A5072774346- Machine Learning in Healthcare
- Explainable Artificial Intelligence (XAI)
- Biomedical Text Mining and Ontologies
- Topic Modeling
- Ethics and Social Impacts of AI
- Opioid Use Disorder Treatment
- Natural Language Processing Techniques
- Sepsis Diagnosis and Treatment
- Insurance, Mortality, Demography, Risk Management
- Data-Driven Disease Surveillance
- Data Stream Mining Techniques
- Gaussian Processes and Bayesian Inference
- Artificial Intelligence in Healthcare
- Bayesian Modeling and Causal Inference
- Parkinson's Disease Mechanisms and Treatments
- Machine Learning and Data Classification
- COVID-19 epidemiological studies
- Hemoglobinopathies and Related Disorders
- Time Series Analysis and Forecasting
- COVID-19 and healthcare impacts
- HIV, Drug Use, Sexual Risk
- Mechanical Circulatory Support Devices
- Healthcare cost, quality, practices
- Public Health Policies and Education
- Nuclear Receptors and Signaling
National Institutes of Health
2023-2025
University of Washington
2007-2025
University of Puget Sound
2024-2025
Carnegie Mellon University
2018-2024
United States National Library of Medicine
2023-2024
Uniformed Services University of the Health Sciences
2021
Madigan Army Medical Center
2021
Yale University
2020
University of Wisconsin–Madison
2012-2015
Sepsis is a heterogeneous syndrome. Identification of distinct clinical phenotypes may allow more precise therapy and improve care.To derive sepsis from data, determine their reproducibility correlation with host-response biomarkers outcomes, assess the potential causal relationship results randomized trials (RCTs).Retrospective analysis data sets using statistical, machine learning, simulation tools. Phenotypes were derived among 20 189 total patients (16 552 unique patients) who met...
<h3>Importance</h3> Current approaches to identifying individuals at high risk for opioid overdose target many patients who are not truly risk. <h3>Objective</h3> To develop and validate a machine-learning algorithm predict among Medicare beneficiaries with least 1 prescription. <h3>Design, Setting, Participants</h3> A prognostic study was conducted between September 1, 2017, December 31, 2018. Participants (n = 560 057) included fee-for-service without cancer filled or more prescriptions...
Electronic health records (EHRs) are an emerging relational domain with large potential to improve clinical outcomes. We apply two statistical learning (SRL) algorithms the task of predicting primary myocardial infarction. show that one SRL algorithm, functional gradient boosting, outperforms propositional learners particularly in medically relevant high‐recall region. observe both predict outcomes better than their analogs and suggest how our methods can augment current epidemiological practices.
Significance To study the COVID-19 pandemic, its effects on society, and measures for reducing spread, researchers need detailed data course of pandemic. Standard public health streams suffer inconsistent reporting frequent, unexpected revisions. They also miss other aspects a population’s behavior that are worthy consideration. We present an open database COVID signals in United States, measured at county level updated daily. This includes traditionally reported cases deaths, many others:...
Recent works in artificial intelligence fairness attempt to mitigate discrimination by proposing constrained optimization programs that achieve parity for some statistic. Most assume availability of the class label, which is impractical many real-world applications such as precision medicine, actuarial analysis and recidivism prediction. Here we consider longitudinal right-censored environments, where time event might be unknown, resulting censorship label inapplicability existing studies....
Objective To develop and validate a machine-learning algorithm to improve prediction of incident OUD diagnosis among Medicare beneficiaries with ≥1 opioid prescriptions. Methods This prognostic study included 361,527 fee-for-service beneficiaries, without cancer, filling prescriptions from 2011–2016. We randomly divided into training, testing, validation samples. measured 269 potential predictors including socio-demographics, health status, patterns use, provider-level regional-level factors...
Electronic health records (EHRs) are an emerging relational domain with large potential to improve clinical outcomes. We apply two statistical learning (SRL) algorithms the task of predicting primary myocardial infarction. show that one SRL algorithm, functional gradient boosting, outperforms propositional learners particularly in medically-relevant high recall region. observe both predict outcomes better than their analogs and suggest how our methods can augment current epidemiological practices.
There has been concern within the artificial intelligence (AI) community and broader society regarding potential lack of fairness AI-based decision-making systems. Surprisingly, there is little work quantifying guaranteeing in presence uncertainty which prevalent many socially sensitive applications, ranging from marketing analytics to actuarial analysis recidivism prediction instruments. To this end, we study a longitudinal censored learning problem subject constraints, where require that...
Geocoding methods vary among spatial epidemiology studies. Errors in the geocoding process and differential match rates may reduce study validity. We compared two using 8,157 Washington State addresses. The multi-stage method implemented by state health department used a sequence of local national reference files. single-stage single file. For each address geocoded both methods, we measured distance between locations assigned method. Area-level characteristics were collected from census...
Reducing preventable hospital re-admissions in Sickle Cell Disease (SCD) could potentially improve outcomes and decrease healthcare costs. In a retrospective study of electronic health records, we hypothesized Machine-Learning (ML) algorithms may outperform standard re-admission scoring systems (LACE HOSPITAL indices). Participants (n = 446) included patients with SCD at least one unplanned inpatient encounter between January 1, 2013, November 2018. Patients were randomly partitioned into...
Nazneen Fatema Rajani, Rui Zhang, Yi Chern Tan, Stephan Zheng, Jeremy Weiss, Aadit Vyas, Abhijit Gupta, Caiming Xiong, Richard Socher, Dragomir Radev. Proceedings of the 58th Annual Meeting Association for Computational Linguistics. 2020.
Abstract The COVID-19 pandemic presented enormous data challenges in the United States. Policy makers, epidemiological modelers, and health researchers all require up-to-date on relevant public behavior, ideally at fine spatial temporal resolution. COVIDcast API is our attempt to fill this need: operational since April 2020, it provides open access both traditional surveillance signals (cases, deaths, hospitalizations) many auxiliary indicators of activity, such as extracted from...
The importance of understanding and correcting algorithmic bias in machine learning (ML) has led to an increase research on fairness ML, which typically assumes that the underlying data is independent identically distributed (IID). However, reality, often represented using non-IID graph structures capture connections among individual units. To address ML systems, it crucial bridge gap between traditional literature designed for IID ubiquity data. In this survey, we review such recent advance...
As Artificial Intelligence (AI) is used in more applications, the need to consider and mitigate biases from learned models has followed. Most works developing fair learning algorithms focus on offline setting. However, many real-world applications data comes an online fashion needs be processed fly. Moreover, practical application, there a trade-off between accuracy fairness that accounted for, but current methods often have multiple hyperparameters with non-trivial interaction achieve...