scholarly journals An Unsupervised Machine Learning Paradigm for Artifact Removal from Electrodermal Activity in an Uncontrolled Clinical Setting

2021 ◽  
Author(s):  
Sandya Subramanian ◽  
Bryan Tseng ◽  
Riccardo Barbieri ◽  
Emery N. Brown
Keyword(s):  
Unsupervised Learning ◽  
Operating Room ◽  
Electrodermal Activity ◽  
High Sensitivity ◽  
Clinical Settings ◽  
Artifact Removal ◽  
Clinical Integration ◽  
Learning Methods ◽  
Artifact Detection ◽  
Small Set

Objective: Electrodermal activity (EDA) reflects sympathetic nervous system activity through sweating-related changes in skin conductance. To enable EDA data to be used robustly in clinical settings, we need to develop artifact detection and removal frameworks that can handle the types of interference experienced in clinical settings. Methods: We collected EDA data from 69 subjects while they were undergoing surgery in the operating room. We then built an artifact removal framework using unsupervised learning methods and informed features to remove the heavy artifact that resulted from the use of surgical electrocautery during the surgery and compared it to other existing methods for artifact removal from EDA data. Results: Our framework was able to remove the vast majority of artifact from the EDA data across all subjects with high sensitivity (94%) and specificity (90%). In contrast, existing methods used for comparison struggled to be sufficiently sensitive and specific, and none effectively removed artifact even if it was identifiable. In addition, the use of unsupervised learning methods in our framework removes the need for manually labeled datasets for training. Conclusion: Our framework allows for robust removal of heavy artifact from EDA data in clinical settings such as surgery. Since this framework only relies on a small set of informed features, it can be expanded to other modalities such as ECG and EEG. Significance: Robust artifact removal from EDA data is the first step to enable clinical integration of EDA as part of standard monitoring in settings such as the operating room.

Evaluating disaster-related tweet credibility using content-based and user-based features

2021 ◽  
Vol ahead-of-print (ahead-of-print) ◽  
Author(s):  
Nasser Assery ◽  
Yuan (Dorothy) Xiaohong ◽  
Qu Xiuli ◽  
Roy Kaushik ◽  
Sultan Almalki
Keyword(s):  
Machine Learning ◽  
Unsupervised Learning ◽  
Supervised Learning ◽  
Emergency Response ◽  
Learning Model ◽  
Performance Comparison ◽  
Supervised Machine Learning ◽  
Learning Methods ◽  
Content Type ◽  
Machine Learning Classification

Purpose This study aims to propose an unsupervised learning model to evaluate the credibility of disaster-related Twitter data and present a performance comparison with commonly used supervised machine learning models. Design/methodology/approach First historical tweets on two recent hurricane events are collected via Twitter API. Then a credibility scoring system is implemented in which the tweet features are analyzed to give a credibility score and credibility label to the tweet. After that, supervised machine learning classification is implemented using various classification algorithms and their performances are compared. Findings The proposed unsupervised learning model could enhance the emergency response by providing a fast way to determine the credibility of disaster-related tweets. Additionally, the comparison of the supervised classification models reveals that the Random Forest classifier performs significantly better than the SVM and Logistic Regression classifiers in classifying the credibility of disaster-related tweets. Originality/value In this paper, an unsupervised 10-point scoring model is proposed to evaluate the tweets’ credibility based on the user-based and content-based features. This technique could be used to evaluate the credibility of disaster-related tweets on future hurricanes and would have the potential to enhance emergency response during critical events. The comparative study of different supervised learning methods has revealed effective supervised learning methods for evaluating the credibility of Tweeter data.

