Abstract WP405: Automated Detection of Hemorrhagic Stroke From Non-Contrast Computed Tomography: A Machine Learning Approach

Introduction: Automated neuroimaging analysis is being used increasingly in the acute ischemic stroke (AIS) evaluation. However, current algorithms do not factor in an assessment of intracranial hemorrhage (ICH) in the workflow. In this study we present a machine learning (ML) algorithm that uses brain symmetry information to detect ICH. Methods: We prospectively collected non-contrast CT (NCCT) images on patients that presented to the Emergency Department for AIS evaluation between 2017 and 2019. Patients were included if they underwent technically adequate NCCT imaging. Diagnoses of ICH, AIS and non-stroke were confirmed by experienced neuroradiologists as well as review of the clinical record. A ML algorithm which integrates symmetry features as well as standard features for the whole brain was trained on 80% of the sample and validated on the remaining images. Training was performed without any prior segmentation. Evaluation of the model performance was conducted using receiver-operator curve and area under the curve (AUC) analysis. Results are given as median [IQR] and [AUC 95% CI]. Results: Among the 568 patients that met inclusion criteria, median age was 65 [55-76], 47% were female and 34% were white. 128 (23%) patients were determined to have ICH and 440 as non-ICH (70% AIS and 30% non-stroke). Among ICH patients, 108 (84%) had a supratentorial ICH. When analyzing the regions of the CT images that most strongly contributed to the algorithm’s diagnostic decisions, they corresponded with the regions of ICH (Fig. 1A). On the external validation data set, the algorithm successfully detected ICH (Fig. 1B) with high accuracy (AUC 0.99 [0.97-1.00]). Conclusion: We have developed a symmetry-sensitive ML method that can with very high fidelity identify ICH in an automated fashion. Without prior training, the algorithm autonomously was able to learn ICH location. These results may help contribute to an automated imaging workflow for all stroke evaluations, not just AIS.

Download Full-text

Artificial intelligence method to classify ophthalmic emergency severity based on symptoms: a validation study

BMJ Open ◽

10.1136/bmjopen-2020-037161 ◽

2020 ◽

Vol 10 (7) ◽

pp. e037161

Author(s):

Hyunmin Ahn

Keyword(s):

Artificial Intelligence ◽

Machine Learning ◽

Model Performance ◽

Armed Forces ◽

Validation Data ◽

Data Set ◽

Orange Yellow ◽

Hospital Visits ◽

Input Variables ◽

Fully Connected

ObjectivesWe investigated the usefulness of machine learning artificial intelligence (AI) in classifying the severity of ophthalmic emergency for timely hospital visits.Study designThis retrospective study analysed the patients who first visited the Armed Forces Daegu Hospital between May and December 2019. General patient information, events and symptoms were input variables. Events, symptoms, diagnoses and treatments were output variables. The output variables were classified into four classes (red, orange, yellow and green, indicating immediate to no emergency cases). About 200 cases of the class-balanced validation data set were randomly selected before all training procedures. An ensemble AI model using combinations of fully connected neural networks with the synthetic minority oversampling technique algorithm was adopted.ParticipantsA total of 1681 patients were included.Major outcomesModel performance was evaluated using accuracy, precision, recall and F1 scores.ResultsThe accuracy of the model was 99.05%. The precision of each class (red, orange, yellow and green) was 100%, 98.10%, 92.73% and 100%. The recalls of each class were 100%, 100%, 98.08% and 95.33%. The F1 scores of each class were 100%, 99.04%, 95.33% and 96.00%.ConclusionsWe provided support for an AI method to classify ophthalmic emergency severity based on symptoms.

Download Full-text

Email Fraud Attack Detection Using Hybrid Machine Learning Approach

Recent Patents on Computer Science ◽

10.2174/2213275912666190617162707 ◽

2019 ◽

Vol 12 ◽

Author(s):

Yousef A. Yaseen ◽

Malik Qasaimeh ◽

Raad S. Al-Qassas ◽

Mustafa Al-Fayoumi

Keyword(s):

