scholarly journals Clustering of patient comorbidities within electronic medical records enables high-precision COVID-19 mortality prediction

2021 ◽  
Author(s):  
Erwann Le Lannou ◽  
Benjamin Post ◽  
Shlomi Haar ◽  
Stephen Brett ◽  
Balasundaram Kadirvelu ◽  
...  
Keyword(s):  
Language Processing ◽  
Latent Dirichlet Allocation ◽  
Clinical History ◽  
Mortality Prediction ◽  
Individual Risk ◽  
Risk Profiling ◽  
Digital Fingerprint ◽  
Disease Associations ◽  
Data Driven Approach ◽  
Digital Fingerprints

We present an explainable AI framework to predict mortality after a positive COVID-19 diagnosis based solely on data routinely collected in electronic healthcare records (EHRs) obtained prior to diagnosis. We grounded our analysis on the 1/2 Million people UK Biobank and linked NHS COVID-19 records. We developed a method to capture the complexities and large variety of clinical codes present in EHRs, and we show that these have a larger impact on risk than all other patient data but age. We use a form of clustering for natural language processing of the clinical codes, specifically, topic modelling by Latent Dirichlet Allocation (LDA), to generate a succinct digital fingerprint of a patient's full secondary care clinical history, i.e. their co-morbidities and past interventions. These digital comorbidity fingerprints offer immediately interpretable clinical descriptions that are meaningful, e.g. grouping cardiovascular disorders with common risk factors but also novel groupings that are not obvious. The comorbidity fingerprints differ in both their breadth and depth from existing observational disease associations in the COVID-19 literature. Taking this data-driven approach allows us to avoid human-induction bias and confirmation bias during the selection of what are important potential predictors of COVID-19 mortality. Together with age, these digital fingerprints are the single most important factor in our predictor. This holds the potential for improving individual risk profiling for clinical decisions and the identification of groups for public health interventions such as vaccine programmes. Combining our digital precondition fingerprints with demographic characteristics allow us to match or exceed the performance of existing state-of-the-art COVID-19 mortality predictors (EHCF) which have been developed through expert consensus. Our precondition fingerprinting and entire mortality prediction analytics pipeline is designed so as to be rapidly redeployable, e.g. for COVID-19 variants or other pre-existing diseases.

scholarly journals Clustering of patient comorbidities within electronic medical records enables high-precision COVID-19 mortality prediction

2021 ◽  
Author(s):  
Aldo Faisal ◽  
Erwann Le Lannou ◽  
Benjamin Post ◽  
Shlomi Haar ◽  
Stephen Brett ◽  
...  
Keyword(s):  
Language Processing ◽  
Latent Dirichlet Allocation ◽  
Clinical History ◽  
Mortality Prediction ◽  
Individual Risk ◽  
Risk Profiling ◽  
Digital Fingerprint ◽  
Disease Associations ◽  
Data Driven Approach ◽  
Digital Fingerprints

Abstract We present an explainable AI framework to predict mortality after a positive COVID-19 diagnosis based solely on data routinely collected in electronic healthcare records (EHRs) obtained prior to diagnosis. We grounded our analysis on the ½ Million people UK Biobank and linked NHS COVID-19 records. We developed a method to capture the complexities and large variety of clinical codes present in EHRs and we show that these have a larger impact on risk than all other patient data but age. We use a form of clustering for natural language processing of the clinical codes, specifically, topic modelling by Latent Dirichlet Allocation (LDA), to generate a succinct digital fingerprint of a patient’s full secondary care clinical history, i.e. their comorbidities and past interventions. These digital comorbidity fingerprints offer immediately interpretable clinical descriptions that are meaningful, e.g. grouping cardiovascular disorders with common risk factors but also novel groupings that are not obvious. The comorbidity fingerprints differ in both their breadth and depth from existing observational disease associations in the COVID-19 literature. Taking this data-driven approach allows us to avoid human-induction bias and confirmation bias during selection of what are important potential predictors of COVID-19 mortality. Together with age these digital fingerprints are the single most important factor in our predictor. This holds the potential for improving individual risk profiling for clinical decisions and the identification of groups for public health interventions such as vaccine programmes. Combining our digital precondition fingerprints with demographic characteristics allow us to match or exceed the performance of existing state-of-the-art COVID-19 mortality predictors (EHCF) which have been developed through expert consensus. Our precondition fingerprinting and entire mortality prediction analytics pipeline are designed so as to be rapidly redeployable, e.g. for COVID-19 variants or other pre-existing diseases.

scholarly journals DERIVATION OF DESCRIPTION FEATURES FOR ENGINEERING CHANGE REQUEST BY AID OF LATENT DIRICHLET ALLOCATION

2020 ◽  
Vol 1 ◽  
pp. 697-706
Author(s):  
M. Riesener ◽  
C. Dölle ◽  
M. Mendl-Heinisch ◽  
G. Schuh ◽  
A. Keuper
Keyword(s):  
Language Processing ◽  
Latent Dirichlet Allocation ◽  
Data Driven ◽  
Automated Classification ◽  
Engineering Change ◽  
Complex Products ◽  
Data Driven Approach ◽  
Engineering Changes ◽  
Processing Techniques ◽  
Change Requests

AbstractComplex products and shorter development cycles lead to an increasing number of engineering changes. In order to be able to process these changes more effectively and efficiently, this paper develops a description model as a first step towards a data driven approach of processing engineering change requests. The description model is systematically derived from literature using text mining and natural language processing techniques. An example of the application is given by an automated classification based on similarity calculations between new and historic engineering change requests.

Designing a Chat-Bot for College Information using Information Retrieval and Automatic Text Summarization Techniques

2020 ◽  
Vol 01 ◽  
Author(s):  
Radha Guha
Keyword(s):  
Information Retrieval ◽  
Language Processing ◽  
Latent Dirichlet Allocation ◽  
Semantic Analysis ◽  
Text Summarization ◽  
The Internet ◽  
Specific Domain ◽  
User Query ◽  
College Information ◽  
Chat Bot

Background:: In the era of information overload it is very difficult for a human reader to make sense of the vast information available in the internet quickly. Even for a specific domain like college or university website it may be difficult for a user to browse through all the links to get the relevant answers quickly. Objective:: In this scenario, design of a chat-bot which can answer questions related to college information and compare between colleges will be very useful and novel. Methods:: In this paper a novel conversational interface chat-bot application with information retrieval and text summariza-tion skill is designed and implemented. Firstly this chat-bot has a simple dialog skill when it can understand the user query intent, it responds from the stored collection of answers. Secondly for unknown queries, this chat-bot can search the internet and then perform text summarization using advanced techniques of natural language processing (NLP) and text mining (TM). Results:: The advancement of NLP capability of information retrieval and text summarization using machine learning tech-niques of Latent Semantic Analysis(LSI), Latent Dirichlet Allocation (LDA), Word2Vec, Global Vector (GloVe) and Tex-tRank are reviewed and compared in this paper first before implementing them for the chat-bot design. This chat-bot im-proves user experience tremendously by getting answers to specific queries concisely which takes less time than to read the entire document. Students, parents and faculty can get the answers for variety of information like admission criteria, fees, course offerings, notice board, attendance, grades, placements, faculty profile, research papers and patents etc. more effi-ciently. Conclusion:: The purpose of this paper was to follow the advancement in NLP technologies and implement them in a novel application.

A guided latent Dirichlet allocation approach to investigate real-time latent topics of Twitter data during Hurricane Laura

2021 ◽  
pp. 016555152110077
Author(s):  
Sulong Zhou ◽  
Pengyu Kan ◽  
Qunying Huang ◽  
Janet Silbernagel
Keyword(s):  
Social Media ◽  
Real Time ◽  
Language Processing ◽  
Disaster Response ◽  
Domain Knowledge ◽  
Latent Dirichlet Allocation ◽  
Situational Awareness ◽  
High Performing ◽  
Latent Topics ◽  
Dirichlet Allocation

Natural disasters cause significant damage, casualties and economical losses. Twitter has been used to support prompt disaster response and management because people tend to communicate and spread information on public social media platforms during disaster events. To retrieve real-time situational awareness (SA) information from tweets, the most effective way to mine text is using natural language processing (NLP). Among the advanced NLP models, the supervised approach can classify tweets into different categories to gain insight and leverage useful SA information from social media data. However, high-performing supervised models require domain knowledge to specify categories and involve costly labelling tasks. This research proposes a guided latent Dirichlet allocation (LDA) workflow to investigate temporal latent topics from tweets during a recent disaster event, the 2020 Hurricane Laura. With integration of prior knowledge, a coherence model, LDA topics visualisation and validation from official reports, our guided approach reveals that most tweets contain several latent topics during the 10-day period of Hurricane Laura. This result indicates that state-of-the-art supervised models have not fully utilised tweet information because they only assign each tweet a single label. In contrast, our model can not only identify emerging topics during different disaster events but also provides multilabel references to the classification schema. In addition, our results can help to quickly identify and extract SA information to responders, stakeholders and the general public so that they can adopt timely responsive strategies and wisely allocate resource during Hurricane events.

scholarly journals Identifying Medication-related Intents from a Bidirectional Text Messaging Platform for Hypertension Management: An Unsupervised Learning Approach

