Sorichetta, A., Bird, T., Ruktanonchai, N. et al. Mapping internal connectivity through human migration in malaria endemic countries. Sci Data 3, 160066 (2016). https://doi.org/10.1038/sdata.2016.66
As human mobility increases continuously, pathogens are carried to countries with streams of people. In 2016, more than one billion people live outside their places of origin and there are more than 740 million migrants. Millions of people travel internationally and domestically every week and such flow of people will continue to increase in the following years. Therefore, for sake of granting people healthy life, it is crucial to quantify human mobility across multiple temporal and spatial scales. This article aims to investigate on the internal human migration flows in malaria epidemic countries and predict the volume of migrants in certain years; thus, health agencies could take measures accordingly to intervene the possible transmission of those contagious pathogens and diseases. Within the sustainable development goals, good health and well-being is related to this article.
This article selects malaria as an example of the epidemics and applied multiple types of data to establish a model that predicts of migrants’ mobility. The article uses several functions to estimate migration flows between administrative units and then provides tables containing the microdata of internal migration from 2005 to 2010 as well as supplementary data to improve the predictive power of the model. The article also applies Generalized Additive Modelling (GAM) to replace linear predictors and build possible non-linear relationship by incorporating the regression coefficients. To present the illustration of the final predictions, the article uses raster/grid data: illustrating estimated internal human migration flows between subnational administrative units for every malaria endemic country in Africa, Asia, Latin America and the Caribbean.
According to Amartya Sen, human development requires the removal of major sources of unfreedom. Diseases caused by contagious pathogens exert health risks to the epidemic countries, which is exactly a manifestation of unfreedom that needs to be eliminated. By establishing a predictive model for malaria epidemic countries, the scholars provide better guidelines at both national and international level and enable health agencies in these countries to design more effective malaria elimination plans.
This study, however, also has some limitations. For example, the modelling sample selected from each country is not sufficiently large; the data (i.e. 2005-2010) may be too old to generalize to temporary situations (major socioeconomic changes may have taken place); this model only estimates permanent human movements, that it to say seasonal movement or forced displacements are not considered. The predict power of this model is inevitably weakened by these factors, but it can still function as a reference for local health bureaus to formulate malaria control measures.
Ruktanonchai, N.W., Bhavnani, D., Sorichetta, A. et al. Census-derived migration data as a tool for informing malaria elimination policy. Malar J 15, 273 (2016). https://doi.org/10.1186/s12936-016-1315-5
The former article discusses the spread of malaria along with human migration and this article, in response, discusses different countries’ approaches to eliminate malaria. Although annual death caused by malaria is decreasing in recent decades and many countries are approaching parasite-elimination, malaria would not completely disappear in a short period of time and, because of human movements internationally, it might be reintroduced to countries that had once overcame it. Since this article is about disease dynamics, it is also related to the goal of health and well-being in human development and “a long and healthy life” dimension.
The scholars use mobile phone data to reflect international movements and then use logistic regression models to predict the proportions of people from certain geographical unit who moved to another unit within a period of time. Another 2 figures are presented to show the correlations between mobile phone data and the microdata of international movements. There are 5 raster data figures that show the population flow, probability of leaving, flow of infected people, relative import/export rate, and community structure of infected people. However, these scholars admit that mobile phone data don’t necessarily represent a complete picture of short-term human movement since mobile phone ownership is known to be demographically biased and the movement pattern data are income-biased. Therefore, the simplicity of this model implies its inability to capture the complex patterns of human activity, nor can it reflect the observed interpersonal heterogeneity, which is essential for enacting malaria elimination plans.
The article compares Haiti people’s movement and Mesoamerican movement and comes to conclude that people tend to migrate into more urbanized cities where population is highly dense and are inclined not to leave an urbanized city. Malaria is thus transmitted between places. Such migration is caused by the inequality between urbanized and unurbanized cities. If more resources, public facilities and more opportunities are available in smaller cities, there would be less migration triggered by urbanization. The existing state of current development is marked by Amartya Sen as unfreedom: less urbanized cities do not enjoy the same level of “instrumental freedoms” as highly urbanized cities. In addition, mobile phone ownership and income difference also represent inequalities, which not only indicate unfreedom, but also make collecting human movement data more difficult and hinder making most effective epidemic control measure due to the insufficiently predictive model.
