Table 3.

An overview of modelling studies of Ebola and study designs (Research aim: Trajectory prediction)

References	Dataset/Factors	Description of Modelling Approaches	Compartments, if applicable	Assumption of population mixing, if applicable	Model fitting and calibration approach, if applicable	Sensitivity analysis, if applicable	Description of account for data bias, if applicable
Area et al. [43]	Case data from WHO [1]	Developed classical differential equations and fractional SEIR model to predict the outbreak trajectory	Susceptible–exposed–infectious–removed (SEIR)	Assumed homogeneous mixing	Fitted models with the real data and obtained parameters values using l^2 norm	NA	NA
Bellan et al. [44]	Unspecified	Compared the projections of two simple models (with and without asymptomatic infection) based on the Ebola epidemic in Liberia	NA	NA	NA	NA	Modelled scenarios that does not account for asymptomatic infections and account for the presence of asymoptomic infection
Dong et al. [45]	Case data from WHO [1]	Ebola case prediction was obtained by a SEIIRF model with the consideration of post-death and hospitalisation compartments	Susceptible–exposed–infected–recovered–hospitalised–buried (SEIIRHF)	Considered those who have died and are in the process of being buried and number of people in hospital who develop to first stage or second stage of infection	Fitted case prediction curve with real data both in total cases and death cases	NA	NA
Drake et al. [46]	Case data from WHO [1], Liberia Ministry of Health or United Nations Office for the Coordination of Humanitarian Affairs (UN-OCHA)	Developed a multi-type branching process model that incorporates key heterogeneities and time-varying parameters to mimic changing human behaviour and controls in Liberia. Focused on the numbers of new infections caused by each case considering offspring distributions	Two generations of infection in a multi-type branching process model	Accounted for subpopulation differences, including hospital treatment vs. community care, transmission at funerals, and scenario-dependent transmission risk differences during care-giving	Fitted ensemble model from plausible parameters with the case data in Liberia and also fitted outcome to infection generations in HCWs and in the community	Investigated the sensitivity of the model to parameters using LHS over a much wider range of the least-squares fit for each parameter	Assumed under-reporting by a factor of 2·5
Fast et al. [47]	Case data from the Liberian Ministry of Health and Social welfare ad local county health offices. ETU admissions data from Github. Social mobilisation data from WHO and UNICEF reports. Parameters from UNICEF reports, WHO reports [1]. Population data from Liberian Institute of Statistics and Geo-information Services	Transmission model based on contact network. Accounts for change in attitudes that underlie behaviour change by simulating the progress of the EVD epidemic with and without population behaviour change, then comparing scenarios with observed data	Susceptible-exposed-infected-hospitalised-buried-unburied fatality-recovered (SEIHFCFBR)	Contact network model considered individuals as nodes and disease-spreading contacts as edges. Hospitalisation and fatality components were included	Model was fitted to weekly cases in Lofa County by considering the fitting metric of mean absolute error	NA	The model considered only cases that sought treatment or were safely buried, but modelled all cases
White et al. [48]	Case data from WHO [1] and daily counts from the public press releases from the Sierra Leonean Ministry of Health [17]	Developed compartmental model with six compartments to describe the outbreaks in Sierra Leone	Susceptible–exposed–infectious (not yet reported)–treated (reported)–dead (unreported)–recovered (reported)–dead (reported) (SEOTRDTRTD)	Considered infected and hospitalised individuals according to their reported status	Implemented an ensemble trajectory model and generated a matrix of plausible parameter values to fit the model for the first 56 days	NA	Used 2·5 correction factor estimated by the CDC to correct for underreporting