Dataset Information

Machine-learning Prognostic Models from the 2014-16 Ebola Outbreak: Data-harmonization Challenges, Validation Strategies, and mHealth Applications.

ABSTRACT:

Background

Ebola virus disease (EVD) plagues low-resource and difficult-to-access settings. Machine learning prognostic models and mHealth tools could improve the understanding and use of evidence-based care guidelines in such settings. However, data incompleteness and lack of interoperability limit model generalizability. This study harmonizes diverse datasets from the 2014-16 EVD epidemic and generates several prognostic models incorporated into the novel Ebola Care Guidelines app that provides informed access to recommended evidence-based guidelines.

Methods

Multivariate logistic regression was applied to investigate survival outcomes in 470 patients admitted to five Ebola treatment units in Liberia and Sierra Leone at various timepoints during 2014-16. We generated a parsimonious model (viral load, age, temperature, bleeding, jaundice, dyspnea, dysphagia, and time-to-presentation) and several fallback models for when these variables are unavailable. All were externally validated against two independent datasets and compared to further models including expert observational wellness assessments. Models were incorporated into an app highlighting the signs/symptoms with the largest contribution to prognosis.

Findings

The parsimonious model approached the predictive power of observational assessments by experienced clinicians (Area-Under-the-Curve, AUC?=?0.70-0.79, accuracy?=?0.64-0.74) and maintained its performance across subcohorts with different healthcare seeking behaviors. Age and viral load contributed > 5-fold the weighting of other features and including them in a minimal model had a similar AUC, albeit at the cost of specificity.

Interpretation

Clinically guided prognostic models can recapitulate clinical expertise and be useful when such expertise is unavailable. Incorporating these models into mHealth tools may facilitate their interpretation and provide informed access to comprehensive clinical guidelines.

Funding

Howard Hughes Medical Institute, US National Institutes of Health, Bill & Melinda Gates Foundation, International Medical Corps, UK Department for International Development, and GOAL Global.

SUBMITTER: Colubri A

PROVIDER: S-EPMC6610774 | biostudies-literature | 2019 May-Jun

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

Machine-learning Prognostic Models from the 2014-16 Ebola Outbreak: Data-harmonization Challenges, Validation Strategies, and mHealth Applications.

Colubri Andres A Hartley Mary-Anne MA Siakor Matthew M Wolfman Vanessa V Felix August A Sesay Tom T Shaffer Jeffrey G JG Garry Robert F RF Grant Donald S DS Levine Adam C AC Sabeti Pardis C PC

EClinicalMedicine 20190501

<h4>Background</h4>Ebola virus disease (EVD) plagues low-resource and difficult-to-access settings. Machine learning prognostic models and mHealth tools could improve the understanding and use of evidence-based care guidelines in such settings. However, data incompleteness and lack of interoperability limit model generalizability. This study harmonizes diverse datasets from the 2014-16 EVD epidemic and generates several prognostic models incorporated into the novel <i>Ebola Care Guidelines</i> a ...[more]

PMID: 31312805

Similar Datasets

Project description:BackgroundPerceived susceptibility to a disease threat (risk perception) can influence protective behaviour. This study aims to determine how exposure to information sources, knowledge and behaviours potentially influenced risk perceptions during the 2014-2015 Ebola Virus Disease outbreak in Sierra Leone.MethodsThe study is based on three cross-sectional, national surveys (August 2014, n = 1413; October 2014, n = 2086; December 2014, n = 3540) that measured Ebola-related knowledge, attitudes, and practices in Sierra Leone. Data were pooled and composite variables were created for knowledge, misconceptions and three Ebola-specific behaviours. Risk perception was measured using a Likert-item and dichotomised into 'no risk perception' and 'some risk perception'. Exposure to five information sources was dichotomised into a binary variable for exposed and unexposed. Multilevel logistic regression models were fitted to examine various associations.ResultsExposure to new media (e.g. internet) and community-level information sources (e.g. religious leaders) were positively associated with expressing risk perception. Ebola-specific knowledge and hand washing were positively associated with expressing risk perception (Adjusted OR [AOR] 1.4, 95% Confidence Interval [CI] 1.2-1.8 and AOR 1.4, 95% CI 1.1-1.7 respectively), whereas misconceptions and avoiding burials were negatively associated with risk perception, (AOR 0.7, 95% CI 0.6-0.8 and AOR 0.8, 95% CI 06-1.0, respectively).ConclusionsOur results illustrate the complexity of how individuals perceived their Ebola acquisition risk based on the way they received information, what they knew about Ebola, and actions they took to protect themselves. Community-level information sources may help to align the public's perceived risk with their actual epidemiological risk. As part of global health security efforts, increased investments are needed for community-level engagements that allow for two-way communication during health emergencies.

Dataset Information

Machine-learning Prognostic Models from the 2014-16 Ebola Outbreak: Data-harmonization Challenges, Validation Strategies, and mHealth Applications.

Background

Methods

Findings

Interpretation

Funding

Publications

Machine-learning Prognostic Models from the 2014-16 Ebola Outbreak: Data-harmonization Challenges, Validation Strategies, and mHealth Applications.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets