T he scientific, academic, medical and data science communities have come together in the face of the COVID-19 pandemic crisis to rapidly assess novel paradigms in artificial intelligence (AI) that are rapid and secure, and potentially incentivize data sharing and model training and testing without the usual privacy and data ownership hurdles of conventional collaborations 1,2 . Healthcare providers, researchers and industry have pivoted their focus to address unmet and critical clinical needs created by the crisis, with remarkable results [3][4][5][6][7][8][9] . Clinical trial recruitment has been expedited and facilitated by national regulatory bodies and an international cooperative spirit 10-12 . The data analytics and AI disciplines have always fostered open

Physicians make critical time-constrained decisions every day. Clinical predictive models can help physicians and administrators make decisions by forecasting clinical and operational events. Existing structured data-based clinical predictive models have limited use in everyday practice owing to complexity in data processing, as well as model development and deployment1–3. Here we show that unstructured clinical notes from the electronic health record can enable the training of clinical language models, which can be used as all-purpose clinical predictive engines with low-resistance development and deployment. Our approach leverages recent advances in natural language processing4,5 to train a large language model for medical language (NYUTron) and subsequently fine-tune it across a wide range of clinical and operational predictive tasks. We evaluated our approach within our health system for five such tasks: 30-day all-cause readmission prediction, in-hospital mortality prediction, comorbidity index prediction, length of stay prediction, and insurance denial prediction. We show that NYUTron has an area under the curve (AUC) of 78.7–94.9%, with an improvement of 5.36–14.7% in the AUC compared with traditional models. We additionally demonstrate the benefits of pretraining with clinical text, the potential for increasing generalizability to different sites through fine-tuning and the full deployment of our system in a prospective, single-arm trial. These results show the potential for using clinical language models in medicine to read alongside physicians and provide guidance at the point of care.

Background Large-scale Granger causality (lsGC) is a recently developed, resting-state functional MRI (fMRI) connectivity analysis approach that estimates multivariate voxel-resolution connectivity. Unlike most commonly used multivariate approaches, which establish coarse-resolution connectivity by aggregating voxel time-series avoiding an underdetermined problem, lsGC estimates voxel-resolution, fine-grained connectivity by incorporating an embedded dimension reduction. New Method We investigate application of lsGC on realistic fMRI simulations, modeling smoothing of neuronal activity by the hemodynamic response function and repetition time (TR), and empirical resting-state fMRI data. Subsequently, functional subnetworks are extracted from lsGC connectivity measures for both datasets and validated quantitatively. We also provide guidelines to select lsGC free parameters. Results Results indicate that lsGC reliably recovers underlying network structure with Area Under receiver operator characteristic Curve (AUC) of 0.93 at TR=1.5s for a 10-minute session of fMRI simulations. Furthermore, subnetworks of closely interacting modules are recovered from the aforementioned lsGC networks. Results on empirical resting-state fMRI data demonstrate recovery of visual and motor cortex in close agreement with spatial maps obtained from (i) visuo-motor fMRI stimulation task-sequence (Accuracy=0.76) and (ii) independent component analysis (ICA) of resting-state fMRI (Accuracy=0.86). Comparison with Existing Method(s) Compared with conventional Granger causality approach (AUC=0.75), lsGC produces better network recovery on fMRI simulations. Furthermore, it cannot recover functional subnetworks from empirical fMRI data, since quantifying voxel-resolution connectivity is not possible as consequence of encountering an underdetermined problem. Conclusions Functional network recovery from fMRI data suggests that lsGC gives useful insight into connectivity patterns from resting-state fMRI at a multivariate voxel-resolution.

HIV is capable of invading the brain soon after seroconversion. This ultimately can lead to deficits in multiple cognitive domains commonly referred to as HIV-associated neurocognitive disorders (HAND). Clinical diagnosis of such deficits requires detailed neuropsychological assessment but clinical signs may be difficult to detect during asymptomatic injury of the central nervous system (CNS). Therefore neuroimaging biomarkers are of particular interest in HAND. In this study, we constructed brain connectivity profiles of 40 subjects (20 HIV positive subjects and 20 age-matched seronegative controls) using two different methods: a non-linear mutual connectivity analysis approach and a conventional method based on Pearson's correlation. These profiles were then summarized using graph-theoretic methods characterizing their topological network properties. Standard clinical and laboratory assessments were performed and a battery of neuropsychological (NP) tests was administered for all participating subjects. Based on NP testing, 14 of the seropositive subjects exhibited mild neurologic impairment. Subsequently, we analyzed associations between the network derived measures and neuropsychological assessment scores as well as common clinical laboratory plasma markers (CD4 cell count, HIV RNA) after adjusting for age and gender. Mutual connectivity analysis derived graph-theoretic measures, Modularity and Small Worldness, were significantly (p < 0.05, FDR adjusted) associated with the Executive as well as Overall z-score of NP performance. In contrast, network measures derived from conventional correlation-based connectivity did not yield any significant results. Thus, changes in connectivity can be captured using advanced time-series analysis techniques. The demonstrated associations between imaging-derived graph-theoretic properties of brain networks with neuropsychological performance, provides opportunities to further investigate the evolution of HAND in larger, longitudinal studies. Our analysis approach, involving non-linear time-series analysis in conjunction with graph theory, is promising and it may prove to be useful not only in HAND but also in other neurodegenerative disorders.

