scholarly journals A Deep Learning Method to Detect Opioid Prescription and Opioid Use Disorder from Electronic Health Records

2021 ◽  
Author(s):  
Aditya Kashyap ◽  
Chris Callison-Burch ◽  
Mary Regina Boland
Keyword(s):  
Deep Learning ◽  
Prediction Models ◽  
The United States ◽  
Opioid Use Disorder ◽  
Unstructured Data ◽  
Learning Approaches ◽  
Learning Models ◽  
Opioid Use ◽  
Opioid Prescription ◽  
Patients At Risk

Objective: As the opioid epidemic continues across the United States, methods are needed to accurately and quickly identify patients at risk for opioid use disorder (OUD). The purpose of this study is to develop two predictive algorithms: one to predict opioid prescription and one to predict OUD. Materials and Methods: We developed an informatics algorithm that trains two deep learning models over patient EHRs using the MIMIC-III database. We utilize both the structured and unstructured parts of the EHR and show that it is possible to predict both of these challenging outcomes. Results: Our deep learning models incorporate both structured and unstructured data elements from the EHRs to predict opioid prescription with an F1-score of 0.88 +/- 0.003 and an AUC-ROC of 0.93 +/- 0.002. We also constructed a model to predict OUD diagnosis achieving an F1-score of 0.82 +/- 0.05 and AUC-ROC of 0.94 +/- 0.008. Discussion: Our model for OUD prediction outperformed prior algorithms for specificity, F1 score and AUC-ROC while achieving equivalent sensitivity. This demonstrates the importance of a.) deep learning approaches in predicting OUD and b.) incorporating both structured and unstructured data for this prediction task. No prediction models for opioid prescription as an outcome were found in the literature and therefore this represents an important contribution of our work as opioid prescriptions are more common than OUDs. Conclusion: Algorithms such as those described in this paper will become increasingly important to understand the drivers underlying this national epidemic.

scholarly journals Modeling Housing Rent in the Atlanta Metropolitan Area Using Textual Information and Deep Learning

2019 ◽  
Vol 8 (8) ◽  
pp. 349 ◽  
Author(s):  
Xiaolu Zhou ◽  
Weitian Tong ◽  
Dongying Li
Keyword(s):  
Neural Network ◽  
Deep Learning ◽  
Information Exchange ◽  
Critical Role ◽  
Rental Housing ◽  
The United States ◽  
Real Estate Market ◽  
Learning Approaches ◽  
Learning Models ◽  
Textual Information

The rental housing market plays a critical role in the United States real estate market. In addition, rent changes are also indicators of urban transformation and social phenomena. However, traditional data sources for market rent prediction are often inaccurate or inadequate at covering large geographies. With the development of housing information exchange platforms such as Craigslist, user-generated rental listings now provide big data that cover wide geographies and are rich in textual information. Given the importance of rent prediction in urban studies, this study aims to develop and evaluate models of rental market dynamics using deep learning approaches on spatial and textual data from Craigslist rental listings. We tested a number of machine learning and deep learning models (e.g., convolutional neural network, recurrent neural network) for the prediction of rental prices based on data collected from Atlanta, GA, USA. With textual information alone, deep learning models achieved an average root mean square error (RMSE) of 288.4 and mean absolute error (MAE) of 196.8. When combining textual information with location and housing attributes, the integrated model achieved an average RMSE of 227.9 and MAE of 145.4. These approaches can be applied to assess the market value of rental properties, and the prediction results can be used as indicators of a variety of urban phenomena and provide practical references for home owners and renters.

Increasing Naloxone Knowledge and Use Through Direct-to-Patient Education

2020 ◽  
Vol 36 (6) ◽  
pp. 237-242
Author(s):  
Kathryn Litten ◽  
Lucas G. Hill ◽  
Aida Garza ◽  
Maaya Srinivasa
Keyword(s):  
At Risk ◽  
Knowledge Score ◽  
The United States ◽  
Opioid Therapy ◽  
Opioid Use Disorder ◽  
Opioid Overdose ◽  
Opioid Use ◽  
Education Access ◽  
Patients At Risk ◽  
Change In Mean

