RESUMO
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, AUCâ¯=â¯0.70-0.79, accuracyâ¯=â¯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.
RESUMO
INTRODUCTION: Previous studies of Ebola Virus Disease (EVD) have focused on clinical symptoms and Ebola virus (EBOV) cycle threshold (CT) values recorded at patient triage. Our study explores EVD symptoms and EBOV CT values from onset of illness to recovery or death in a diverse population of patients. METHODOLOGY/PRINCIPAL FINDINGS: We analyzed clinical care data from EBOV positive patients admitted to five Ebola treatment units in West Africa from 2014-2015. Prevalence of clinical signs/symptoms and CT values were explored using descriptive statistics. Logistic regression was used to examine their association with mortality. Survival was analyzed using Kaplan-Meier estimators from symptom onset date to death. During the first week of illness, dyspnea (OR = 2.44, 95% CI: 1.07-5.85) and tachycardia (OR = 10.22, 95% CI: 2.20-56.71) were associated with higher odds of mortality. Dyspnea (OR = 2.33, 95% CI: 1.210-4.581), bleeding (OR = 2.51, 95% CI: 1.219-5.337), and diarrhoea (OR = 2.79, 95% CI: 1.171-6.970) at any point during the illness course were associated with higher odds of mortality. Higher initial (OR = 0.85, 95% CI: 0.81-0.89) and mean (OR = 0.60, 95% CI: 0.53-0.66) CT values were associated with lower odds of mortality. CT values reached their nadir after 3-5 days of illness and then rose in both survivors and non-survivors until recovery or death. CONCLUSIONS/SIGNIFICANCE: Our study demonstrates the population prevalence of clinical signs/symptoms and EBOV CT values over time in a large, diverse cohort of patients with EVD, as well as associations between symptoms/EBOV CT values and mortality. These findings have implications on surveillance, operational planning, and clinical care for future EVD outbreaks.