Skip to main content

Edmon Begoli

Oak Ridge National Laboratory, National Security Sciences Directorate, Director, Center for AI Security Research (CAISER)

UC Berkeley, EECS, Research Scientist (Visiting)

University of Tennessee Knoxville, EECS, Adjunct

Followers

41

Following

7

Co-authors

7

Public Views

Director, Center for AI Security Research (CAISER)
Oak Ridge National Laboratory (ORNL)
Phone: 865.576.0599
Address: 1 Bethel Valley Rd.
Oak Ridge, TN 37831

less

Julita Vassileva

University of Saskatchewan

Knox College

Jayjit Majumdar

University of Kalyani

Roshan Chitrakar

Nepal College of Information Technology

Graduate Center of the City University of New York

Université de Tunis

Viacheslav Kuleshov

Stockholm University

José Francisco Pessanha

UERJ - Universidade do Estado do Rio de Janeiro / Rio de Janeiro State University

PALIMOTE JUSTICE

RIVERS STATE POLYTECHNIC

Nikos Mantzakouras

National & Kapodistrian University of Athens

Interests

Uploads

Papers by Edmon Begoli

What Clinical Trials Can Teach Us about the Development of More Resilient AI for Cybersecurity

arXiv (Cornell University), May 13, 2021

This manuscript has been authored by UT-Battelle LLC under contract DE-AC05-00OR22725 with the US... more This manuscript has been authored by UT-Battelle LLC under contract DE-AC05-00OR22725 with the US Department of Energy (DOE). The US government retains and the publisher, by accepting the article for publication, acknowledges that the US government retains a nonexclusive, paid-up, irrevocable, worldwide license to publish or reproduce the published form of this manuscript, or allow others to do so, for US government purposes. DOE will provide public access to these results of federally sponsored research in accordance with the DOE Public Access Plan (http://energy.gov/downloads/doe-public-access-plan).

Diversity and Scale: Genetic Architecture of 2,068 Traits in the VA Million Veteran Program

medRxiv (Cold Spring Harbor Laboratory), Jun 29, 2023

Genome-wide association studies (GWAS) have underrepresented individuals from non-European popula... more Genome-wide association studies (GWAS) have underrepresented individuals from non-European populations, impeding progress in characterizing the genetic architecture and consequences of health and disease traits. To address this, we present a population-stratified phenome-wide GWAS followed by a multi-population meta-analysis for 2,068 traits derived from electronic health records of 635,969 participants in the Million Veteran Program (MVP), a longitudinal cohort study of diverse U.S. Veterans genetically similar to the respective African (121,177), Admixed American (59,048), East Asian (6,702), and European (449,042) superpopulations defined by the 1000 Genomes Project. We identified 38,270 independent variants associating with one or more traits at experiment-wide (P < 4.6x10-11) significance; finemapping 6,318 signals identified from 613 traits to single-variant resolution. Among these, a third (2,069) of the associations were found only among participants genetically similar to non-European reference populations, demonstrating the importance of expanding diversity in genetic studies. Our work provides a comprehensive atlas of phenome-wide genetic associations for future studies dissecting the architecture of complex traits in diverse populations.

Visualizing novel connections and genetic similarities across diseases using a network-medicine based approach

Scientific Reports

Understanding the genetic relationships between human disorders could lead to better treatment an... more Understanding the genetic relationships between human disorders could lead to better treatment and prevention strategies, especially for individuals with multiple comorbidities. A common resource for studying genetic-disease relationships is the GWAS Catalog, a large and well curated repository of SNP-trait associations from various studies and populations. Some of these populations are contained within mega-biobanks such as the Million Veteran Program (MVP), which has enabled the genetic classification of several diseases in a large well-characterized and heterogeneous population. Here we aim to provide a network of the genetic relationships among diseases and to demonstrate the utility of quantifying the extent to which a given resource such as MVP has contributed to the discovery of such relations. We use a network-based approach to evaluate shared variants among thousands of traits in the GWAS Catalog repository. Our results indicate many more novel disease relationships that di...

1346: Markers in Unstructured Progress Notes Predict Imminent Icu Admission Using Machine Learning

Critical Care Medicine, 2020

Standardized Architecture for a Mega-Biobank Phenomic Library: The Million Veteran Program (MVP)

PubMed, 2020

Electronic health records (EHRs) provide a wealth of data for phenotype development in population... more Electronic health records (EHRs) provide a wealth of data for phenotype development in population health studies, and researchers invest considerable time to curate data elements and validate disease definitions. The ability to reproduce well-defined phenotypes increases data quality, comparability of results and expedites research. In this paper, we present a standardized approach to organize and capture phenotype definitions, resulting in the creation of an open, online repository of phenotypes. This resource captures phenotype development, provenance and process from the Million Veteran Program, a national mega-biobank embedded in the Veterans Health Administration (VHA). To ensure that the repository is searchable, extendable, and sustainable, it is necessary to develop both a proper digital catalog architecture and underlying metadata infrastructure to enable effective management of the data fields required to define each phenotype. Our methods provide a resource for VHA investigators and a roadmap for researchers interested in standardizing their phenotype definitions to increase portability.

Effects of Real-Life Traffic Sign Alteration on YOLOv7- an Object Recognition Model

arXiv (Cornell University), May 9, 2023

The advancement of Image Processing has led to the widespread use of Object Recognition (OR) mode... more The advancement of Image Processing has led to the widespread use of Object Recognition (OR) models in various applications, such as airport security and mail sorting. These models have become essential in signifying the capabilities of AI and supporting vital services like national postal operations. However, the performance of OR models can be impeded by real-life scenarios, such as traffic sign alteration. Therefore, this research investigates the effects of altered traffic signs on the accuracy and performance of object recognition models. To this end, a publicly available dataset was used to create different types of traffic sign alterations, including changes to size, shape, color, visibility, and angles. The impact of these alterations on the YOLOv7 (You Only Look Once) model's detection and classification abilities were analyzed. It reveals that the accuracy of object detection models decreases significantly when exposed to modified traffic signs under unlikely conditions. This study highlights the significance of enhancing the robustness of object detection models in real-life scenarios and the need for further investigation in this area to improve their accuracy and reliability.

Performance Profile of Transformer Fine-Tuning in Multi-GPU Cloud Environments

2021 IEEE International Conference on Big Data (Big Data), Dec 15, 2021

The study presented here focuses on performance characteristics and trade-offs associated with ru... more The study presented here focuses on performance characteristics and trade-offs associated with running machine-learning tasks in multi-GPU environments on both on-site cloud computing resources and commercial cloud services (Azure). Specifically, this study examines these tradeoffs by examining the performance of training and fine-tuning of transformer-based deep-learning (DL) networks on clinical notes and data, a task of critical importance in the medical domain. To this end, we perform DL-related experiments on the widely deployed NVIDIA V100 GPUs and on the newer A100 GPUs connected via NVLink or PCIe. This study analyzes the execution time of major operations to train DL models and investigate popular options to optimize each of them. We examine and present the findings on the impacts that various operations (e.g. data loading into GPUs, training, fine-tuning), optimizations, and system configurations (single vs. multi-GPU, NVLink vs. PCIe) have on the overall training performance.

Understanding the pros and cons of big data analytics

PubMed, Sep 6, 2014

Data Management and Analytics for Medicine and Healthcare

Lecture Notes in Computer Science, 2017

The use of general descriptive names, registered names, trademarks, service marks, etc. in this p... more The use of general descriptive names, registered names, trademarks, service marks, etc. in this publication does not imply, even in the absence of a specific statement, that such names are exempt from the relevant protective laws and regulations and therefore free for general use. The publisher, the authors and the editors are safe to assume that the advice and information in this book are believed to be true and accurate at the date of publication. Neither the publisher nor the authors or the editors give a warranty, express or implied, with respect to the material contained herein or for any errors or omissions that may have been made. The publisher remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Real-Time Discovery Services over Large, Heterogeneous and Complex Healthcare Datasets Using Schema-Less, Column-Oriented Methods

We present a service platform for schema-leess exploration of data and discovery of patient-relat... more We present a service platform for schema-leess exploration of data and discovery of patient-related statistics from healthcare data sets. The architecture of this platform is motivated by the need for fast, schema-less, and flexible approaches to SQL-based exploration and discovery of information embedded in the common, heterogeneously structured healthcare data sets and supporting components (electronic health records, practice management systems, etc.) The motivating use cases described in the paper are clinical trials candidate discovery, and a treatment effectiveness analysis. Following the use cases, we discuss the key features and software architecture of the platform, the underlying core components (Apache Parquet, Drill, the web services server), and the runtime profiles and performance characteristics of the platform. We conclude by showing dramatic speedup with some approaches, and the performance tradeoffs and limitations of others.

Question-Answering System Extracts Information on Injection Drug Use from Clinical Progress Notes

arXiv (Cornell University), May 15, 2023

model. This paper also demonstrates the QA model's ability to extract IDU-related information on ... more model. This paper also demonstrates the QA model's ability to extract IDU-related information on temporally out-of-distribution data. The results indicate that the majority (51%) of the extracted information by the QA model exactly matches the gold-standard answer and 73% of them contain the gold-standard answer with some additional surrounding words. This manuscript has been authored by UT-Battelle, LLC, under contract DE-AC05-00OR22725 with the US Department of Energy (DOE). The US government retains and the publisher, by accepting the article for publication, acknowledges that the US government retains a nonexclusive, paid-up, irrevocable, worldwide license to publish or reproduce the published form of this manuscript, or allow others to do so, for US government purposes.

A Compound Data Poisoning Technique with Significant Adversarial Effects on Transformer-based Text Classification Tasks

Research Square (Research Square), Jan 10, 2023

Transformer-based models have demonstrated much success in various natural language processing (N... more Transformer-based models have demonstrated much success in various natural language processing (NLP) tasks. However, they are often vulnerable to adversarial attacks, such as data poisoning, that can intentionally fool the model into generating incorrect results. In this paper, we present a novel, compound variant of a data poisoning attack on a transformer-based model that maximizes the poisoning effect while minimizing the scope of poisoning. We do so by combining the established data poisoning technique (label flipping) with a novel adversarial artifact selection and insertion technique aimed at minimizing detectability and the scope of the poisoning footprint. We find that using a combination of these two techniques, we achieve a state-of-the-art attack success rate (ASR) of ∼ 90% while poisoning only 0.5% of the original training set, thus minimizing the scope and detectability of the poisoning action.

cpgQA: A Benchmark Dataset for Machine Reading Comprehension Tasks on Clinical Practice Guidelines and a Case Study Using Transfer Learning

IEEE Access, 2023

Biomedical machine reading comprehension (bio-MRC), a crucial task in natural language processing... more Biomedical machine reading comprehension (bio-MRC), a crucial task in natural language processing, is a vital application of a computer-assisted clinical decision support system. It can help clinicians extract critical information effortlessly for clinical decision-making by comprehending and answering questions from biomedical text data. While recent advances in bio-MRC consider text data from resources such as clinical notes and scholarly articles, the clinical practice guidelines (CPGs) are still unexplored in this regard. CPGs are a pivotal component of clinical decision-making at the point of care as they provide recommendations for patient care based on the most up-to-date information available. Although CPGs are inherently terse compared to a multitude of articles, often, clinicians find them lengthy and complicated to use. In this paper, we define a new problem domain-bio-MRC on CPGs-where the ultimate goal is to assist clinicians in efficiently interpreting the clinical practice guidelines using MRC systems. To that end, we develop a manually annotated and subject-matter expert-validated benchmark dataset for the bio-MRC task on CPGs-cpgQA. This dataset aims to evaluate intelligent systems performing MRC tasks on CPGs. Hence, we employ the state-of-the-art MRC models to present a case study illustrating an extensive evaluation of the proposed dataset. We address the problem of lack of training data in this newly defined domain by applying transfer learning. The results show that while the current state-of-the-art models perform well with 78% exact match scores on the dataset, there is still room for improvement, warranting further research on this problem domain. We release the dataset at https://github.com/mmahbub/cpgQA. INDEX TERMS Biomedical machine reading comprehension, clinical practice guidelines, deep learning, natural language processing, transfer learning, transformers.