Background: In the United States, opioid overdoses account for 130 deaths daily. Barriers to obtaining naloxone, the drug-of-choice for opioid overdose reversal, include limited education, access, and perceptions of provider judgement. Objectives: This study aimed to assess the efficacy of mailed education about naloxone, with or without a live teaching seminar, to patients at risk for opioid overdose. Methods: This observational study was conducted in a federally qualified health system. A phone presurvey was administered to patients on long-term opioid therapy or with a diagnosis of opioid use disorder to assess opioid overdose-related knowledge. Subjects were mailed a handout about naloxone and an invitation to receive naloxone at no cost at a seminar. Three-month phone postsurveys were conducted. The primary outcome was change in mean knowledge score from presurvey to postsurvey. Secondary outcomes included scores on individual survey items, naloxone prescriptions provided, and overdose reversals reported. Results: Ninety-four patients received mailed education. Sixty-two subjects took presurveys and 23 took 3-month follow-up surveys. Five subjects attended the live seminar. The mean cumulative knowledge score improved by 8.7% from the presurvey to the postsurvey. During the study period, one new naloxone prescription was written and one overdose reversal was reported. Conclusion: Direct-to-patient mailed education slightly improved knowledge regarding naloxone and opioid overdose response, and it may have led to one successful overdose reversal. Mailing education to a larger population of patients at risk for opioid overdose may be necessary to observe a substantial clinical impact.

scholarly journals Comparison of deep learning with traditional models to predict preventable acute care use and spending among heart failure patients

2021 ◽  
Vol 11 (1) ◽  
Author(s):  
Maor Lewis ◽  
Guy Elad ◽  
Moran Beladev ◽  
Gal Maor ◽  
Kira Radinsky ◽  
...  
Keyword(s):  
Heart Failure ◽  
Deep Learning ◽  
Prediction Models ◽  
Evaluation Metrics ◽  
Gradient Boosting ◽  
Learning Approaches ◽  
Learning Models ◽  
Preventable Hospitalizations ◽  
Targeted Interventions ◽  
Ed Visits

AbstractRecent health reforms have created incentives for cardiologists and accountable care organizations to participate in value-based care models for heart failure (HF). Accurate risk stratification of HF patients is critical to efficiently deploy interventions aimed at reducing preventable utilization. The goal of this paper was to compare deep learning approaches with traditional logistic regression (LR) to predict preventable utilization among HF patients. We conducted a prognostic study using data on 93,260 HF patients continuously enrolled for 2-years in a large U.S. commercial insurer to develop and validate prediction models for three outcomes of interest: preventable hospitalizations, preventable emergency department (ED) visits, and preventable costs. Patients were split into training, validation, and testing samples. Outcomes were modeled using traditional and enhanced LR and compared to gradient boosting model and deep learning models using sequential and non-sequential inputs. Evaluation metrics included precision (positive predictive value) at k, cost capture, and Area Under the Receiver operating characteristic (AUROC). Deep learning models consistently outperformed LR for all three outcomes with respect to the chosen evaluation metrics. Precision at 1% for preventable hospitalizations was 43% for deep learning compared to 30% for enhanced LR. Precision at 1% for preventable ED visits was 39% for deep learning compared to 33% for enhanced LR. For preventable cost, cost capture at 1% was 30% for sequential deep learning, compared to 18% for enhanced LR. The highest AUROCs for deep learning were 0.778, 0.681 and 0.727, respectively. These results offer a promising approach to identify patients for targeted interventions.

scholarly journals Deep learning systems detect dysplasia with human-like accuracy using histopathology and probe-based confocal laser endomicroscopy

2021 ◽  
Vol 11 (1) ◽  
Author(s):  
Shan Guleria ◽  
Tilak U. Shah ◽  
J. Vincent Pulido ◽  
Matthew Fasullo ◽  
Lubaina Ehsan ◽  
...  
Keyword(s):  
Deep Learning ◽  
Diagnostic Accuracy ◽  
High Sensitivity ◽  
Confocal Laser Endomicroscopy ◽  
Confocal Laser ◽  
Learning Approaches ◽  
Learning Models ◽  
Whole Slide Image ◽  
Slide Image ◽  
Level Model

AbstractProbe-based confocal laser endomicroscopy (pCLE) allows for real-time diagnosis of dysplasia and cancer in Barrett’s esophagus (BE) but is limited by low sensitivity. Even the gold standard of histopathology is hindered by poor agreement between pathologists. We deployed deep-learning-based image and video analysis in order to improve diagnostic accuracy of pCLE videos and biopsy images. Blinded experts categorized biopsies and pCLE videos as squamous, non-dysplastic BE, or dysplasia/cancer, and deep learning models were trained to classify the data into these three categories. Biopsy classification was conducted using two distinct approaches—a patch-level model and a whole-slide-image-level model. Gradient-weighted class activation maps (Grad-CAMs) were extracted from pCLE and biopsy models in order to determine tissue structures deemed relevant by the models. 1970 pCLE videos, 897,931 biopsy patches, and 387 whole-slide images were used to train, test, and validate the models. In pCLE analysis, models achieved a high sensitivity for dysplasia (71%) and an overall accuracy of 90% for all classes. For biopsies at the patch level, the model achieved a sensitivity of 72% for dysplasia and an overall accuracy of 90%. The whole-slide-image-level model achieved a sensitivity of 90% for dysplasia and 94% overall accuracy. Grad-CAMs for all models showed activation in medically relevant tissue regions. Our deep learning models achieved high diagnostic accuracy for both pCLE-based and histopathologic diagnosis of esophageal dysplasia and its precursors, similar to human accuracy in prior studies. These machine learning approaches may improve accuracy and efficiency of current screening protocols.