Machine Learning ◽

Area Under The Curve ◽

Attack Detection ◽

Majority Voting ◽

Data Set ◽

Hybrid Classifier ◽

Machine Learning Approach ◽

Hybrid Machine ◽

Communication Methods ◽

E Mail

Background: : E-mail is an efficient way to communicate. It is one of the most commonly used communication methods, and it can be used for achieving legitimate and illegitimate activities. Many features that can be effective in detecting email fraud attacks, are still under investigation. Method: : This paper proposes an improved classification accuracy for fraudulent emails that is implemented through feature extraction and hybrid Machine Learning (ML) classifier that combines Adaboost and Majority Voting. Eleven ML classifiers are evaluated experimentally within the hybrid classifier, and the performance of the email fraud filtering is evaluated by using WEKA and R tool on a data set of 9298 email messages. Results: : The performance evaluation shows that the hybrid model of Voting using Adaboost outperforms all other classifiers, with the lowest Error Rate of 0.6991%, highest f1-measure of 99.30%, and highest Area Under the Curve (AUC) of 99.9%. Conclusion:: The utilized proposed email features with the combination of Adaboost and Voting algorithms prove the efficiency of fraud email detection.

Download Full-text

Machine-Learning Models for Combinatorial Catalyst Discovery

MRS Proceedings ◽

10.1557/proc-804-jj11.5 ◽

2003 ◽

Vol 804 ◽

Cited By ~ 1

Author(s):

Gregory A. Landrum ◽

Julie Penzotti ◽

Santosh Putta

Keyword(s):

Machine Learning ◽

External Validation ◽

Machine Learning Algorithms ◽

Homogeneous Catalyst ◽

Validation Data ◽

Data Set ◽

Molecular Weights ◽

Equivalent Models ◽

Average Accuracy ◽

Better Than

ABSTRACTStandard machine-learning algorithms were used to build models capable of predicting the molecular weights of polymers generated by a homogeneous catalyst. Using descriptors calculated from only the two-dimensional structures of the ligands, the average accuracy of the models on an external validation data set was approximately 70%. Because the models show no bias and perform significantly better than equivalent models built using randomized data, we conclude that they learned useful rules and did not overfit the data.

Download Full-text

Early Prediction of Ventilator-associated Pneumonia in Critical Care Patients: a Machine Learning Model

10.21203/rs.3.rs-129953/v1 ◽

2020 ◽

Author(s):

Yingjian Liang ◽

Chengrui Zhu ◽

Cong Tian ◽

Qizhong Lin ◽

Zhiliang Li ◽

...

Keyword(s):

Machine Learning ◽

Prediction Model ◽

Sensitivity And Specificity ◽

External Validation ◽

Model Performance ◽

Learning Model ◽

Ventilator Associated Pneumonia ◽

Data Set ◽

Machine Learning Model ◽

Mimic Iii

Abstract Background: This study was performed to develop and validate machine learning models for the early detection of ventilator-associated pneumonia (VAP) in patients 24 h before the diagnosis that enables VAP patients to receive early intervention and reduces the occurrence of complications.Patients and Methods: This study was based on the MIMIC-III dataset, which was a retrospective cohort. The random forest algorithm was applied to construct a base classifier, and the area under the receiver operating characteristic (ROC) curve (AUC), sensitivity and specificity of the prediction model were evaluated. Meanwhile, a Clinical Pulmonary Infection Score (CPIS)-based model (threshold value≥3) using the same training and test data set was used as the control model.Results: A total of 38,515 ventilation durations occurred in 61,532 ICU admissions. VAP occurred in 212 of these durations. We incorporated 42 VAP risk factors on admission and routinely measured vital characteristics and laboratory results. Five-fold cross-validation was performed to evaluate the model performance, and the model achieved an AUC of 84.4%±1.7% on validation, 74.3%±2.5% sensitivity and 70.7.6%±1.2% specificity 24 h before the gold standard time (at least 48 h after ventilation). Our VAP machine learning model improved the AUC of the CPIS-based model by almost 25%, and the sensitivity and specificity were also improved by almost 14% and 15%, respectively.Conclusions: We developed and internally validated an automated model of VAP prediction in the MIMIC-III cohort. The VAP prediction model achieved high performance for AUC, sensitivity and specificity. and its performance was superior to that of the CPIS model. External validation and prospective interventional or outcome studies using this prediction model are envisioned as future work.

