scholarly journals Violence detection explanation via semantic roles embeddings

2020 ◽  
Vol 20 (1) ◽  
Author(s):  
Enrico Mensa ◽  
Davide Colla ◽  
Marco Dalmasso ◽  
Marco Giustini ◽  
Carlo Mamo ◽  
...  
Keyword(s):  
Emergency Room ◽  
Language Processing ◽  
Detection System ◽  
Real Data ◽  
Hospital Staff ◽  
The Body ◽  
Hybrid Technique ◽  
Narrative Texts ◽  
Semantic Roles ◽  
Violence Detection

Abstract Background Emergency room reports pose specific challenges to natural language processing techniques. In this setting, violence episodes on women, elderly and children are often under-reported. Categorizing textual descriptions as containing violence-related injuries (V) vs. non-violence-related injuries (NV) is thus a relevant task to the ends of devising alerting mechanisms to track (and prevent) violence episodes. Methods We present ViDeS (so dubbed after Violence Detection System), a system to detect episodes of violence from narrative texts in emergency room reports. It employs a deep neural network for categorizing textual ER reports data, and complements such output by making explicit which elements corroborate the interpretation of the record as reporting about violence-related injuries. To these ends we designed a novel hybrid technique for filling semantic frames that employs distributed representations of terms herein, along with syntactic and semantic information. The system has been validated on real data annotated with two sorts of information: about the presence vs. absence of violence-related injuries, and about some semantic roles that can be interpreted as major cues for violent episodes, such as the agent that committed violence, the victim, the body district involved, etc.. The employed dataset contains over 150K records annotated with class (V,NV) information, and 200 records with finer-grained information on the aforementioned semantic roles. Results We used data coming from an Italian branch of the EU-Injury Database (EU-IDB) project, compiled by hospital staff. Categorization figures approach full precision and recall for negative cases and.97 precision and.94 recall on positive cases. As regards as the recognition of semantic roles, we recorded an accuracy varying from.28 to.90 according to the semantic roles involved. Moreover, the system allowed unveiling annotation errors committed by hospital staff. Conclusions Explaining systems’ results, so to make their output more comprehensible and convincing, is today necessary for AI systems. Our proposal is to combine distributed and symbolic (frame-like) representations as a possible answer to such pressing request for interpretability. Although presently focused on the medical domain, the proposed methodology is general and, in principle, it can be extended to further application areas and categorization tasks.

scholarly journals ER documents categorization and explanation

2020 ◽  
Author(s):  
Enrico Mensa ◽  
Davide Colla ◽  
Marco Dalmasso ◽  
Marco Giustini ◽  
Carlo Mamo ◽  
...  
Keyword(s):  
Natural Language Processing ◽  
Natural Language ◽  
Emergency Room ◽  
Women Elderly ◽  
Language Processing ◽  
Real Data ◽  
Hospital Staff ◽  
Hybrid Technique ◽  
Relevant Task ◽  
Processing Techniques

Abstract Background Emergency room reports are a specific kind of text, posing specific challenges to natural language processing techniques. In this setting, violence episodes on women, elderly and children are often under-reported. Categorizing textual descriptions as containing violence-related injuries vs. non-violence-related injuries, is thus a relevant task, to the ends of devising alerting mechanisms to track violence episodes. Methods We present a system to detect episodes of violence from the textual descriptions contained in emergency room reports. It employs a deep neural network for categorizing textual ER reports data. Additionally, the system complements such output by making explicit which elements corroborate the interpretation of the record as reporting about violence-related injuries. To these ends we designed a novel hybrid technique for filling semantic frames that employs distributed representations of the terms herein, along with syntactic and semantic information. Results We tested our system on a set of real data of emergency room reports, coming from an Italian branch of the EU-Injury Database (EU-IDB) project, annotated by hospital staff. Our experimentation shows that the system produces accurate categorization (of violent vs. non violent records), paired with interesting results on the explanation of such output. At times, it also allowed unveiling annotation errors committed by hospital staff. Conclusions In the last few years deep architectures and word embeddings have been compared to a tsunami hitting AI and the area concerned with natural language processing. Only at a later time we have been realizing that the stunning output of deep networks needed to be explained: our proposal, combining distributed and symbolic (frame-like) representations are a possible answer to this pressing request for interpretability. Although the present application is focused on the medical domain, the proposed methodology is general and, in principle, it can be extended to further application areas and categorization tasks.

