent clinical structure. Conclusion: This capability enhances the versatility of predictive healthcare models and reduces the need for task-specific training, enabling more scalable applications in clinical settings.

Cytology is essential for cancer diagnostics and screening due to its minimally invasive nature. However, the development of robust deep learning models for digital cytology is challenging due to the heterogeneity in staining and preparation methods of samples, differences across organs, and the limited availability of large, diverse, annotated datasets. Developing a task-specific model for every cytology application is impractical and non-cytology-specific foundation models struggle to generalize to tasks in this domain where the emphasis is on cell morphology. To address these challenges, we introduce CytoFM, the first cytology self-supervised foundation model. Using iBOT, a self-supervised Vision Transformer (ViT) training framework incorporating masked image modeling and self-distillation, we pretrain CytoFM on a diverse collection of cytology datasets to learn robust, transferable representations. We evaluate CytoFM on multiple downstream cytology tasks, including breast cancer classification and cell type identification, using an attention-based multiple instance learning framework. Our results demonstrate that CytoFM performs better on two out of three downstream tasks than existing foundation models pretrained on histopathology (UNI) or natural images (iBOT-Imagenet). Visualizations of learned representations demonstrate our model is able to attend to cytologically relevant features. Despite a small pre-training dataset, CytoFM's promising results highlight the ability of task-agnostic pre-training approaches to learn robust and generalizable features from cytology data.