Download Full-text

Development and temporal external validation of a simple risk score tool for prediction of outcomes after severe head injury based on admission characteristics from level-1 trauma centre of India using retrospectively collected data

BMJ Open ◽

10.1136/bmjopen-2020-040778 ◽

2021 ◽

Vol 11 (1) ◽

pp. e040778

Author(s):

Vineet Kumar Kamal ◽

Ravindra Mohan Pandey ◽

Deepak Agrawal

Keyword(s):

Hospital Mortality ◽

External Validation ◽

Trauma Centre ◽

Unfavourable Outcome ◽

Motor Score ◽

Validation Data ◽

Data Set ◽

Development Data ◽

Level 1 ◽

Pupillary Reactivity

ObjectiveTo develop and validate a simple risk scores chart to estimate the probability of poor outcomes in patients with severe head injury (HI).DesignRetrospective.SettingLevel-1, government-funded trauma centre, India.ParticipantsPatients with severe HI admitted to the neurosurgery intensive care unit during 19 May 2010–31 December 2011 (n=946) for the model development and further, data from same centre with same inclusion criteria from 1 January 2012 to 31 July 2012 (n=284) for the external validation of the model.Outcome(s)In-hospital mortality and unfavourable outcome at 6 months.ResultsA total of 39.5% and 70.7% had in-hospital mortality and unfavourable outcome, respectively, in the development data set. The multivariable logistic regression analysis of routinely collected admission characteristics revealed that for in-hospital mortality, age (51–60, >60 years), motor score (1, 2, 4), pupillary reactivity (none), presence of hypotension, basal cistern effaced, traumatic subarachnoid haemorrhage/intraventricular haematoma and for unfavourable outcome, age (41–50, 51–60, >60 years), motor score (1–4), pupillary reactivity (none, one), unequal limb movement, presence of hypotension were the independent predictors as its 95% confidence interval (CI) of odds ratio (OR)_did not contain one. The discriminative ability (area under the receiver operating characteristic curve (95% CI)) of the score chart for in-hospital mortality and 6 months outcome was excellent in the development data set (0.890 (0.867 to 912) and 0.894 (0.869 to 0.918), respectively), internal validation data set using bootstrap resampling method (0.889 (0.867 to 909) and 0.893 (0.867 to 0.915), respectively) and external validation data set (0.871 (0.825 to 916) and 0.887 (0.842 to 0.932), respectively). Calibration showed good agreement between observed outcome rates and predicted risks in development and external validation data set (p>0.05).ConclusionFor clinical decision making, we can use of these score charts in predicting outcomes in new patients with severe HI in India and similar settings.

Download Full-text

Glioblastoma and primary central nervous system lymphoma: differentiation using MRI derived first-order texture analysis – a machine learning study

The Neuroradiology Journal ◽

10.1177/1971400921998979 ◽

2021 ◽

pp. 197140092199897

Author(s):

Sarv Priya ◽

Caitlin Ward ◽

Thomas Locke ◽

Neetu Soni ◽

Ravishankar Pillenahalli Maheshwarappa ◽

...

Keyword(s):

Machine Learning ◽

Central Nervous System ◽

Nervous System ◽

Diagnostic Performance ◽

Model Performance ◽

Area Under The Curve ◽

Central Nervous System Lymphoma ◽

First Order ◽

Single Slice ◽

Contrast Enhanced

Objectives To evaluate the diagnostic performance of multiple machine learning classifier models derived from first-order histogram texture parameters extracted from T1-weighted contrast-enhanced images in differentiating glioblastoma and primary central nervous system lymphoma. Methods Retrospective study with 97 glioblastoma and 46 primary central nervous system lymphoma patients. Thirty-six different combinations of classifier models and feature selection techniques were evaluated. Five-fold nested cross-validation was performed. Model performance was assessed for whole tumour and largest single slice using receiver operating characteristic curve. Results The cross-validated model performance was relatively similar for the top performing models for both whole tumour and largest single slice (area under the curve 0.909–0.924). However, there was a considerable difference between the worst performing model (logistic regression with full feature set, area under the curve 0.737) and the highest performing model for whole tumour (least absolute shrinkage and selection operator model with correlation filter, area under the curve 0.924). For single slice, the multilayer perceptron model with correlation filter had the highest performance (area under the curve 0.914). No significant difference was seen between the diagnostic performance of the top performing model for both whole tumour and largest single slice. Conclusions T1 contrast-enhanced derived first-order texture analysis can differentiate between glioblastoma and primary central nervous system lymphoma with good diagnostic performance. The machine learning performance can vary significantly depending on the model and feature selection methods. Largest single slice and whole tumour analysis show comparable diagnostic performance.