Advanced Brain Tumor Detection System

2020 ◽  
Vol 3 (2) ◽  
pp. 31-45
Author(s):  
Monica S. Kumar ◽  
Swathi K. Bhat ◽  
Vaishali R. Thakare
Keyword(s):  
Neural Network ◽  
Brain Tumor ◽  
Language Processing ◽  
Detection System ◽  
The Body ◽  
Tumor Segmentation ◽  
Cancer Type ◽  
Brain Tumor Segmentation ◽  
Secondary Tumor ◽  
Tumor Region

Brain tumor segmentation and detection is one of the most critical parts in the field of medical regions. Tumor is a cancer type that can be visible in any part of the body in case of primary and secondary tumor. The different type of brain tumor is glioma, benign, malignant, meningioma. This research helps in retrieving the tumor region in the brain with the help of 2D MRI images. The system predicts using MATLAB which is a programming platform and analyze the tumor from different method like canny edge, Otsu's binary, fuzzy c-means (FCM), and k-means clustering to improve the borders using the pixel technique. Using convolution neural network (CNN), neural network, and natural language processing, the system detects brain tumor based on the pre-processing and post-processing feature. Moreover, the authors figure out which tumor affected is the most important feature to protect the lifespan in the initial stages. Finally, it acknowledges the result in the mail format to the doctor or patient.

scholarly journals Sensors for the detection of ammonia as a potential biomarker for health screening

2021 ◽  
Vol 11 (1) ◽  
Author(s):  
Peter P. Ricci ◽  
Otto J. Gregory
Keyword(s):  
Vapor Phase ◽  
Detection System ◽  
Health Screening ◽  
The Body ◽  
Peptic Ulcers ◽  
Non Invasive ◽  
Sensing Platform ◽  
Potential Biomarker ◽  
Screening Protocol ◽  
H Pylori

AbstractThe presence of ammonia within the body has long been linked to complications stemming from the liver, kidneys, and stomach. These complications can be the result of serious conditions such as chronic kidney disease (CKD), peptic ulcers, and recently COVID-19. Limited liver and kidney function leads to increased blood urea nitrogen (BUN) within the body resulting in elevated levels of ammonia in the mouth, nose, and skin. Similarly, peptic ulcers, commonly from H. pylori, result in ammonia production from urea within the stomach. The presence of these biomarkers enables a potential screening protocol to be considered for frequent, non-invasive monitoring of these conditions. Unfortunately, detection of ammonia in these mediums is rather challenging due to relatively small concentrations and an abundance of interferents. Currently, there are no options available for non-invasive screening of these conditions continuously and in real-time. Here we demonstrate the selective detection of ammonia using a vapor phase thermodynamic sensing platform capable of being employed as part of a health screening protocol. The results show that our detection system has the remarkable ability to selectively detect trace levels of ammonia in the vapor phase using a single catalyst. Additionally, detection was demonstrated in the presence of interferents such as carbon dioxide (CO2) and acetone common in human breath. These results show that our thermodynamic sensors are well suited to selectively detect ammonia at levels that could potentially be useful for health screening applications.

Small angle reconstruction of plasma luminosity for spherical tokamak

2020 ◽  
Author(s):  
А.Н. Баженов ◽  
П.А. Затылкин
Keyword(s):  
Power Plants ◽  
Essential Feature ◽  
Real Data ◽  
The Body ◽  
Magnetic Sensors ◽  
Test Problems ◽  
General View ◽  
Central Projection ◽  
The Matrix ◽  
Matrix Detector

Публикация посвящена применению методов вычислительной геометрии, интервального анализа и линейного программирования к задачам физики управляемого термоядерного синтеза. Рассмотрены геометрические аспекты проблемы, получены проекции светимостей различных объемов сферического токамака на плоскость матричного детектора, изучены изображения предполагаемых макроскопических структур и микроскопических включений. Для набора модельных распределений светимости объема токамака поставлена задача восстановления сигнала. Решение получено с использованием задач линейного программирования. The problems of reconstruction of plasma luminosity are important for physics and technology of power plants-tokamaks. The Globus-M research tokamak obtained a large amount of data using a matrix detector in pinhole camera geometry. From the mathematical point of view, finding the luminosity for different regions of the plasma volume according to the matrix detector is an inverse problem related to the field of integral geometry. An essential feature of the particular task is the use of a single fixed camera with a small viewing angle. In this regard, application of methods of harmonic analysis of data is not enough. The paper investigates the geometric aspects of the problem. In the general view, a threedimensional object is projected onto a two-dimensional plane through a diaphragm. Under the assumption of azimuthal symmetry, there is a central projection of the luminosity of the body of rotation onto a flat matrix detector. The initial information for the calculation is the plasma boundary obtained from magnetic sensors. There is no reliable information about the internal structure of the plasma, so its division into regions of the equal luminosity is not unambiguous. The paper presents an algorithm for finding the projections of the luminosity of plasma volumes on the plane of the matrix detector. A set of model direct problems for the construction of algorithms for their recognition according to the detector data was investigated. Images of supposed macroscopic structures and microscopic inclusions were obtained. The methodological basis of the work is the use of interval analysis methods for solving geometric and algebraic problems. This approach allows obtaining qualitative and quantitative results that takes into account the uncertainty of the input data with the minimum amount of computational costs. Algebraic solvability is investigated in the interval formulation using response functionality. Solutions for a set of test problems are obtained, which demonstrate the availability of successful reconstruction for real data. An important result of the study is an information about the presence of uncertainties in geometric data and related calculations by obtaining results about the luminosity of the plasma by solving linear programming problems.