Alegana, V.A., Wright, J., Bosco, C. et al. Malaria prevalence metrics in low- and middle-income countries: an assessment of precision in nationally-representative surveys. Malar J 16, 475 (2017). https://doi.org/10.1186/s12936-017-2127-y
The authors believe that the data collected from nationally-representative surveys in low- and middle-income countries that are used for monitoring and evaluating health metrics, remain unquantified. In order to strengthen the scientific basis for decision making in health areas, this article attempts to make a retrospective analysis for the precision of prevalence from those surveys. By showing that the required precision of national surveys can be better achieved by defining a sample size that is indicator relevant, the study calls for more health-related investments.
The researchers used a Bayesian hierarchical statistical model to estimate between- and within-cluster variability for fever and malaria prevalence, and used insecticide-treated bed nets (ITNs) in children under the age of 5 years. The main source of data comes from a national population-based representative household survey, which aims to assess population health outcomes and risk factors. Data on three indicators were collected-fever prevalence reported in the previous two weeks, rapid diagnostic test (RDT) results based on finger prick samples, and ITN use-which are available along with geographic coordinates at cluster level.
Intra-class correlation coefficient (ICC) (ρ) is used to estimate the proportion of total change between and within clusters. According to the RDT results, the use of fever prevalence or ITN is stratified according to the age of the person and the survey area in urban or rural areas. When estimating ICC, the researchers assumed that the number of individuals examined in the cluster has a binomial distribution. Then, based on ICC’s estimation, the survey design effect is modeled in the same framework based on the same number of clusters used in the original survey. There’s a table that shows the results for the 5839 clusters are split into multiple data sources across nine countries. Another figure is provided showing the spatial distribution of these clusters in the nine countries. The Bayesian mean, median and 95% credible intervals are presented along with estimates of ICC the ESS.
This study is very useful for health interventions. For example, when ICC is large, a combination of universal and population-based targeting is beneficial, while when the degree of clustering is relatively low, universal coverage for the entire at-risk population should be emphasized. Considering the applicability of this study, as long as new methods are available to improve the effectiveness and accuracy of current surveys, including the connection between household surveys and administrative-level health information system data, it will also be beneficial to perform extended analysis on the variability of clustering levels for other indicators.
Lai, S., Erbach-Schoenberg, E.z., Pezzulo, C. et al. Exploring the use of mobile phone data for national migration statistics. Palgrave Commun 5, 34 (2019). https://doi.org/10.1057/s41599-019-0242-9
This article is about justifying the promising prospect of using of mobile phone data to keep track of population number update and thus estimate migration flows. Statistics on migration is important for many human development aspects like urban planning, infrastructure development, resource allocation, public service provision, etc. As mobile phone penetration rate in low-income countries continues to increase in recent years, call detailed records (CDRs) become a pretty precise and accurate indicator in measuring the migration at multiple temporal and spatial scales, which is proved by CDRs datasets in Namibia. The researchers obtained a larger dataset from the largest network operator in Namibia which obtains more than 76% of market share and provides network spatial coverage 75% population. The place of residence of a user is estimated as the location where the user is most frequently observed and the researchers only include users that are active for more than 30 days every year in order to eliminate noise.
Three types of models to census data are applied to explore whether CDR-derived migration data can accurately replicate traditional census-derived migration statistics: CDR-based linear models, gravity-type spatial interaction models and GTSIMs. The gravity-type spatial interaction models have been extensively applied to estimate migration flows based on a range of migration-related push-pull factors like urbanization and natural disaster.
CDRs-based models and census statistics-based models are compared by using a leave-one-out-cross-validation approach to subset data and calculate goodness-of-fit indicators: root mean square error, Akaike Information Criterion and R square. Data of migration flow between region are presented in circular plot and the logarithmic relation between census derived population and CDRs derived model is also provide with a linear regression fit and p and R square value.
Although result shows that studying mobile phone data is a good way to estimate population, as mobile users are only a proportion of the population, the model might be biased, especially when considering that 11.5% of households are uneducated or have low-income that prevent them from affording a mobile phone. Despite that the researchers attempt to adjust these biases, the performances of both CDRs based linear model and CGTSIMs were not significantly improved. But CDRs data offer additional benefits in terms of updating intercensal migration numbers and understanding changing patterns of annual internal migration. Additionally, the methodologies presented are easy to implement considering the impact of heterogeneous phone ownership across regions and years.