Search for: All records

Automatic assessment of impairment and disease severity is a key challenge in data-driven medicine. We propose a framework to address this challenge, which leverages AI models trained exclusively on healthy individuals. The COnfidence-Based chaRacterization of Anomalies (COBRA) score exploits the decrease in confidence of these models when presented with impaired or diseased patients to quantify their deviation from the healthy population. We applied the COBRA score to address a key limitation of current clinical evaluation of upper-body impairment in stroke patients. The gold-standard Fugl-Meyer Assessment (FMA) requires in-person administration by a trained assessor for 30-45 minutes, which restricts monitoring frequency and precludes physicians from adapting rehabilitation protocols to the progress of each patient. The COBRA score, computed automatically in under one minute, is shown to be strongly correlated with the FMA on an independent test cohort for two different data modalities: wearable sensors (ρ = 0.814, 95% CI [0.700,0.888]) and video (ρ = 0.736, 95% C.I [0.584, 0.838]). To demonstrate the generalizability of the approach to other conditions, the COBRA score was also applied to quantify severity of knee osteoarthritis from magnetic-resonance imaging scans, again achieving significant correlation with an independent clinical assessment (ρ = 0.644, 95% C.I [0.585,0.696]). 

Observational constraints on the element abundances in metal-poor stars in the Galactic halo have grown dramatically over the past decade. Such data is now available for a few up to dozens of elements in the Milky Way as well as extragalactic halos. Despite this greatly increased observational database and important strides in the theoretical understanding of the first stars, interpreting the signature of primordial supernovae remains challenging. This signature may manifest through element abundance patterns in Galactic hyper-iron-poor stars or ultra-metal-poor stars, or in local dSph galaxies. These patterns can be tied through modeling (with some caveats) to the nucleosynthetic output of individual supernovae from first-generation stars. We share ongoing calculations of the current constraints on the first supernovae and some fascinating puzzles that remain unresolved to date. A first-stars supernova model needs to explain the widely-noted enhancement of carbon and many alpha-elements at the lowest [Fe/H], but also must consistently account for the trends in N, Li, Ca and Ti in the broader framework of the structure, rotation and binarity of early stellar generations. Last, we note that Ti may not always behave like an alpha-element, especially below [Fe/H] of about -3, providing an additional constraint on the first supernovae. This work was supported by the University of San Francisco (USF) Faculty Development Fund, and by the Undergraduate ALFALFA Team through NSF grant AST-1637339. 

We present our results on calculations of the escape of Ly-alpha and Ly-continuum radiation from low- and intermediate-mass galaxies. Such systems may have played a crucial role in reionization at early times. We use simple analytic models for the underlying galaxy profiles and compare them with semi-analytic and numerical computations of escaping radiation from such systems. We comment on the possible range of values for the critical spectral index of the source radiation at which H and He ionization start to compete, under a variety of physical conditions. Last, we examine data of low- and intermediate-mass galaxy populations in the local volume, including strong-emission line systems like green pea galaxies and Ly-alpha emitting systems, that closely resemble the earliest halos that hosted the first stars. We share a set of observable galaxy properties that could characterize the "leakers", whose high-redshift counterparts would have had significant escape of Ly-alpha and Ly-continuum radiation. This work was supported by the University of San Francisco (USF) Faculty Development Fund, the USF Student Travel Fund, and by the Undergraduate ALFALFA Team through NSF grant AST-1637339. 

The Arecibo Pisces-Perseus Supercluster Survey is a targeted HI survey of galaxies that began its second observing season in October 2016. The survey is conducted by members of the Undergraduate ALFALFA Team (UAT) and extensively involves undergraduates in observations, data reduction, and analysis. It aims to complement the HI sources identified by the ALFALFA extragalactic HI line survey by probing deeper in HI mass (to lower masses) than the legacy survey itself. Measurements of the HI line velocity widths will be combined with uniform processing of images obtained in the SDSS and GALEX public databases to localize the sample within the baryonic Tully Fisher relation, allowing estimates of their redshift-independent distances and thus their peculiar velocities. The survey is designed to constrain Pisces-Perseus Supercluster infall models by producing 5-σ detections of infall velocities to a precision of about 500 km/s. By targeting galaxies based on SDSS and GALEX photometry, we have achieved detection rates of 68% of the galaxies in our sample. We will discuss the target selection process, HI velocities and mass estimates from the 2015 fall observing season, preliminary results from 2016 observations, and preliminary comparisons with inflow models predicted by numerical simulations. This work has been supported by NSF grants AST-1211005, AST-1637339, AST-1637262.