Download Full-text

INeo-Epp: A novel T-cell HLA class-I immunogenicity or neoantigenic epitope prediction method based on sequence related amino acid features

10.1101/697011 ◽

2019 ◽

Author(s):

Guangzhi Wang ◽

Huihui Wan ◽

Xingxing Jian ◽

Yuyu Li ◽

Jian Ouyang ◽

...

Keyword(s):

Amino Acid ◽

T Cell ◽

Antigen Processing ◽

External Validation ◽

Cell Epitope ◽

Epitope Prediction ◽

Validation Data ◽

Data Set ◽

Immunogenic Peptide ◽

Related Amino Acid

AbstractIn silico T-cell epitope prediction plays an important role in immunization experimental design and vaccine preparation. Currently, most epitope prediction research focuses on peptide processing and presentation, e.g. proteasomal cleavage, transporter associated with antigen processing (TAP) and major histocompatibility complex (MHC) combination. To date, however, the mechanism for immunogenicity of epitopes remains unclear. It is generally agreed upon that T-cell immunogenicity may be influenced by the foreignness, accessibility, molecular weight, molecular structure, molecular conformation, chemical properties and physical properties of target peptides to different degrees. In this work, we tried to combine these factors. Firstly, we collected significant experimental HLA-I T-cell immunogenic peptide data, as well as the potential immunogenic amino acid properties. Several characteristics were extracted, including amino acid physicochemical property of epitope sequence, peptide entropy, eluted ligand likelihood percentile rank (EL rank(%)) score and frequency score for immunogenic peptide. Subsequently, a random forest classifier for T cell immunogenic HLA-I presenting antigen epitopes and neoantigens was constructed. The classification results for the antigen epitopes outperformed the previous research (the optimal AUC=0.81, external validation data set AUC=0.77). As mutational epitopes generated by the coding region contain only the alterations of one or two amino acids, we assume that these characteristics might also be applied to the classification of the endogenic mutational neoepitopes also called ‘neoantigens’. Based on mutation information and sequence related amino acid characteristics, a prediction model of neoantigen was established as well (the optimal AUC=0.78). Further, an easy-to-use web-based tool ‘INeo-Epp’ was developed (available at http://www.biostatistics.online/INeo-Epp/neoantigen.php)for the prediction of human immunogenic antigen epitopes and neoantigen epitopes.

Download Full-text

Discovery of Highly Polymorphic Organic Materials: A New Machine Learning Approach

10.26434/chemrxiv.9524219 ◽

2019 ◽

Author(s):

Zied Hosni ◽

Annalisa Riccardi ◽

Stephanie Yerdelen ◽

Alan R. G. Martin ◽

Deborah Bowering ◽

...

Keyword(s):

Machine Learning ◽

Structure Prediction ◽

External Validation ◽

New Drugs ◽

Training Dataset ◽

Validation Dataset ◽

Machine Learning Classification ◽

Novel Approach ◽

Physical Form ◽

Machine Learning Approach

<div><div><p>Polymorphism is the capacity of a molecule to adopt different conformations or molecular packing arrangements in the solid state. This is a key property to control during pharmaceutical manufacturing because it can impact a range of properties including stability and solubility. In this study, a novel approach based on machine learning classification methods is used to predict the likelihood for an organic compound to crystallise in multiple forms. A training dataset of drug-like molecules was curated from the Cambridge Structural Database (CSD) and filtered according to entries in the Drug Bank database. The number of separate forms in the CSD for each molecule was recorded. A metaclassifier was trained using this dataset to predict the expected number of crystalline forms from the compound descriptors. This approach was used to estimate the number of crystallographic forms for an external validation dataset. These results suggest this novel methodology can be used to predict the extent of polymorphism of new drugs or not-yet experimentally screened molecules. This promising method complements expensive ab initio methods for crystal structure prediction and as integral to experimental physical form screening, may identify systems that with unexplored potential.</p> </div> </div>

Download Full-text

Imaging-based prediction of histological clot composition from admission CT imaging

Journal of NeuroInterventional Surgery ◽

10.1136/neurintsurg-2020-016774 ◽

2021 ◽

pp. neurintsurg-2020-016774

Author(s):

Uta Hanning ◽

Peter B Sporns ◽

Marios N Psychogios ◽

Astrid Jeibmann ◽

Jens Minnerup ◽

...