scholarly journals Deep learning model for fully automated breast cancer detection system from thermograms

PLoS ONE ◽  
2022 ◽  
Vol 17 (1) ◽  
pp. e0262349
Author(s):  
Esraa A. Mohamed ◽  
Essam A. Rashed ◽  
Tarek Gaber ◽  
Omar Karam
Keyword(s):  
Breast Cancer ◽  
Deep Learning ◽  
Cancer Detection ◽  
Detection System ◽  
Diagnostic Technique ◽  
Learning Model ◽  
The Body ◽  
Breast Cancer Detection ◽  
Benchmark Database ◽  
Deep Learning Model

Breast cancer is one of the most common diseases among women worldwide. It is considered one of the leading causes of death among women. Therefore, early detection is necessary to save lives. Thermography imaging is an effective diagnostic technique which is used for breast cancer detection with the help of infrared technology. In this paper, we propose a fully automatic breast cancer detection system. First, U-Net network is used to automatically extract and isolate the breast area from the rest of the body which behaves as noise during the breast cancer detection model. Second, we propose a two-class deep learning model, which is trained from scratch for the classification of normal and abnormal breast tissues from thermal images. Also, it is used to extract more characteristics from the dataset that is helpful in training the network and improve the efficiency of the classification process. The proposed system is evaluated using real data (A benchmark, database (DMR-IR)) and achieved accuracy = 99.33%, sensitivity = 100% and specificity = 98.67%. The proposed system is expected to be a helpful tool for physicians in clinical use.

scholarly journals Geostatistical Learning: Challenges and Opportunities

2021 ◽  
Vol 7 ◽  
Author(s):  
Júlio Hoffimann ◽  
Maciel Zortea ◽  
Breno de Carvalho ◽  
Bianca Zadrozny
Keyword(s):  
Statistical Learning ◽  
Spatial Correlation ◽  
Language Processing ◽  
Real Data ◽  
Model Errors ◽  
Process Data ◽  
Applied Machine Learning ◽  
Geophysical Surveys ◽  
Challenges And Opportunities ◽  
Geophysical Processes

Statistical learning theory provides the foundation to applied machine learning, and its various successful applications in computer vision, natural language processing and other scientific domains. The theory, however, does not take into account the unique challenges of performing statistical learning in geospatial settings. For instance, it is well known that model errors cannot be assumed to be independent and identically distributed in geospatial (a.k.a. regionalized) variables due to spatial correlation; and trends caused by geophysical processes lead to covariate shifts between the domain where the model was trained and the domain where it will be applied, which in turn harm the use of classical learning methodologies that rely on random samples of the data. In this work, we introduce the geostatistical (transfer) learning problem, and illustrate the challenges of learning from geospatial data by assessing widely-used methods for estimating generalization error of learning models, under covariate shift and spatial correlation. Experiments with synthetic Gaussian process data as well as with real data from geophysical surveys in New Zealand indicate that none of the methods are adequate for model selection in a geospatial context. We provide general guidelines regarding the choice of these methods in practice while new methods are being actively researched.

Anterior clinoidectomy using an extradural and intradural 2-step hybrid technique

2018 ◽  
Vol 130 (1) ◽  
pp. 238-247 ◽  
Author(s):  
Ali Tayebi Meybodi ◽  
Michael T. Lawton ◽  
Sonia Yousef ◽  
Xiaoming Guo ◽  
Jose Juan González Sánchez ◽  
...  
Keyword(s):  
Internal Carotid ◽  
Surgical Simulation ◽  
Skull Base Surgery ◽  
The Body ◽  
Hybrid Technique ◽  
Dust Deposition ◽  
Anterior Clinoid Process ◽  
Anterior Clinoidectomy ◽  
Optic Strut ◽  
Neurovascular Structures

