Erratum: Measuring disparate outcomes of content recommendation algorithms with distributional inequality metrics.

[This corrects the article DOI: 10.1016/j.patter.2022.100568.].

Measuring disparate outcomes of content recommendation algorithms with distributional inequality metrics.

The harmful impacts of algorithmic decision systems have recently come into focus, with many examples of machine learning (ML) models amplifying societal biases. In this paper, we propose adapting income inequality metrics from economics to complement existing model-level fairness metrics, which focus on intergroup differences of model performance. In particular, we evaluate their ability to measure disparities between exposures that individuals receive in a production recommendation system, the Twitter algorithmic timeline. We define desirable criteria for metrics to be used in an operational setting by ML practitioners. We characterize engagements with content on Twitter using these metrics and use the results to evaluate the metrics with respect to our criteria. We also show that we can use these metrics to identify content suggestion algorithms that contribute more strongly to skewed outcomes between users. Overall, we conclude that these metrics can be a useful tool for auditing algorithms in production settings.

Alterations of functional circuitry in aging brain and the impact of mutated APP expression.

Alzheimer's disease (AD) is a disease of aging that results in cognitive impairment, dementia, and death. Pathognomonic features of AD are amyloid plaques composed of proteolytic fragments of the amyloid precursor protein (APP) and neurofibrillary tangles composed of hyperphosphorylated tau protein. One type of familial AD occurs when mutant forms of APP are inherited. Both APP and tau are components of the microtubule-based axonal transport system, which prompts the hypothesis that axonal transport is disrupted in AD, and that such disruption impacts cognitive function. Transgenic mice expressing mutated forms of APP provide preclinical experimental systems to study AD. Here, we perform manganese-enhanced magnetic resonance imaging to study transport from hippocampus to forebrain in four cohorts of living mice: young and old wild-type and transgenic mice expressing a mutant APP with both Swedish and Indiana mutations (APPSwInd). We find that transport is decreased in normal aging and further altered in aged APPSwInd plaque-bearing mice. These findings support the hypothesis that transport deficits are a component of AD pathology and thus may contribute to cognitive deficits.

Automated Computational Processing of 3-D MR Images of Mouse Brain for Phenotyping of Living Animals.

Magnetic resonance (MR) imaging provides a method to obtain anatomical information from the brain in vivo that is not typically available by optical imaging because of this organ's opacity. MR is nondestructive and obtains deep tissue contrast with 100-µm3 voxel resolution or better. Manganese-enhanced MRI (MEMRI) may be used to observe axonal transport and localized neural activity in the living rodent and avian brain. Such enhancement enables researchers to investigate differences in functional circuitry or neuronal activity in images of brains of different animals. Moreover, once MR images of a number of animals are aligned into a single matrix, statistical analysis can be done comparing MR intensities between different multi-animal cohorts comprising individuals from different mouse strains or different transgenic animals, or at different time points after an experimental manipulation. Although preprocessing steps for such comparisons (including skull stripping and alignment) are automated for human imaging, no such automated processing has previously been readily available for mouse or other widely used experimental animals, and most investigators use in-house custom processing. This protocol describes a stepwise method to perform such preprocessing for mouse. © 2017 by John Wiley & Sons, Inc.

Analysis of endovascular therapy for femoropopliteal disease with the Supera stent.