scholarly journals Assessing the Relationships Between COVID-19 Stay-at-Home Orders and Opioid Overdoses in the State of Pennsylvania

2021 ◽  
pp. 002204262110063
Author(s):  
Brian King ◽  
Ruchi Patel ◽  
Andrea Rishworth
Keyword(s):  
Age Groups ◽  
The United States ◽  
Opioid Use Disorder ◽  
Opioid Overdose ◽  
Policy Recommendations ◽  
Opioid Use ◽  
Fentanyl Analogs ◽  
Measure Change ◽  
Using Data ◽  
At Home

COVID-19 is compounding opioid use disorder throughout the United States. While recent commentaries provide useful policy recommendations, few studies examine the intersection of COVID-19 policy responses and patterns of opioid overdose. We examine opioid overdoses prior to and following the Pennsylvania stay-at-home order implemented on April 1, 2020. Using data from the Pennsylvania Overdose Information Network, we measure change in monthly incidents of opioid-related overdose pre- versus post-April 1, and the significance of change by gender, age, race, drug class, and naloxone doses administered. Findings demonstrate statistically significant increases in overdose incidents among both men and women, White and Black groups, and several age groups, most notably the 30–39 and 40–49 ranges, following April 1. Significant increases were observed for overdoses involving heroin, fentanyl, fentanyl analogs or other synthetic opioids, pharmaceutical opioids, and carfentanil. The study emphasizes the need for opioid use to be addressed alongside efforts to mitigate and manage COVID-19 infection.

scholarly journals Deep Learning with Neuroimaging and Genomics in Alzheimer’s Disease

2021 ◽  
Vol 22 (15) ◽  
pp. 7911
Author(s):  
Eugene Lin ◽  
Chieh-Hsin Lin ◽  
Hsien-Yuan Lane
Keyword(s):  
Alzheimer’S Disease ◽  
Alzheimer's Disease ◽  
Deep Learning ◽  
Future Research ◽  
Learning Approaches ◽  
Learning Models ◽  
Learning Techniques ◽  
Neuroimaging Data ◽  
Similarities And Differences ◽  
Normal Controls

A growing body of evidence currently proposes that deep learning approaches can serve as an essential cornerstone for the diagnosis and prediction of Alzheimer’s disease (AD). In light of the latest advancements in neuroimaging and genomics, numerous deep learning models are being exploited to distinguish AD from normal controls and/or to distinguish AD from mild cognitive impairment in recent research studies. In this review, we focus on the latest developments for AD prediction using deep learning techniques in cooperation with the principles of neuroimaging and genomics. First, we narrate various investigations that make use of deep learning algorithms to establish AD prediction using genomics or neuroimaging data. Particularly, we delineate relevant integrative neuroimaging genomics investigations that leverage deep learning methods to forecast AD on the basis of incorporating both neuroimaging and genomics data. Moreover, we outline the limitations as regards to the recent AD investigations of deep learning with neuroimaging and genomics. Finally, we depict a discussion of challenges and directions for future research. The main novelty of this work is that we summarize the major points of these investigations and scrutinize the similarities and differences among these investigations.

scholarly journals “It’s way more than just writing a prescription”: A qualitative study of preferences for integrated versus non-integrated treatment models among individuals with opioid use disorder

2021 ◽  
Vol 16 (1) ◽  
Author(s):  
Elizabeth C. Saunders ◽  
Sarah K. Moore ◽  
Olivia Walsh ◽  
Stephen A. Metcalf ◽  
Alan J. Budney ◽  
...  
Keyword(s):  
Qualitative Study ◽  
The United States ◽  
Opioid Use Disorder ◽  
Integrated Treatment ◽  
Medical Setting ◽  
Integrated Models ◽  
Opioid Use ◽  
Medical Settings ◽  
Psychosocial Services ◽  
Treatment Models