D-dimer Testing and Venous Thromboembolism

2009 ◽  
Vol 03 (01) ◽  
pp. 37
Author(s):  
Domenico Prisco ◽  
Elisa Grifoni ◽  
Daniela Poli ◽  
◽  
◽  
...  
Keyword(s):  
Venous Thromboembolism ◽  
Diagnostic Yield ◽  
Prediction Rule ◽  
High Sensitivity ◽  
Clinical Prediction Rule ◽  
Patient Characteristics ◽  
Optimal Timing ◽  
Clinical Settings ◽  
Diagnostic Work ◽  
D Dimer

With its high sensitivity and negative predictive value, D-dimer (DD) testing has gained a role in the diagnostic work-up of suspected venous thromboembolism (VTE) for the exclusion of the disease, potentially reducing the need for imaging tests. The diagnostic yield of DD testing is affected not only by the choice of the appropriate assay for its measurement, but also by patient characteristics. As a consequence, its clinical usefulness for the exclusion of suspected VTE should be carefully evaluated in special clinical settings. There is increasing evidence that DD testing after anticoagulation withdrawal for a first unprovoked VTE episode may be useful to identify patients at higher risk of recurrence, and may help clinicians with the decision of whether to continue or stop anticoagulant treatment. However, further studies are needed to establish the optimal timing of DD testing and the best DD cut-off level that predicts recurrence, and to develop a clinical prediction rule for recurrent VTE.

Unsupervised Learning Methods for Identification of Defects in Heterogeneous Materials

2020 ◽  
Author(s):  
MUTHU RAM PRABHU ELENCHEZHIAN ◽  
VAMSEE VADLAMUDI ◽  
RASSEL RAIHAN ◽  
KENNETH REIFSNIDER
Keyword(s):  
Unsupervised Learning ◽  
Heterogeneous Materials ◽  
Learning Methods

Validation of an exome and transcriptome based diagnostic platform enabling clinical cancer therapy selection and emerging composite biomarkers for immunotherapy.

2020 ◽  
Vol 38 (15_suppl) ◽  
pp. e15583-e15583
Author(s):  
Juan-Sebstian Saldivar ◽  
Jason Harris ◽  
Sejal Desai ◽  
Erin Ayash ◽  
Prateek Tandon ◽  
...  
Keyword(s):  
Biomarker Discovery ◽  
High Sensitivity ◽  
Cancer Genes ◽  
Patient Response ◽  
Diagnostic Biomarkers ◽  
Dna And Rna ◽  
Clinical Biomarkers ◽  
Whole Exome ◽  
Ffpe Samples ◽  
Small Set

e15583 Background: While immunotherapy has become a pillar of cancer treatment, diagnostic biomarkers that consistently predict patient response to these therapies have remained elusive. There is an increasing need for the development of integrative, composite biomarkers that can model the complex biology driving response and/or resistance to immunotherapy more effectively than existing single-analyte approaches. However, the majority of current cancer diagnostic panels, with their focus on a small set of genes, provide limited ability to support these emerging advanced biomarkers. Methods: To address these limitations, we developed and validated NeXT Dx, a comprehensive enhanced exome and transcriptome based diagnostic platform designed to simultaneously characterize tumor and immune genomics from a single limited FFPE sample. To achieve higher accuracy and sensitivity for an exome scale diagnostic platform, we developed an augmented exome assay that improves uniformity of coverage across all ~20,000 genes, including boosted coverage of 248 clinically-relevant cancer genes. We validated this assay using genomic DNA and RNA extracted from tumor-derived cell-lines, constructs, clinical FFPE samples, and proficiency testing samples. The assay utilizes > = 25ng of co-extracted DNA and RNA which were sequenced using Illumina NovaSeq instruments at our CAP-accredited, CLIA-certified laboratory. Additional assay enhancements for HLA, immune repertoire, and oncoviruses were designed to further optimize the platform for immunotherapy biomarker discovery applications. Results: Validation of NeXT Dx demonstrated a performance of 99.5% sensitivity and 99.8% positive predictive value (PPV) for SNVs with > = 5% AF; 98.7% sensitivity and 97.4% PPV for indels with > = 10% AF; 97.2% sensitivity and 94.6% PPV for CNAs in samples with > = 30% tumor content; 94.9% sensitivity and 94.9% PPV for fusions; and a 2.1% error rate for MSI classification. TMB was calculated using gold-standard whole exome data from SNVs and indels. Typical median coverage depth was > 1,000X for 248 clinically-relevant genes, ~300X for the remaining (whole exome) footprint. Conclusions: With NeXT Dx, we demonstrate a exome/transcriptome scale diagnostic platform that can detect current clinical biomarkers with high sensitivity as well as support emerging, advanced biomarkers that integrate across both tumor and immune features.

Sign in / Sign up

Export Citation Format

Share Document