Unified Medical Language System resources improve sieve-based generation and Bidirectional Encoder Representations from Transformers (BERT)–based ranking for concept normalization

Journal of the American Medical Informatics Association, Jul 27, 2020

Objective: Concept normalization, the task of linking phrases in text to concepts in an ontology,... more Objective: Concept normalization, the task of linking phrases in text to concepts in an ontology, is useful for many downstream tasks including relation extraction, information retrieval, etc. We present a generate-andrank concept normalization system based on our participation in the 2019 National NLP Clinical Challenges Shared Task Track 3 Concept Normalization. Materials and Methods: The shared task provided 13 609 concept mentions drawn from 100 discharge summaries. We first design a sieve-based system that uses Lucene indices over the training data, Unified Medical Language System (UMLS) preferred terms, and UMLS synonyms to generate a list of possible concepts for each mention. We then design a listwise classifier based on the BERT (Bidirectional Encoder Representations from Transformers) neural network to rank the candidate concepts, integrating UMLS semantic types through a regularizer. Results: Our generate-and-rank system was third of 33 in the competition, outperforming the candidate generator alone (81.66% vs 79.44%) and the previous state of the art (76.35%). During postevaluation, the model's accuracy was increased to 83.56% via improvements to how training data are generated from UMLS and incorporation of our UMLS semantic type regularizer. Discussion: Analysis of the model shows that prioritizing UMLS preferred terms yields better performance, that the UMLS semantic type regularizer results in qualitatively better concept predictions, and that the model performs well even on concepts not seen during training. Conclusions: Our generate-and-rank framework for UMLS concept normalization integrates key UMLS features like preferred terms and semantic types with a neural network-based ranking model to accurately link phrases in text to UMLS concepts.

Poster: StreamToxWatch – Data Poisoning Detector in Distributed, Event-based Environments

Poster: Cognitive Cyber – Dynamic, Adaptive Cyber Defense Systems for Massively Distributed, Autonomous, and Ad-hoc Computing Environments

Evaluating Text Analytic Frameworks for Mental Health Surveillance

Reducing suicide incidence among US veterans is one of the highest priorities for the US Departme... more Reducing suicide incidence among US veterans is one of the highest priorities for the US Department of Veterans Affairs (VA). We are implementing a suicide risk detection system, in collaboration with the VA, that would serve as a surveillance system for risk factors appearing in clinical text data. Primary requirements for this system are fast search capability, feature and information extraction, and delivery of data to up-stream natural language processing models. As such, we are evaluating scalable storage solutions on the basis of performance, fault tolerance, and scalability. In this paper we present our current approach to evaluation, preliminary findings, and the work in progress towards a more robust text analysis pipeline.

Improving Quality of Observational Streaming Medical Data by Using Long Short-Term Memory Networks (LSTMs)

We present an exploration of the encoder-decoder structured Long Short-Term Memory Network (LSTM)... more We present an exploration of the encoder-decoder structured Long Short-Term Memory Network (LSTM) as a detector of the anomalous missing observations in streaming medical data by using the difference between the LSTM-reconstructed and observed values as the anomaly detector. We experiment with time-series data from bedside monitoring devices from the available Medical Information Mart for Intensive Care Database (MIMIC). Our results show that not only encoder-decoder LSTM approach works well for detecting the difference between anomalous and normal missing observations in streaming medical data, but also has an imputation potential for the missing data.

Procedural Reasoning System (PRS) architecture for agent-mediated behavioral interventions

We present an architecture in support of intelligent agent-mediated, behavioral interventions in ... more We present an architecture in support of intelligent agent-mediated, behavioral interventions in special education programs for individuals with Autism Spectrum Disorder (ASD). The proposed architecture is a derivative of the Procedural Reasoning System (PRS) architecture with representative, inter-pretative, reasoning, knowledge-based, and procedural control components abstracted from the physical and locomotoric aspects of the agent's placement in the environment. The architecture is designed to serve as the unifying foundation for virtual, mixed reality and embodied implementations, so its behavior-oriented control, reasoning, knowledge base, and inference mechanisms are designed as abstractive.

sciCAN: Single-cell chromatin accessibility and gene expression data integration via Cycle-consistent Adversarial Network

bioRxiv (Cold Spring Harbor Laboratory), Dec 1, 2021

The boom in single-cell technologies has brought a surge of high dimensional data that come from ... more The boom in single-cell technologies has brought a surge of high dimensional data that come from different sources and represent cellular systems from different views. With advances in these single-cell technologies, integrating single-cell data across modalities arises as a new computational challenge. Here, we present an adversarial approach, sciCAN, to integrate single-cell chromatin accessibility and gene expression data in an unsupervised manner. We benchmarked sciCAN with 5 existing methods in 5 scATACseq/scRNA-seq datasets, and we demonstrated that our method dealt with data integration with consistent performance across datasets and better balance of mutual transferring between modalities than the other 5 existing methods. We further applied sciCAN to 10X Multiome data and confirmed that the integrated representation preserves biological relationships within the hematopoietic hierarchy. Finally, we investigated CRISPR-perturbed single-cell K562 ATAC-seq and RNA-seq data to identify cells with related responses to different perturbations in these different modalities.

