Adriel Saporta

I’m a PhD candidate in Computer Science at the Courant Institute at NYU working on machine learning for health and advised by Rajesh Ranganath. I’m a DeepMind Scholar and a Shakti VC Fellow. I also co-host The AI Health Podcast with Pranav Rajpurkar.

Before joining NYU, I worked on research on Apple’s Health AI team and in Andrew Ng’s Stanford Machine Learning Group. In a previous life, I was Anna Wintour’s executive assistant at Vogue (remember The Devil Wears Prada?), and I’ve held engineering and product roles across both big tech (Apple, Amazon) and start-ups (SeatGeek, Common).

I have an MBA from the Stanford Graduate School of Business, an MS in Computer Science from Stanford University, and a BA in Comparative Literature from Yale University. I was born and raised in Brooklyn, and am half-Cuban and half-Greek.

publications

Don’t be fooled: label leakage in explanation methods and the importance of their quantitative evaluation

Neil Jethani*, Adriel Saporta*, and Rajesh Ranganath

AISTATS 2023 (notable paper, oral presentation)

Abs HTML PDF

Feature attribution methods identify which features of an input most influence a model’s output. Most widely-used feature attribution methods (such as SHAP, LIME, and Grad-CAM) are "class-dependent" methods in that they generate a feature attribution vector as a function of class. In this work, we demonstrate that class-dependent methods can "leak" information about the selected class, making that class appear more likely than it is. Thus, an end user runs the risk of drawing false conclusions when interpreting an explanation generated by a class-dependent method. In contrast, we introduce "distribution-aware" methods, which favor explanations that keep the label’s distribution close to its distribution given all features of the input. We introduce SHAP-KL and FastSHAP-KL, two baseline distribution-aware methods that compute Shapley values. Finally, we perform a comprehensive evaluation of seven class-dependent and three distribution-aware methods on three clinical datasets of different high-dimensional data types: images, biosignals, and text.
Benchmarking saliency methods for chest X-ray interpretation

Adriel Saporta*, Xiaotong Gui*, Ashwin Agrawal*, Anuj Pareek, Steven QH Truong, Chanh DT Nguyen, Van-Doan Ngo, Jayne Seekins, Francis G. Blankenberg, Andrew Y. Ng, Matthew P. Lungren, and Pranav Rajpurkar

Nature Machine Intelligence 2022

Abs HTML PDF Code

Saliency methods, which “explain” deep neural networks by producing heat maps that highlight the areas of the medical image that influence model prediction, are often presented to clinicians as an aid in diagnostic decision-making. Although many saliency methods have been proposed for medical imaging interpretation, rigorous investigation of the accuracy and reliability of these strategies is necessary before they are integrated into the clinical setting. In this work, we quantitatively evaluate seven saliency methods—including Grad-CAM, Grad-CAM++, and Integrated Gradients—across multiple neural network architectures using two evaluation metrics. We establish the first human benchmark for chest X-ray segmentation in a multilabel classification set up, and examine under what clinical conditions saliency maps might be more prone to failure in localizing important pathologies compared to a human expert benchmark. We find that (i) while Grad-CAM generally localized pathologies better than the other evaluated saliency methods, all seven performed significantly worse compared with the human benchmark; (ii) the gap in localization performance between Grad-CAM and the human benchmark was largest for pathologies that were smaller in size and had shapes that were more complex; (iii) model confidence was positively correlated with Grad-CAM localization performance. While it is difficult to know whether poor localization performance is attributable to the model or to the saliency method, our work demonstrates that several important limitations of saliency methods must be addressed before we can rely on them for deep learning explainability in medical imaging.
Learning Invariant Representations with Missing Data

Mark Goldstein, Jörn-Henrik Jacobsen, Olina Chau, Adriel Saporta, Aahlad Manas Puli, Rajesh Ranganath, and Andrew C. Miller

CLeaR (Causal Learning and Reasoning) 2022

Abs PDF

Spurious correlations allow flexible models to predict well during training but poorly on related test distributions. Recent work has shown that models that satisfy particular independencies involving correlation-inducing nuisance variables have guarantees on their test performance. Enforcing such independencies requires nuisances to be observed during training. However, nuisances, such as demographics or image background labels, are often missing. Enforcing independence on just the observed data does not imply independence on the entire population. Here we derive MMD estimators used for invariance objectives under missing nuisances. On simulations and clinical data, optimizing through these estimates achieves test performance similar to using estimators that make use of the full data.
RadGraph: Extracting Clinical Entities and Relations from Radiology Reports

Saahil Jain*, Ashwin Agrawal*, Adriel Saporta*, Steven Truong, Du Nguyen Duong, Tan Bui, Pierre Chambon, Yuhao Zhang, Matthew Lungren, Andrew Ng, Curtis Langlotz, Pranav Rajpurkar, and Pranav Rajpurkar

NeurIPS Datasets and Benchmarks 2021 (oral presentation)

Abs PDF

Extracting structured clinical information from free-text radiology reports can enable the use of radiology report information for a variety of critical healthcare applications. In our work, we present RadGraph, a dataset of entities and relations in full-text chest X-ray radiology reports based on a novel information extraction schema we designed to structure radiology reports. We release a development dataset, which contains board-certified radiologist annotations for 500 radiology reports from the MIMIC-CXR dataset (14,579 entities and 10,889 relations), and a test dataset, which contains two independent sets of board-certified radiologist annotations for 100 radiology reports split equally across the MIMIC-CXR and CheXpert datasets. Using these datasets, we train and test a deep learning model, RadGraph Benchmark, that achieves a micro F1 of 0.82 and 0.73 on relation extraction on the MIMIC-CXR and CheXpert test sets respectively. Additionally, we release an inference dataset, which contains annotations automatically generated by RadGraph Benchmark across 220,763 MIMIC-CXR reports (around 6 million entities and 4 million relations) and 500 CheXpert reports (13,783 entities and 9,908 relations) with mappings to associated chest radiographs. Our freely available dataset can facilitate a wide range of research in medical natural language processing, as well as computer vision and multi-modal learning when linked to chest radiographs.
Q-Pain: A Question Answering Dataset to Measure Social Bias in Pain Management

Cécile Logé*, Emily Ross*, David Dadey, Saahil Jain, Adriel Saporta, Andrew Ng, Pranav Rajpurkar, and Pranav Rajpurkar

NeurIPS Datasets and Benchmarks 2021

Abs PDF

Recent advances in Natural Language Processing (NLP), and specifically automated Question Answering (QA) systems, have demonstrated both impressive linguistic fluency and a pernicious tendency to reflect social biases. In this study, we introduce Q-Pain, a dataset for assessing bias in medical QA in the context of pain management, one of the most challenging forms of clinical decision-making. Along with the dataset, we propose a new, rigorous framework, including a sample experimental design, to measure the potential biases present when making treatment decisions. We demonstrate its use by assessing two reference Question-Answering systems, GPT-2 and GPT-3, and find statistically significant differences in treatment between intersectional race-gender subgroups, thus reaffirming the risks posed by AI in medical settings, and the need for datasets like ours to ensure safety before medical AI applications are deployed.