Anterior clinoidectomy is a difficult yet essential technique in skull base surgery. Two main techniques (extradural and intradural) with multiple modifications have been proposed to increase efficiency and avoid complications. In this study, the authors sought to develop a hybrid technique based on localization of the optic strut (OS) to combine the advantages and avoid the disadvantages of both techniques.Ten cadaveric specimens were prepared for surgical simulation. After a standard pterional craniotomy, the anterior clinoid process (ACP) was resected in 2 steps. The segment anterior to the OS was resected extradurally, while the segment posterior to the OS was resected intradurally. The proposed technique was performed in 6 clinical cases to evaluate its safety and efficiency.Anterior clinoidectomy was successfully performed in all cadaveric specimens and all 6 patients by using the proposed technique. The extradural phase enabled early decompression of the optic nerve while avoiding the adjacent internal carotid artery. The OS was drilled intradurally under direct visualization of the adjacent neurovascular structures. The described landmarks were easily identifiable and applicable in the surgically treated patients. No operative complication was encountered.A proposed 2-step hybrid technique combines the advantages of the extradural and intradural techniques while avoiding their disadvantages. This technique allows reduced intradural drilling and subarachnoid bone dust deposition. Moreover, the most critical part of the clinoidectomy—that is, drilling of the OS and removal of the body of the ACP—is left for the intradural phase, when critical neurovascular structures can be directly viewed.

The Unfolding of Cognitive Effort During Sentence Processing: Pupillometric Evidence From People With and Without Aphasia

2021 ◽  
pp. 1-18
Author(s):  
Laura Roche Chapman ◽  
Brooke Hallowell
Keyword(s):  
Language Processing ◽  
Sentence Processing ◽  
Time Course ◽  
Time Windows ◽  
Cognitive Effort ◽  
Task Engagement ◽  
The Body ◽  
Group Differences ◽  
Language Deficits ◽  
Pupillary Responses

Purpose: Arousal and cognitive effort are relevant yet often overlooked components of attention during language processing. Pupillometry can be used to provide a psychophysiological index of arousal and cognitive effort. Given that much is unknown regarding the relationship between cognition and language deficits seen in people with aphasia (PWA), pupillometry may be uniquely suited to explore those relationships. The purpose of this study was to examine arousal and the time course of the allocation of cognitive effort related to sentence processing in people with and without aphasia. Method: Nineteen PWA and age- and education-matched control participants listened to relatively easy (subject-relative) and relatively difficult (object-relative) sentences and were required to answer occasional comprehension questions. Tonic and phasic pupillary responses were used to index arousal and the unfolding of cognitive effort, respectively, while sentences were processed. Group differences in tonic and phasic responses were examined. Results: Group differences were observed for both tonic and phasic responses. PWA exhibited greater overall arousal throughout the task compared with controls, as evidenced by larger tonic pupil responses. Controls exhibited more effort (greater phasic responses) for difficult compared with easy sentences; PWA did not. Group differences in phasic responses were apparent during end-of-sentence and postsentence time windows. Conclusions: Results indicate that the attentional state of PWA in this study was not consistently supportive of adequate task engagement. PWA in our sample may have relatively limited attentional capacity or may have challenges with allocating existing capacity in ways that support adequate task engagement and performance. This work adds to the body of evidence supporting the validity of pupillometric tasks for the study of aphasia and contributes to a better understanding of the nature of language deficits in aphasia. Supplemental Material https://doi.org/10.23641/asha.16959376

A Proof of Concept for Assessing Emergency Room Use with Primary Care Data and Natural Language Processing

2013 ◽  
Vol 52 (01) ◽  
pp. 33-42 ◽  
Author(s):  
M.-H. Kuo ◽  
P. Gooch ◽  
J. St-Maurice
Keyword(s):  
Mental Health ◽  
Primary Care ◽  
Natural Language Processing ◽  
Natural Language ◽  
Emergency Room ◽  
Language Processing ◽  
Free Text ◽  
Proof Of Concept ◽  
Term Extraction ◽  
Primary Care Data

SummaryObjective: The objective of this study was to undertake a proof of concept that demonstrated the use of primary care data and natural language processing and term extraction to assess emergency room use. The study extracted biopsychosocial concepts from primary care free text and related them to inappropriate emergency room use through the use of odds ratios.Methods: De-identified free text notes were extracted from a primary care clinic in Guelph, Ontario and analyzed with a software toolkit that incorporated General Architecture for Text Engineering (GATE) and MetaMap components for natural language processing and term extraction.Results: Over 10 million concepts were extracted from 13,836 patient records. Codes found in at least 1% percent of the sample were regressed against inappropriate emergency room use. 77 codes fell within the realm of biopsychosocial, were very statistically significant (p < 0.001) and had an OR > 2.0. Thematically, these codes involved mental health and pain related concepts.Conclusions: Analyzed thematically, mental health issues and pain are important themes; we have concluded that pain and mental health problems are primary drivers for inappropriate emergency room use. Age and sex were not significant. This proof of concept demonstrates the feasibly of combining natural language processing and primary care data to analyze a system use question. As a first work it supports further research and could be applied to investigate other, more complex problems.

Sign in / Sign up

Export Citation Format

Share Document