What Clinical Trials Can Teach Us about the Development of More Resilient AI for Cybersecurity

arXiv (Cornell University), May 13, 2021

This manuscript has been authored by UT-Battelle LLC under contract DE-AC05-00OR22725 with the US... more This manuscript has been authored by UT-Battelle LLC under contract DE-AC05-00OR22725 with the US Department of Energy (DOE). The US government retains and the publisher, by accepting the article for publication, acknowledges that the US government retains a nonexclusive, paid-up, irrevocable, worldwide license to publish or reproduce the published form of this manuscript, or allow others to do so, for US government purposes. DOE will provide public access to these results of federally sponsored research in accordance with the DOE Public Access Plan (http://energy.gov/downloads/doe-public-access-plan).

Diversity and Scale: Genetic Architecture of 2,068 Traits in the VA Million Veteran Program

medRxiv (Cold Spring Harbor Laboratory), Jun 29, 2023

Genome-wide association studies (GWAS) have underrepresented individuals from non-European popula... more Genome-wide association studies (GWAS) have underrepresented individuals from non-European populations, impeding progress in characterizing the genetic architecture and consequences of health and disease traits. To address this, we present a population-stratified phenome-wide GWAS followed by a multi-population meta-analysis for 2,068 traits derived from electronic health records of 635,969 participants in the Million Veteran Program (MVP), a longitudinal cohort study of diverse U.S. Veterans genetically similar to the respective African (121,177), Admixed American (59,048), East Asian (6,702), and European (449,042) superpopulations defined by the 1000 Genomes Project. We identified 38,270 independent variants associating with one or more traits at experiment-wide (P < 4.6x10-11) significance; finemapping 6,318 signals identified from 613 traits to single-variant resolution. Among these, a third (2,069) of the associations were found only among participants genetically similar to non-European reference populations, demonstrating the importance of expanding diversity in genetic studies. Our work provides a comprehensive atlas of phenome-wide genetic associations for future studies dissecting the architecture of complex traits in diverse populations.

Visualizing novel connections and genetic similarities across diseases using a network-medicine based approach

Scientific Reports

Understanding the genetic relationships between human disorders could lead to better treatment an... more Understanding the genetic relationships between human disorders could lead to better treatment and prevention strategies, especially for individuals with multiple comorbidities. A common resource for studying genetic-disease relationships is the GWAS Catalog, a large and well curated repository of SNP-trait associations from various studies and populations. Some of these populations are contained within mega-biobanks such as the Million Veteran Program (MVP), which has enabled the genetic classification of several diseases in a large well-characterized and heterogeneous population. Here we aim to provide a network of the genetic relationships among diseases and to demonstrate the utility of quantifying the extent to which a given resource such as MVP has contributed to the discovery of such relations. We use a network-based approach to evaluate shared variants among thousands of traits in the GWAS Catalog repository. Our results indicate many more novel disease relationships that di...

1346: Markers in Unstructured Progress Notes Predict Imminent Icu Admission Using Machine Learning

Critical Care Medicine, 2020

Standardized Architecture for a Mega-Biobank Phenomic Library: The Million Veteran Program (MVP)

PubMed, 2020

Electronic health records (EHRs) provide a wealth of data for phenotype development in population... more Electronic health records (EHRs) provide a wealth of data for phenotype development in population health studies, and researchers invest considerable time to curate data elements and validate disease definitions. The ability to reproduce well-defined phenotypes increases data quality, comparability of results and expedites research. In this paper, we present a standardized approach to organize and capture phenotype definitions, resulting in the creation of an open, online repository of phenotypes. This resource captures phenotype development, provenance and process from the Million Veteran Program, a national mega-biobank embedded in the Veterans Health Administration (VHA). To ensure that the repository is searchable, extendable, and sustainable, it is necessary to develop both a proper digital catalog architecture and underlying metadata infrastructure to enable effective management of the data fields required to define each phenotype. Our methods provide a resource for VHA investigators and a roadmap for researchers interested in standardizing their phenotype definitions to increase portability.