BACKGROUND: The structural limitations of currently available laser-cut nitinol stents in the highly diseased arterial system of the lower extremity are well described. These devices are frequently used in long-segment occlusions, at the adductor hiatus, and across the knee into the popliteal artery and tibial-peroneal trunk despite concerns related to the high mobility of the knee joint and the potential for external stent compression, fractures, and occlusion. The Supera stent (Abbott Vascular, Santa Clara, Calif) was developed to provide superior radial strength, fracture resistance, and flexibility compared with laser-cut nitinol stents. This study represents the initial U.S. experience in the management of lower extremity artery atherosclerotic disease with the Supera interwoven wire stent. METHODS: Patients undergoing stent implantation in the 20-month period after the 2008 Food and Drug Administration clearance were included. Medical records, radiographic imaging, and procedural data were examined. Procedural angiograms were classified according to TransAtlantic Inter-Society Consensus criteria. Patency and limb loss rates were calculated using Kaplan-Meier analysis. RESULTS: There were 305 stents implanted in 147 patients. Clinical follow-up was a mean of 12.7 months (range, 0.2-33.7 months), and radiologic follow-up was a mean of 8.4 months (range, 0.2-26.8 months). Most patients had critical limb ischemia, with tissue loss (38.1%) or rest pain (28.6%) as the indication for intervention. Of the 147 patients, 75 (51.02%) underwent a concomitant adjunctive procedure in the treated extremity. Primary, assisted primary, and secondary patency rates at 12 months by duplex ultrasound imaging were 89.8%, 91.2%, and 93.2%, respectively, by Kaplan-Meier estimates, with a mean lesion length of 184.5 ± 131.80 mm and mean stented length of 197.5 ± 113.65 mm. Seventeen patients experienced an event requiring successful reintervention in the stented segment (13 for type I or II restenosis; four for type III). There were eight major amputations, with five of those eight having a patent stent at the time of limb sacrifice. The overall mortality was 11.6% during the study period. Knee radiography was performed in 47 patients with devices extending into the popliteal and tibial-peroneal segment, and no stent fractures were identified. CONCLUSIONS: Stenting of the superficial femoral and popliteal arteries using the Supera stent system appears to be safe and effective. The interwoven stent design may better serve areas under extreme mechanical stress. Our results with this population of highly diseased patients including very long lesion lengths are consistent with outcomes of other publications reporting the use of this device.

A simple rapid process for semi-automated brain extraction from magnetic resonance images of the whole mouse head.

BACKGROUND: Magnetic resonance imaging (MRI) is a well-developed technique in neuroscience. Limitations in applying MRI to rodent models of neuropsychiatric disorders include the large number of animals required to achieve statistical significance, and the paucity of automation tools for the critical early step in processing, brain extraction, which prepares brain images for alignment and voxel-wise statistics. NEW METHOD: This novel timesaving automation of template-based brain extraction ("skull-stripping") is capable of quickly and reliably extracting the brain from large numbers of whole head images in a single step. The method is simple to install and requires minimal user interaction. RESULTS: This method is equally applicable to different types of MR images. Results were evaluated with Dice and Jacquard similarity indices and compared in 3D surface projections with other stripping approaches. Statistical comparisons demonstrate that individual variation of brain volumes are preserved. COMPARISON WITH EXISTING METHODS: A downloadable software package not otherwise available for extraction of brains from whole head images is included here. This software tool increases speed, can be used with an atlas or a template from within the dataset, and produces masks that need little further refinement. CONCLUSIONS: Our new automation can be applied to any MR dataset, since the starting point is a template mask generated specifically for that dataset. The method reliably and rapidly extracts brain images from whole head images, rendering them useable for subsequent analytical processing. This software tool will accelerate the exploitation of mouse models for the investigation of human brain disorders by MRI.

Quantitative measurements and modeling of cargo-motor interactions during fast transport in the living axon.

The kinesins have long been known to drive microtubule-based transport of sub-cellular components, yet the mechanisms of their attachment to cargo remain a mystery. Several different cargo-receptors have been proposed based on their in vitro binding affinities to kinesin-1. Only two of these-phosphatidyl inositol, a negatively charged lipid, and the carboxyl terminus of the amyloid precursor protein (APP-C), a trans-membrane protein-have been reported to mediate motility in living systems. A major question is how these many different cargo, receptors and motors interact to produce the complex choreography of vesicular transport within living cells. Here we describe an experimental assay that identifies cargo-motor receptors by their ability to recruit active motors and drive transport of exogenous cargo towards the synapse in living axons. Cargo is engineered by derivatizing the surface of polystyrene fluorescent nanospheres (100 nm diameter) with charged residues or with synthetic peptides derived from candidate motor receptor proteins, all designed to display a terminal COOH group. After injection into the squid giant axon, particle movements are imaged by laser-scanning confocal time-lapse microscopy. In this report we compare the motility of negatively charged beads with APP-C beads in the presence of glycine-conjugated non-motile beads using new strategies to measure bead movements. The ensuing quantitative analysis of time-lapse digital sequences reveals detailed information about bead movements: instantaneous and maximum velocities, run lengths, pause frequencies and pause durations. These measurements provide parameters for a mathematical model that predicts the spatiotemporal evolution of distribution of the two different types of bead cargo in the axon. The results reveal that negatively charged beads differ from APP-C beads in velocity and dispersion, and predict that at long time points APP-C will achieve greater progress towards the presynaptic terminal. The significance of this data and accompanying model pertains to the role transport plays in neuronal function, connectivity, and survival, and has implications in the pathogenesis of neurological disorders, such as Alzheimer's, Huntington and Parkinson's diseases.