2021 ◽  
Author(s):  
Anahita Davoudi ◽  
Natalie Lee ◽  
Thaibinh Luong ◽  
Timothy Delaney ◽  
Elizabeth Asch ◽  
...  
Keyword(s):  
Blood Pressure ◽  
Unsupervised Learning ◽  
Language Processing ◽  
Text Messaging ◽  
Latent Dirichlet Allocation ◽  
Clinical Care ◽  
Hypertension Management ◽  
Free Text ◽  
Significant Heterogeneity ◽  
Text Data

Background: Free-text communication between patients and providers is playing an increasing role in chronic disease management, through platforms varying from traditional healthcare portals to more novel mobile messaging applications. These text data are rich resources for clinical and research purposes, but their sheer volume render them difficult to manage. Even automated approaches such as natural language processing require labor-intensive manual classification for developing training datasets, which is a rate-limiting step. Automated approaches to organizing free-text data are necessary to facilitate the use of free-text communication for clinical care and research. Objective: We applied unsupervised learning approaches to 1) understand the types of topics discussed and 2) to learn medication-related intents from messages sent between patients and providers through a bi-directional text messaging system for managing participant blood pressure. Methods: This study was a secondary analysis of de-identified messages from a remote mobile text-based employee hypertension management program at an academic institution. In experiment 1, we trained a Latent Dirichlet Allocation (LDA) model for each message type (inbound-patient and outbound-provider) and identified the distribution of major topics and significant topics (probability >0.20) across message types. In experiment 2, we annotated all medication-related messages with a single medication intent. Then, we trained a second LDA model (medLDA) to assess how well the unsupervised method could identify more fine-grained medication intents. We encoded each medication message with n-grams (n-1-3 words) using spaCy, clinical named entities using STANZA, and medication categories using MedEx, and then applied Chi-square feature selection to learn the most informative features associated with each medication intent. Results: A total of 253 participants and 5 providers engaged in the program generating 12,131 total messages: 47% patient messages and 53% provider messages. Most patient messages correspond to blood pressure (BP) reporting, BP encouragement, and appointment scheduling. In contrast, most provider messages correspond to BP reporting, medication adherence, and confirmatory statements. In experiment 1, for both patient and provider messages, most messages contained 1 topic and few with more than 3 topics identified using LDA. However, manual review of some messages within topics revealed significant heterogeneity even within single-topic messages as identified by LDA. In experiment 2, among the 534 medication messages annotated with a single medication intent, most of the 282 patient medication messages referred to medication request (48%; n=134) and medication taking (28%; n=79); most of the 252 provider medication messages referred to medication question (69%; n=173). Although medLDA could identify a majority intent within each topic, the model could not distinguish medication intents with low prevalence within either patient or provider messages. Richer feature engineering identified informative lexical-semantic patterns associated with each medication intent class. Conclusion: LDA can be an effective method for generating subgroups of messages with similar term usage and facilitate the review of topics to inform annotations. However, few training cases and shared vocabulary between intents precludes the use of LDA for fully automated deep medication intent classification.

scholarly journals Application of Latent Dirichlet Allocation (LDA) for clustering financial tweets

2021 ◽  
Vol 297 ◽  
pp. 01071
Author(s):  
Sifi Fatima-Zahrae ◽  
Sabbar Wafae ◽  
El Mzabi Amal
Keyword(s):  
Language Processing ◽  
Latent Dirichlet Allocation ◽  
Sentiment Classification ◽  
Research Areas ◽  
Preprocessing Method ◽  
Long Time ◽  
Standard Text ◽  
The Given ◽  
Dirichlet Allocation

Sentiment classification is one of the hottest research areas among the Natural Language Processing (NLP) topics. While it aims to detect sentiment polarity and classification of the given opinion, requires a large number of aspect extractions. However, extracting aspect takes human effort and long time. To reduce this, Latent Dirichlet Allocation (LDA) method have come out recently to deal with this issue.In this paper, an efficient preprocessing method for sentiment classification is presented and will be used for analyzing user’s comments on Twitter social network. For this purpose, different text preprocessing techniques have been used on the dataset to achieve an acceptable standard text. Latent Dirichlet Allocation has been applied on the obtained data after this fast and accurate preprocessing phase. The implementation of different sentiment analysis methods and the results of these implementations have been compared and evaluated. The experimental results show that the combined uses of the preprocessing method of this paper and Latent Dirichlet Allocation have an acceptable results compared to other basic methods.