Effects of Real-Life Traffic Sign Alteration on YOLOv7- an Object Recognition Model

arXiv (Cornell University), May 9, 2023

The advancement of Image Processing has led to the widespread use of Object Recognition (OR) mode... more The advancement of Image Processing has led to the widespread use of Object Recognition (OR) models in various applications, such as airport security and mail sorting. These models have become essential in signifying the capabilities of AI and supporting vital services like national postal operations. However, the performance of OR models can be impeded by real-life scenarios, such as traffic sign alteration. Therefore, this research investigates the effects of altered traffic signs on the accuracy and performance of object recognition models. To this end, a publicly available dataset was used to create different types of traffic sign alterations, including changes to size, shape, color, visibility, and angles. The impact of these alterations on the YOLOv7 (You Only Look Once) model's detection and classification abilities were analyzed. It reveals that the accuracy of object detection models decreases significantly when exposed to modified traffic signs under unlikely conditions. This study highlights the significance of enhancing the robustness of object detection models in real-life scenarios and the need for further investigation in this area to improve their accuracy and reliability.

Performance Profile of Transformer Fine-Tuning in Multi-GPU Cloud Environments

2021 IEEE International Conference on Big Data (Big Data), Dec 15, 2021

The study presented here focuses on performance characteristics and trade-offs associated with ru... more The study presented here focuses on performance characteristics and trade-offs associated with running machine-learning tasks in multi-GPU environments on both on-site cloud computing resources and commercial cloud services (Azure). Specifically, this study examines these tradeoffs by examining the performance of training and fine-tuning of transformer-based deep-learning (DL) networks on clinical notes and data, a task of critical importance in the medical domain. To this end, we perform DL-related experiments on the widely deployed NVIDIA V100 GPUs and on the newer A100 GPUs connected via NVLink or PCIe. This study analyzes the execution time of major operations to train DL models and investigate popular options to optimize each of them. We examine and present the findings on the impacts that various operations (e.g. data loading into GPUs, training, fine-tuning), optimizations, and system configurations (single vs. multi-GPU, NVLink vs. PCIe) have on the overall training performance.

Understanding the pros and cons of big data analytics

PubMed, Sep 6, 2014

Data Management and Analytics for Medicine and Healthcare

Lecture Notes in Computer Science, 2017

The use of general descriptive names, registered names, trademarks, service marks, etc. in this p... more The use of general descriptive names, registered names, trademarks, service marks, etc. in this publication does not imply, even in the absence of a specific statement, that such names are exempt from the relevant protective laws and regulations and therefore free for general use. The publisher, the authors and the editors are safe to assume that the advice and information in this book are believed to be true and accurate at the date of publication. Neither the publisher nor the authors or the editors give a warranty, express or implied, with respect to the material contained herein or for any errors or omissions that may have been made. The publisher remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Real-Time Discovery Services over Large, Heterogeneous and Complex Healthcare Datasets Using Schema-Less, Column-Oriented Methods

We present a service platform for schema-leess exploration of data and discovery of patient-relat... more We present a service platform for schema-leess exploration of data and discovery of patient-related statistics from healthcare data sets. The architecture of this platform is motivated by the need for fast, schema-less, and flexible approaches to SQL-based exploration and discovery of information embedded in the common, heterogeneously structured healthcare data sets and supporting components (electronic health records, practice management systems, etc.) The motivating use cases described in the paper are clinical trials candidate discovery, and a treatment effectiveness analysis. Following the use cases, we discuss the key features and software architecture of the platform, the underlying core components (Apache Parquet, Drill, the web services server), and the runtime profiles and performance characteristics of the platform. We conclude by showing dramatic speedup with some approaches, and the performance tradeoffs and limitations of others.