Phase contrast X-ray computed tomography (PCI-CT) has been demonstrated to be effective for visualization of the human cartilage matrix at micrometer resolution, thereby capturing osteoarthritis induced changes to chondrocyte organization. This study aims to systematically assess the efficacy of deep transfer learning methods for classifying between healthy and diseased tissue patterns. We extracted features from two different convolutional neural network architectures, CaffeNet and Inception-v3 for characterizing such patterns. These features were quantitatively evaluated in a classification task measured by the area (AUC) under the Receiver Operating Characteristic (ROC) curve as well as qualitative visualization through a dimension reduction approach t-Distributed Stochastic Neighbor Embedding (t-SNE). The best classification performance, for CaffeNet, was observed when using features from the last convolutional layer and the last fully connected layer (AUCs >0.91). Meanwhile, off-the-shelf features from Inception-v3 produced similar classification performance (AUC >0.95). Visualization of features from these layers further confirmed adequate characterization of chondrocyte patterns for reliably distinguishing between healthy and osteoarthritic tissue classes. Such techniques, can be potentially used for detecting the presence of osteoarthritis related changes in the human patellar cartilage.

We explore a computational framework for functional connectivity analysis in resting-state functional MRI (fMRI) data acquired from the human brain for recovering the underlying network structure and understanding causality between network components. Termed mutual connectivity analysis (MCA), this framework involves two steps, the first of which is to evaluate the pair-wise cross-prediction performance between fMRI pixel time series within the brain. In a second step, the underlying network structure is subsequently recovered from the affinity matrix using non-metric network clustering approaches, such as the so-called Louvain method. Finally, we use convergent cross-mapping (CCM) to study causality between different network components. We demonstrate our MCA framework in the problem of recovering the motor cortex network associated with hand movement from resting state fMRI data. Results are compared with a ground truth of active motor cortex regions as identified by a task-based fMRI sequence involving a finger-tapping stimulation experiment. Our results regarding causation between regions of the motor cortex revealed a significant directional variability and were not readily interpretable in a consistent manner across subjects. However, our results on whole-slice fMRI analysis demonstrate that MCA-based model-free recovery of regions associated with the primary motor cortex and supplementary motor area are in close agreement with localization of similar regions achieved with a task-based fMRI acquisition. Thus, we conclude that our MCA methodology can extract and visualize valuable information concerning the underlying network structure between different regions of the brain in resting state fMRI.

‘Federated Learning’ (FL) is a method to train Artificial Intelligence (AI) models with data from multiple sources while maintaining anonymity of the data thus removing many barriers to data sharing. During the SARS-COV-2 pandemic, 20 institutes collaborated on a healthcare FL study to predict future oxygen requirements of infected patients using inputs of vital signs, laboratory data, and chest x-rays, constituting the “EXAM” (EMR CXR AI Model) model. EXAM achieved an average Area Under the Curve (AUC) of over 0.92, an average improvement of 16%, and a 38% increase in generalisability over local models. The FL paradigm was successfully applied to facilitate a rapid data science collaboration without data exchange, resulting in a model that generalised across heterogeneous, unharmonized datasets. This provided the broader healthcare community with a validated model to respond to COVID-19 challenges, as well as set the stage for broader use of FL in healthcare.

HIV-associated neurocognitive disorders (HAND) represent an important source of neurologic complications in individuals with HIV. The dynamic, often subclinical, course of HAND has rendered diagnosis, which currently depends on neuropsychometric (NP) evaluation, a challenge for clinicians. Here, we present evidence that functional brain connectivity, derived by large-scale Granger causality (lsGC) analysis of resting-state functional MRI (rs-fMRI) time-series, represents a potential biomarker to address this critical diagnostic need. Brain graph properties were used as features in machine learning tasks to 1) classify individuals as HIV + or HIV − and 2) to predict overall cognitive performance, as assessed by NP scores, in a 22-subject (13 HIV − , 9 HIV +) cohort. Over nearly all seven brain parcellation templates considered, support vector machine (SVM) classifiers based on lsGC-derived brain graph features significantly outperformed those based on conventional Pearson correlation (PC)-derived features (p < 0.05, Bonferroni-corrected). In a second task for which the objective was to predict the overall NP score of each subject, the lsGC-based SVM regressors consistently outperformed the PC-based regressors (p < 0.05, Bonferroni-corrected) on nearly all templates. With the widely used Automated Anatomical Labeling (AAL90) template, it was determined that the brain regions that figured most strongly in *