Understanding Language Features in Loneliness Forum Posts that Elicit Comments (Preprint)

2021 ◽  
Author(s):  
Anietie Andy
Keyword(s):  
Language Processing ◽  
Latent Dirichlet Allocation ◽  
Well Being ◽  
Time Of Day ◽  
Online Forums ◽  
Cohen’S D ◽  
Increased Risk ◽  
Social Media Platforms ◽  
Cohen's D ◽  
To Receive

BACKGROUND Loneliness is a threat to the well-being of individuals and in older adults is associated with increased risk of early mortality. Studies have shown that some individuals seek support around loneliness on online forums/social media platforms. A common challenge in online forums is that some posts do not receive comments. In some non-health related forums, posts not receiving comments may not be a serious concern, however, in an online health forum such as those focused on discussions around loneliness, posts not receiving comments could translate to individuals seeking support around loneliness not receiving adequate support. OBJECTIVE The aim of this work is to analyze posts published on an active online forum focused on discussions around loneliness (loneliness forum) to determine the language features associated with posts that elicit comments from members of the forum. METHODS For the analysis in this work, 15,012 posts published on an online loneliness forum by 9,956 users were analyzed. Of these posts, 6,450 received five or more comments, 13,221 received one or more comments, and 1,791 received no comments. Using the natural language processing method, latent dirichlet allocation (LDA) and a psycholinguistic dictionary, Linguistics Inquiry and Word Count (LIWC), the language features expressed in posts that elicit comments from members of the forum were determined. RESULTS The findings from this work show that posts related to topics themes on relationships (Cohen’s D = 0.319) and the use of negation words (Cohen’s D = 0.149) tend to receive one or more comments. Also, posts associated with LIWC categories on first person singular pronouns (Cohen’s D = 0.264) tend to elicit one or more comments. Posts on topic themes related to spending time around holidays/birthdays/year/time of day or week (Cohen’s D = 0.79) and affection relative to relationships (Cohen’s D = 0.102) tend to receive five or more comments. CONCLUSIONS This work identifies language features expressed in loneliness forum posts that elicit comments. The findings from this work can provide members of online loneliness forums tips on how to write posts that potentially elicit comments from members of the forum.

Big Data and Natural Language Processing for Analysing Railway Safety

2018 ◽  
pp. 240-267
Author(s):  
Kanza Noor Syeda ◽  
Syed Noorulhassan Shirazi ◽  
Syed Asad Ali Naqvi ◽  
Howard J Parkinson ◽  
Gary Bamford
Keyword(s):  
Big Data ◽  
Natural Language Processing ◽  
Natural Language ◽  
Language Processing ◽  
Machine Intelligence ◽  
Data Availability ◽  
Accident Data ◽  
Data Driven Approach ◽  
Advanced Analytics ◽  
The Uk

Due to modern powerful computing and the explosion in data availability and advanced analytics, there should be opportunities to use a Big Data approach to proactively identify high risk scenarios on the railway. In this chapter, we comprehend the need for developing machine intelligence to identify heightened risk on the railway. In doing so, we have explained a potential for a new data driven approach in the railway, we then focus the rest of the chapter on Natural Language Processing (NLP) and its potential for analysing accident data. We review and analyse investigation reports of railway accidents in the UK, published by the Rail Accident Investigation Branch (RAIB), aiming to reveal the presence of entities which are informative of causes and failures such as human, technical and external. We give an overview of a framework based on NLP and machine learning to analyse the raw text from RAIB reports which would assist the risk and incident analysis experts to study causal relationship between causes and failures towards the overall safety in the rail industry.

Open-Ended Questions

2020 ◽  
pp. 202-218
Author(s):  
Subhadra Dutta ◽  
Eric M. O’Rourke
Keyword(s):  
Machine Learning ◽  
Language Processing ◽  
Latent Dirichlet Allocation ◽  
Semantic Analysis ◽  
Written Language ◽  
Text Data ◽  
Employee Survey ◽  
Trade Offs ◽  
Word Relatedness ◽  
Survey Responses

Natural language processing (NLP) is the field of decoding human written language. This chapter responds to the growing interest in using machine learning–based NLP approaches for analyzing open-ended employee survey responses. These techniques address scalability and the ability to provide real-time insights to make qualitative data collection equally or more desirable in organizations. The chapter walks through the evolution of text analytics in industrial–organizational psychology and discusses relevant supervised and unsupervised machine learning NLP methods for survey text data, such as latent Dirichlet allocation, latent semantic analysis, sentiment analysis, word relatedness methods, and so on. The chapter also lays out preprocessing techniques and the trade-offs of growing NLP capabilities internally versus externally, points the readers to available resources, and ends with discussing implications and future directions of these approaches.

Sign in / Sign up

Export Citation Format

Share Document