Question-Answering System Extracts Information on Injection Drug Use from Clinical Progress Notes

arXiv (Cornell University), May 15, 2023

model. This paper also demonstrates the QA model's ability to extract IDU-related information on ... more model. This paper also demonstrates the QA model's ability to extract IDU-related information on temporally out-of-distribution data. The results indicate that the majority (51%) of the extracted information by the QA model exactly matches the gold-standard answer and 73% of them contain the gold-standard answer with some additional surrounding words. This manuscript has been authored by UT-Battelle, LLC, under contract DE-AC05-00OR22725 with the US Department of Energy (DOE). The US government retains and the publisher, by accepting the article for publication, acknowledges that the US government retains a nonexclusive, paid-up, irrevocable, worldwide license to publish or reproduce the published form of this manuscript, or allow others to do so, for US government purposes.

A Compound Data Poisoning Technique with Significant Adversarial Effects on Transformer-based Text Classification Tasks

Research Square (Research Square), Jan 10, 2023

Transformer-based models have demonstrated much success in various natural language processing (N... more Transformer-based models have demonstrated much success in various natural language processing (NLP) tasks. However, they are often vulnerable to adversarial attacks, such as data poisoning, that can intentionally fool the model into generating incorrect results. In this paper, we present a novel, compound variant of a data poisoning attack on a transformer-based model that maximizes the poisoning effect while minimizing the scope of poisoning. We do so by combining the established data poisoning technique (label flipping) with a novel adversarial artifact selection and insertion technique aimed at minimizing detectability and the scope of the poisoning footprint. We find that using a combination of these two techniques, we achieve a state-of-the-art attack success rate (ASR) of ∼ 90% while poisoning only 0.5% of the original training set, thus minimizing the scope and detectability of the poisoning action.

cpgQA: A Benchmark Dataset for Machine Reading Comprehension Tasks on Clinical Practice Guidelines and a Case Study Using Transfer Learning

IEEE Access, 2023

Biomedical machine reading comprehension (bio-MRC), a crucial task in natural language processing... more Biomedical machine reading comprehension (bio-MRC), a crucial task in natural language processing, is a vital application of a computer-assisted clinical decision support system. It can help clinicians extract critical information effortlessly for clinical decision-making by comprehending and answering questions from biomedical text data. While recent advances in bio-MRC consider text data from resources such as clinical notes and scholarly articles, the clinical practice guidelines (CPGs) are still unexplored in this regard. CPGs are a pivotal component of clinical decision-making at the point of care as they provide recommendations for patient care based on the most up-to-date information available. Although CPGs are inherently terse compared to a multitude of articles, often, clinicians find them lengthy and complicated to use. In this paper, we define a new problem domain-bio-MRC on CPGs-where the ultimate goal is to assist clinicians in efficiently interpreting the clinical practice guidelines using MRC systems. To that end, we develop a manually annotated and subject-matter expert-validated benchmark dataset for the bio-MRC task on CPGs-cpgQA. This dataset aims to evaluate intelligent systems performing MRC tasks on CPGs. Hence, we employ the state-of-the-art MRC models to present a case study illustrating an extensive evaluation of the proposed dataset. We address the problem of lack of training data in this newly defined domain by applying transfer learning. The results show that while the current state-of-the-art models perform well with 78% exact match scores on the dataset, there is still room for improvement, warranting further research on this problem domain. We release the dataset at https://github.com/mmahbub/cpgQA. INDEX TERMS Biomedical machine reading comprehension, clinical practice guidelines, deep learning, natural language processing, transfer learning, transformers.

Unified Medical Language System resources improve sieve-based generation and Bidirectional Encoder Representations from Transformers (BERT)–based ranking for concept normalization