Keyword(s):

Machine Learning ◽

Vascular Tissue ◽

Large Diameter ◽

Small Diameter ◽

Area Under The Curve ◽

Quantitative Information ◽

Youden Index ◽

Quantitative Image ◽

Characteristic Area ◽

Contrast Ct

BackgroundThrombus composition has been shown to be a major determinant of recanalization success and occurrence of complications in mechanical thrombectomy. The most important parameters of thrombus behavior during interventional procedures are relative fractions of fibrin and red blood cells (RBCs). We hypothesized that quantitative information from admission non-contrast CT (NCCT) and CT angiography (CTA) can be used for machine learning based prediction of thrombus composition.MethodsThe analysis included 112 patients with occlusion of the carotid-T or middle cerebral artery who underwent thrombectomy. Thrombi samples were histologically analyzed and fractions of fibrin and RBCs were determined. Thrombi were semi-automatically delineated in CTA scans and NCCT scans were registered to the same space. Two regions of interest (ROIs) were defined for each thrombus: small-diameter ROIs capture vessel walls and thrombi, large-diameter ROIs reflect peri-vascular tissue responses. 4844 quantitative image markers were extracted and evaluated for their ability to predict thrombus composition using random forest algorithms in a nested fivefold cross validation.ResultsTest set receiver operating characteristic area under the curve was 0.83 (95% CI 0.80 to 0.87) for differentiating RBC-rich thrombi and 0.84 (95% CI 0.80 to 0.87) for differentiating fibrin-rich thrombi. At maximum Youden-Index, RBC-rich thrombi were identified at 77% sensitivity and 74% specificity; for fibrin-rich thrombi the classifier reached 81% sensitivity at 73% specificity.ConclusionsMachine learning based analysis of admission imaging allows for prediction of clot composition. Perspectively, such an approach could allow selection of clot-specific devices and retrieval procedures for personalized thrombectomy strategies.

Download Full-text

A comprehensive review and performance evaluation of bioinformatics tools for HLA class I peptide-binding prediction

Briefings in Bioinformatics ◽

10.1093/bib/bbz051 ◽

2020 ◽

Vol 21 (4) ◽

pp. 1119-1135 ◽

Cited By ~ 27

Author(s):

Shutao Mei ◽

Fuyi Li ◽

André Leier ◽

Tatiana T Marquez-Lago ◽

Kailin Giam ◽

...

Keyword(s):

Machine Learning ◽

T Cell ◽

Peptide Binding ◽

Hla Class I ◽

Machine Learning Algorithms ◽

Class I ◽

Target Cells ◽

Binding Prediction ◽

Validation Data ◽

Data Set

Abstract Human leukocyte antigen class I (HLA-I) molecules are encoded by major histocompatibility complex (MHC) class I loci in humans. The binding and interaction between HLA-I molecules and intracellular peptides derived from a variety of proteolytic mechanisms play a crucial role in subsequent T-cell recognition of target cells and the specificity of the immune response. In this context, tools that predict the likelihood for a peptide to bind to specific HLA class I allotypes are important for selecting the most promising antigenic targets for immunotherapy. In this article, we comprehensively review a variety of currently available tools for predicting the binding of peptides to a selection of HLA-I allomorphs. Specifically, we compare their calculation methods for the prediction score, employed algorithms, evaluation strategies and software functionalities. In addition, we have evaluated the prediction performance of the reviewed tools based on an independent validation data set, containing 21 101 experimentally verified ligands across 19 HLA-I allotypes. The benchmarking results show that MixMHCpred 2.0.1 achieves the best performance for predicting peptides binding to most of the HLA-I allomorphs studied, while NetMHCpan 4.0 and NetMHCcons 1.1 outperform the other machine learning-based and consensus-based tools, respectively. Importantly, it should be noted that a peptide predicted with a higher binding score for a specific HLA allotype does not necessarily imply it will be immunogenic. That said, peptide-binding predictors are still very useful in that they can help to significantly reduce the large number of epitope candidates that need to be experimentally verified. Several other factors, including susceptibility to proteasome cleavage, peptide transport into the endoplasmic reticulum and T-cell receptor repertoire, also contribute to the immunogenicity of peptide antigens, and some of them can be considered by some predictors. Therefore, integrating features derived from these additional factors together with HLA-binding properties by using machine-learning algorithms may increase the prediction accuracy of immunogenic peptides. As such, we anticipate that this review and benchmarking survey will assist researchers in selecting appropriate prediction tools that best suit their purposes and provide useful guidelines for the development of improved antigen predictors in the future.

Download Full-text