Receive a weekly summary and discussion of the top papers of the week by leading researchers in the field.

General General

The Rise of AI-Driven Simulators: Building a New Crystal Ball

ArXiv Preprint

The use of computational simulation is by now so pervasive in society that it is no exaggeration to say that continued U.S. and international prosperity, security, and health depend in part on continued improvements in simulation capabilities. What if we could predict weather two weeks out, guide the design of new drugs for new viral diseases, or manage new manufacturing processes that cut production costs and times by an order of magnitude? What if we could predict collective human behavior, for example, response to an evacuation request during a natural disaster, or labor response to fiscal stimulus? (See also the companion CCC Quad Paper on Pandemic Informatics, which discusses features that would be essential to solving large-scale problems like preparation for, and response to, the inevitable next pandemic.) The past decade has brought remarkable advances in complementary areas: in sensors, which can now capture enormous amounts of data about the world, and in AI methods capable of learning to extract predictive patterns from those data. These advances may lead to a new era in computational simulation, in which sensors of many kinds are used to produce vast quantities of data, AI methods identify patterns in those data, and new AI-driven simulators combine machine-learned and mathematical rules to make accurate and actionable predictions. At the same time, there are new challenges -- computers in some important regards are no longer getting faster, and in some areas we are reaching the limits of mathematical understanding, or at least of our ability to translate mathematical understanding into efficient simulation. In this paper, we lay out some themes that we envision forming part of a cohesive, multi-disciplinary, and application-inspired research agenda on AI-driven simulators.

Ian Foster, David Parkes, Stephan Zheng


General General

Deep learning reveals 3D atherosclerotic plaque distribution and composition.

In Scientific reports ; h5-index 158.0

Complications of atherosclerosis are the leading cause of morbidity and mortality worldwide. Various genetically modified mouse models are used to investigate disease trajectory with classical histology, currently the preferred methodology to elucidate plaque composition. Here, we show the strength of light-sheet fluorescence microscopy combined with deep learning image analysis for characterising and quantifying plaque burden and composition in whole aorta specimens. 3D imaging is a non-destructive method that requires minimal ex vivo handling and can be up-scaled to large sample sizes. Combined with deep learning, atherosclerotic plaque in mice can be identified without any ex vivo staining due to the autofluorescent nature of the tissue. The aorta and its branches can subsequently be segmented to determine how anatomical position affects plaque composition and progression. Here, we find the highest plaque accumulation in the aortic arch and brachiocephalic artery. Simultaneously, aortas can be stained for markers of interest (for example the pan immune cell marker CD45) and quantified. In ApoE-/- mice we observe that levels of CD45 reach a plateau after which increases in plaque volume no longer correlate to immune cell infiltration. All underlying code is made publicly available to ease adaption of the method.

Jurtz Vanessa Isabell, Skovbjerg Grethe, Salinas Casper Gravesen, Roostalu Urmas, Pedersen Louise, Hecksher-Sørensen Jacob, Rolin Bidda, Nyberg Michael, van de Bunt Martijn, Ingvorsen Camilla


Ophthalmology Ophthalmology

A novel deep learning conditional generative adversarial network for producing angiography images from retinal fundus photographs.

In Scientific reports ; h5-index 158.0

Fluorescein angiography (FA) is a procedure used to image the vascular structure of the retina and requires the insertion of an exogenous dye with potential adverse side effects. Currently, there is only one alternative non-invasive system based on Optical coherence tomography (OCT) technology, called OCT angiography (OCTA), capable of visualizing retina vasculature. However, due to its cost and limited view, OCTA technology is not widely used. Retinal fundus photography is a safe imaging technique used for capturing the overall structure of the retina. In order to visualize retinal vasculature without the need for FA and in a cost-effective, non-invasive, and accurate manner, we propose a deep learning conditional generative adversarial network (GAN) capable of producing FA images from fundus photographs. The proposed GAN produces anatomically accurate angiograms, with similar fidelity to FA images, and significantly outperforms two other state-of-the-art generative algorithms ([Formula: see text] and [Formula: see text]). Furthermore, evaluations by experts shows that our proposed model produces such high quality FA images that are indistinguishable from real angiograms. Our model as the first application of artificial intelligence and deep learning to medical image translation, by employing a theoretical framework capable of establishing a shared feature-space between two domains (i.e. funduscopy and fluorescein angiography) provides an unrivaled way for the translation of images from one domain to the other.

Tavakkoli Alireza, Kamran Sharif Amit, Hossain Khondker Fariha, Zuckerbrod Stewart Lee


General General

Multidimensional machine learning algorithms to learn liquid velocity inside a cylindrical bubble column reactor.

In Scientific reports ; h5-index 158.0

For understanding the complex behavior of fluids in a multiphase chemical bubble column reactor, a combination of the computational fluid dynamic (CFD) method and the adaptive network-based fuzzy inference system (ANFIS) method is used to predict bubble flow inside a reactor based on the function of column height. In this study, the Euler-Euler model is employed as a CFD method. In the Eulerian method, continuity and momentum governing equations are mathematically computed for each phase, while the equations are connected together by source terms. After calculating the flow pattern and turbulence flow in the reactor, all data sets are used to prepare a fully artificial method for further prediction. This algorithm contains different learning dimensions such as learning in different directions of reactor or large amount of input parameters and data set representing "big data". The ANFIS method was evaluated in three steps by using one, two, and three inputs in each one to predict the liquid velocity in the x-direction (Ux). The x, y, and z coordinates of the location of the node of the liquid were considered as the inputs. Different percentages of data and various iterations and membership functions were used for training in the ANFIS method. The ANFIS method showed the best prediction using three inputs. This combination also shows the ability of computer science and computational methods in learning physical and chemical phenomena.