Abstract Background Increasingly, treatment for opioid use disorder (OUD) is offered in integrated treatment models addressing both substance use and other health conditions within the same system. This often includes offering medications for OUD in general medical settings. It remains uncertain whether integrated OUD treatment models are preferred to non-integrated models, where treatment is provided within a distinct treatment system. This study aimed to explore preferences for integrated versus non-integrated treatment models among people with OUD and examine what factors may influence preferences. Methods This qualitative study recruited participants (n = 40) through Craigslist advertisements and flyers posted in treatment programs across the United States. Participants were 18 years of age or older and scored a two or higher on the heroin or opioid pain reliever sections of the Tobacco, Alcohol, Prescription Medications, and Other Substances (TAPS) Tool. Each participant completed a demographic survey and a telephone interview. The interviews were coded and content analyzed. Results While some participants preferred receiving OUD treatment from an integrated model in a general medical setting, the majority preferred non-integrated models. Some participants preferred integrated models in theory but expressed concerns about stigma and a lack of psychosocial services. Tradeoffs between integrated and non-integrated models were centered around patient values (desire for anonymity and personalization, fear of consequences), the characteristics of the provider and setting (convenience, perceived treatment effectiveness, access to services), and the patient-provider relationship (disclosure, trust, comfort, stigma). Conclusions Among this sample of primarily White adults, preferences for non-integrated versus integrated OUD treatment were mixed. Perceived benefits of integrated models included convenience, potential for treatment personalization, and opportunity to extend established relationships with medical providers. Recommendations to make integrated treatment more patient-centered include facilitating access to psychosocial services, educating patients on privacy, individualizing treatment, and prioritizing the patient-provider relationship. This sample included very few minorities and thus findings may not be fully generalizable to the larger population of persons with OUD. Nonetheless, results suggest a need for expansion of both OUD treatment in specialty and general medical settings to ensure access to preferred treatment for all.

scholarly journals Discovering Non-Invasive Biomarkers Predictive of Opioid Use Disorder Treatment Response

CNS Spectrums ◽  
2021 ◽  
Vol 26 (2) ◽  
pp. 173-173
Author(s):  
Amir Levine ◽  
Kelly Clemenza ◽  
Shira Weiss ◽  
Adam Bisaga ◽  
Erez Eitan ◽  
...  
Keyword(s):  
Treatment Response ◽  
Mononuclear Cells ◽  
Specific Treatment ◽  
The United States ◽  
Opioid Use Disorder ◽  
Response To Treatment ◽  
Post Translational Modification ◽  
Opioid Use ◽  
Peripheral Blood Mononuclear ◽  
Long Term Treatment

AbstractBackgroundOpioid use disorder (OUD) continues to be the driving force behind drug overdoses in the United States, killing nearly 47,000 people in 2018 alone. The increasing presence of deadlier fentanyl analogues in the heroin drug supply are putting users at a greater risk for overdose than ever before. Admissions to treatment programs for OUD have also nearly doubled since 2006, yet relapse rates remain high. In response to these alarming statistics, developing approaches to reduce overdose deaths has become an area of high priority. As it is not yet known which patients are most likely to benefit from a specific treatment, there is a dire need to utilize new molecular tools to guide precision medicine approaches and improve treatment outcomes. Here we describe a proof-of-concept study evaluating plasma-derived extracellular vesicle (EV) signatures and how they differ in patients who responded to two pharmacologically contrasting treatments for OUD: the μOR agonist methadone, and the μOR antagonist naltrexone.MethodsWe obtained blood samples from patients with OUD who remained abstinent from illicit opioids for at least 3 months during treatment with methadone (n=5) and naltrexone (n=5), as well as matched healthy controls (n=5). EVs were isolated from plasma and histones were isolated from peripheral blood mononuclear cells (PBMCs). EVs were then analyzed for lipid and histone post-translational modification (PTM) content using liquid chromatography-mass spectrometry. EV miRNA cargo was determined by RNA sequencing.ResultsWe found one lipid class and six miRNAs that differed significantly between the naltrexone group and the methadone and control groups. We also found that histone H3acK9acK14 was increasingly acetylated in PMBCs from both the methadone and naltrexone groups compared to controls.DiscussionNaltrexone, which is used in treatment of OUD and other substance use disorders as well as disorders of impulse control, was found to have multiple potential corresponding molecular signatures that can be identified after long-term treatment. It remains to be seen if these markers can also be a good predictor for treatment response. In addition, significant gender differences in EV content are found between men and women with OUD, which supports the importance of examining changes in response to treatment in a gender informed way.

Opioid use disorder and overdose among youth following an initial opioid prescription

Addiction ◽  
2021 ◽  
Author(s):  
Scott E. Hadland ◽  
Sarah M. Bagley ◽  
Mam Jarra Gai ◽  
Joel J. Earlywine ◽  
Samantha F. Schoenberger ◽  
...  
Keyword(s):  
Opioid Use Disorder ◽  
Opioid Use ◽  
Opioid Prescription
Sign in / Sign up

Export Citation Format

Share Document