Minimax Pareto Fairness: A Multi Objective Perspective.

In this work we formulate and formally characterize group fairness as a multi-objective optimization problem, where each sensitive group risk is a separate objective. We propose a fairness criterion where a classifier achieves minimax risk and is Pareto-efficient w.r.t. all groups, avoiding unnecessary harm, and can lead to the best zero-gap model if policy dictates so. We provide a simple optimization algorithm compatible with deep neural networks to satisfy these constraints. Since our method does not require test-time access to sensitive attributes, it can be applied to reduce worst-case classification errors between outcomes in unbalanced classification problems. We test the proposed methodology on real case-studies of predicting income, ICU patient mortality, skin lesions classification, and assessing credit risk, demonstrating how our framework compares favorably to other approaches.

Attention-Based Network for Weak Labels in Neonatal Seizure Detection.

Seizures are a common emergency in the neonatal intesive care unit (NICU) among newborns receiving therapeutic hypothermia for hypoxic ischemic encephalopathy. The high incidence of seizures in this patient population necessitates continuous electroencephalographic (EEG) monitoring to detect and treat them. Due to EEG recordings being reviewed intermittently throughout the day, inevitable delays to seizure identification and treatment arise. In recent years, work on neonatal seizure detection using deep learning algorithms has started gaining momentum. These algorithms face numerous challenges: first, the training data for such algorithms comes from individual patients, each with varying levels of label imbalance since the seizure burden in NICU patients differs by several orders of magnitude. Second, seizures in neonates are usually localized in a subset of EEG channels, and performing annotations per channel is very time-consuming. Hence models which make use of labels only per time periods, and not per channels, are preferable. In this work we assess how different deep learning models and data balancing methods influence learning in neonatal seizure detection in EEGs. We propose a model which provides a level of importance to each of the EEG channels - a proxy to whether a channel exhibits seizure activity or not, and we provide a quantitative assessment of how well this mechanism works. The model is portable to EEG devices with differing layouts without retraining, facilitating its potential deployment across different medical centers. We also provide a first assessment of how a deep learning model for neonatal seizure detection agrees with human rater decisions - an important milestone for deployment to clinical practice. We show that high AUC values in a deep learning model do not necessarily correspond to agreement with a human expert, and there is still a need to further refine such algorithms for optimal seizure discrimination.

Active learning of cortical connectivity from two-photon imaging data.

Understanding how groups of neurons interact within a network is a fundamental question in system neuroscience. Instead of passively observing the ongoing activity of a network, we can typically perturb its activity, either by external sensory stimulation or directly via techniques such as two-photon optogenetics. A natural question is how to use such perturbations to identify the connectivity of the network efficiently. Here we introduce a method to infer sparse connectivity graphs from in-vivo, two-photon imaging of population activity in response to external stimuli. A novel aspect of the work is the introduction of a recommended distribution, incrementally learned from the data, to optimally refine the inferred network. Unlike existing system identification techniques, this "active learning" method automatically focuses its attention on key undiscovered areas of the network, instead of targeting global uncertainty indicators like parameter variance. We show how active learning leads to faster inference while, at the same time, provides confidence intervals for the network parameters. We present simulations on artificial small-world networks to validate the methods and apply the method to real data. Analysis of frequency of motifs recovered show that cortical networks are consistent with a small-world topology model.

An open tool for input function estimation and quantification of dynamic PET FDG brain scans.

PURPOSE: Positron emission tomography (PET) analysis of clinical studies is mostly restricted to qualitative evaluation. Quantitative analysis of PET studies is highly desirable to be able to compute an objective measurement of the process of interest in order to evaluate treatment response and/or compare patient data. But implementation of quantitative analysis generally requires the determination of the input function: the arterial blood or plasma activity which indicates how much tracer is available for uptake in the brain. The purpose of our work was to share with the community an open software tool that can assist in the estimation of this input function, and the derivation of a quantitative map from the dynamic PET study. METHODS: Arterial blood sampling during the PET study is the gold standard method to get the input function, but is uncomfortable and risky for the patient so it is rarely used in routine studies. To overcome the lack of a direct input function, different alternatives have been devised and are available in the literature. These alternatives derive the input function from the PET image itself (image-derived input function) or from data gathered from previous similar studies (population-based input function). In this article, we present ongoing work that includes the development of a software tool that integrates several methods with novel strategies for the segmentation of blood pools and parameter estimation. RESULTS: The tool is available as an extension to the 3D Slicer software. Tests on phantoms were conducted in order to validate the implemented methods. We evaluated the segmentation algorithms over a range of acquisition conditions and vasculature size. Input function estimation algorithms were evaluated against ground truth of the phantoms, as well as on their impact over the final quantification map. End-to-end use of the tool yields quantification maps with [Formula: see text] relative error in the estimated influx versus ground truth on phantoms. CONCLUSIONS: The main contribution of this article is the development of an open-source, free to use tool that encapsulates several well-known methods for the estimation of the input function and the quantification of dynamic PET FDG studies. Some alternative strategies are also proposed and implemented in the tool for the segmentation of blood pools and parameter estimation. The tool was tested on phantoms with encouraging results that suggest that even bloodless estimators could provide a viable alternative to blood sampling for quantification using graphical analysis. The open tool is a promising opportunity for collaboration among investigators and further validation on real studies.