Babanezhad Meisam, Marjani Azam, Shirazian Saeed


Public Health Public Health

Vital signs assessed in initial clinical encounters predict COVID-19 mortality in an NYC hospital system.

In Scientific reports ; h5-index 158.0

Timely and effective clinical decision-making for COVID-19 requires rapid identification of risk factors for disease outcomes. Our objective was to identify characteristics available immediately upon first clinical evaluation related COVID-19 mortality. We conducted a retrospective study of 8770 laboratory-confirmed cases of SARS-CoV-2 from a network of 53 facilities in New-York City. We analysed 3 classes of variables; demographic, clinical, and comorbid factors, in a two-tiered analysis that included traditional regression strategies and machine learning. COVID-19 mortality was 12.7%. Logistic regression identified older age (OR, 1.69 [95% CI 1.66-1.92]), male sex (OR, 1.57 [95% CI 1.30-1.90]), higher BMI (OR, 1.03 [95% CI 1.102-1.05]), higher heart rate (OR, 1.01 [95% CI 1.00-1.01]), higher respiratory rate (OR, 1.05 [95% CI 1.03-1.07]), lower oxygen saturation (OR, 0.94 [95% CI 0.93-0.96]), and chronic kidney disease (OR, 1.53 [95% CI 1.20-1.95]) were associated with COVID-19 mortality. Using gradient-boosting machine learning, these factors predicted COVID-19 related mortality (AUC = 0.86) following cross-validation in a training set. Immediate, objective and culturally generalizable measures accessible upon clinical presentation are effective predictors of COVID-19 outcome. These findings may inform rapid response strategies to optimize health care delivery in parts of the world who have not yet confronted this epidemic, as well as in those forecasting a possible second outbreak.

Rechtman Elza, Curtin Paul, Navarro Esmeralda, Nirenberg Sharon, Horton Megan K


Radiology Radiology

Feasibility of Deep Learning Algorithms for Reporting in Routine Spine Magnetic Resonance Imaging.

In International journal of spine surgery

BACKGROUND : Artificial intelligence is gaining traction in automated medical imaging analysis. Development of more accurate magnetic resonance imaging (MRI) predictors of successful clinical outcomes is necessary to better define indications for surgery, improve clinical outcomes with targeted minimally invasive and endoscopic procedures, and realize cost savings by avoiding more invasive spine care.

OBJECTIVE : To demonstrate the ability for deep learning neural network models to identify features in MRI DICOM datasets that represent varying intensities or severities of common spinal pathologies and injuries and to demonstrate the feasibility of generating automated verbal MRI reports comparable to those produced by reading radiologists.

METHODS : A 3-dimensional (3D) anatomical model of the lumbar spine was fitted to each of the patient's MRIs by a team of technicians. MRI T1, T2, sagittal, axial, and transverse reconstruction image series were used to train segmentation models by the intersection of the 3D model through these image sequences. Class definitions were extracted from the radiologist report for the central canal: (0) no disc bulge/protrusion/canal stenosis, (1) disc bulge without canal stenosis, (2) disc bulge resulting in canal stenosis, and (3) disc herniation/protrusion/extrusion resulting in canal stenosis. Both the left and right neural foramina were assessed with either (0) neural foraminal stenosis absent, or (1) neural foramina stenosis present. Reporting criteria for the pathologies at each disc level and, when available, the grading of severity were extracted, and a natural language processing model was used to generate a verbal and written report. These data were then used to train a set of very deep convolutional neural network models, optimizing for minimal binary cross-entropy for each classification.

RESULTS : The initial prediction validation of the implemented deep learning algorithm was done on 20% of the dataset, which was not used for artificial intelligence training. Of the 17,800 total disc locations for which MRI images and radiology reports were available, 14,720 were used to train the model, and 3560 were used to validate against. The convergence of validation accuracy achieved with the deep learning algorithm for the foraminal stenosis detector was 81% (sensitivity = 72.4.4%, specificity = 83.1%) after 25 complete iterations through the entire training dataset (epoch). The accuracy was 86.2% (sensitivity = 91.1%, specificity = 82.5%) for the central stenosis detector and 85.2% (sensitivity = 81.8%, specificity = 87.4%) for the disc herniation detector.

CONCLUSIONS : Deep learning algorithms may be used for routine reporting in spine MRI. There was a minimal disparity among accuracy, sensitivity, and specificity, indicating that the data were not overfitted to the training set. We concluded that variability in the training data tends to reduce overfitting and overtraining as the deep neural network models learn to focus on the common pathologies. Future studies should demonstrate the accuracy of deep neural network models and the predictive value of favorable clinical outcomes with intervention and surgery.


CLINICAL RELEVANCE : Feasibility, clinical teaching, and evaluation study.

LewandrowskI Kai-Uwe, Muraleedharan Narendran, Eddy Steven Allen, Sobti Vikram, Reece Brian D, Ramírez León Jorge Felipe, Shah Sandeep


artificial intelligence, deep neural network learning, feasibility analysis, magnetic resonance imaging, spinal pathologies