Journal of the American Medical Informatics Association, Jul 27, 2020

Objective: Concept normalization, the task of linking phrases in text to concepts in an ontology,... more Objective: Concept normalization, the task of linking phrases in text to concepts in an ontology, is useful for many downstream tasks including relation extraction, information retrieval, etc. We present a generate-andrank concept normalization system based on our participation in the 2019 National NLP Clinical Challenges Shared Task Track 3 Concept Normalization. Materials and Methods: The shared task provided 13 609 concept mentions drawn from 100 discharge summaries. We first design a sieve-based system that uses Lucene indices over the training data, Unified Medical Language System (UMLS) preferred terms, and UMLS synonyms to generate a list of possible concepts for each mention. We then design a listwise classifier based on the BERT (Bidirectional Encoder Representations from Transformers) neural network to rank the candidate concepts, integrating UMLS semantic types through a regularizer. Results: Our generate-and-rank system was third of 33 in the competition, outperforming the candidate generator alone (81.66% vs 79.44%) and the previous state of the art (76.35%). During postevaluation, the model's accuracy was increased to 83.56% via improvements to how training data are generated from UMLS and incorporation of our UMLS semantic type regularizer. Discussion: Analysis of the model shows that prioritizing UMLS preferred terms yields better performance, that the UMLS semantic type regularizer results in qualitatively better concept predictions, and that the model performs well even on concepts not seen during training. Conclusions: Our generate-and-rank framework for UMLS concept normalization integrates key UMLS features like preferred terms and semantic types with a neural network-based ranking model to accurately link phrases in text to UMLS concepts.

Poster: StreamToxWatch – Data Poisoning Detector in Distributed, Event-based Environments

Poster: Cognitive Cyber – Dynamic, Adaptive Cyber Defense Systems for Massively Distributed, Autonomous, and Ad-hoc Computing Environments

Evaluating Text Analytic Frameworks for Mental Health Surveillance

Reducing suicide incidence among US veterans is one of the highest priorities for the US Departme... more Reducing suicide incidence among US veterans is one of the highest priorities for the US Department of Veterans Affairs (VA). We are implementing a suicide risk detection system, in collaboration with the VA, that would serve as a surveillance system for risk factors appearing in clinical text data. Primary requirements for this system are fast search capability, feature and information extraction, and delivery of data to up-stream natural language processing models. As such, we are evaluating scalable storage solutions on the basis of performance, fault tolerance, and scalability. In this paper we present our current approach to evaluation, preliminary findings, and the work in progress towards a more robust text analysis pipeline.

Improving Quality of Observational Streaming Medical Data by Using Long Short-Term Memory Networks (LSTMs)

We present an exploration of the encoder-decoder structured Long Short-Term Memory Network (LSTM)... more We present an exploration of the encoder-decoder structured Long Short-Term Memory Network (LSTM) as a detector of the anomalous missing observations in streaming medical data by using the difference between the LSTM-reconstructed and observed values as the anomaly detector. We experiment with time-series data from bedside monitoring devices from the available Medical Information Mart for Intensive Care Database (MIMIC). Our results show that not only encoder-decoder LSTM approach works well for detecting the difference between anomalous and normal missing observations in streaming medical data, but also has an imputation potential for the missing data.

Procedural Reasoning System (PRS) architecture for agent-mediated behavioral interventions

We present an architecture in support of intelligent agent-mediated, behavioral interventions in ... more We present an architecture in support of intelligent agent-mediated, behavioral interventions in special education programs for individuals with Autism Spectrum Disorder (ASD). The proposed architecture is a derivative of the Procedural Reasoning System (PRS) architecture with representative, inter-pretative, reasoning, knowledge-based, and procedural control components abstracted from the physical and locomotoric aspects of the agent's placement in the environment. The architecture is designed to serve as the unifying foundation for virtual, mixed reality and embodied implementations, so its behavior-oriented control, reasoning, knowledge base, and inference mechanisms are designed as abstractive.

sciCAN: Single-cell chromatin accessibility and gene expression data integration via Cycle-consistent Adversarial Network

bioRxiv (Cold Spring Harbor Laboratory), Dec 1, 2021

The boom in single-cell technologies has brought a surge of high dimensional data that come from ... more The boom in single-cell technologies has brought a surge of high dimensional data that come from different sources and represent cellular systems from different views. With advances in these single-cell technologies, integrating single-cell data across modalities arises as a new computational challenge. Here, we present an adversarial approach, sciCAN, to integrate single-cell chromatin accessibility and gene expression data in an unsupervised manner. We benchmarked sciCAN with 5 existing methods in 5 scATACseq/scRNA-seq datasets, and we demonstrated that our method dealt with data integration with consistent performance across datasets and better balance of mutual transferring between modalities than the other 5 existing methods. We further applied sciCAN to 10X Multiome data and confirmed that the integrated representation preserves biological relationships within the hematopoietic hierarchy. Finally, we investigated CRISPR-perturbed single-cell K562 ATAC-seq and RNA-seq data to identify cells with related responses to different perturbations in these different modalities.