Katherine Battle’s scientific contributions

What is this page?


This page lists works of an author who doesn't have a ResearchGate profile or hasn't added the works to their profile yet. It is automatically generated from public (personal) data to further our legitimate goal of comprehensive and accurate scientific recordkeeping. If you are this author and want this page removed, please let us know.

Publications (8)


Description of the model assumptions in the four scenarios
The maximum walking time is fixed to 60 minutes. The maximum number of inhabitants per CHW is varied across the four scenarios. In scenario A, the entire territory is covered by CHWs, with a maximum population of 1000 per CHW in rural areas, 2500 in urban areas and 4000 in the metropolitan area. In Scenario B, only areas situated at more than a 30 minutes’ walk from a community health centre (CCS) are covered by CHWs, with a maximum population of 2500 per CHW in urban and metropolitan areas and 1000 in rural areas. In scenarios C and C2, the entire territory is covered by CHWs but the maximum population thresholds depend on the distance to the nearest CCS. In scenario C, less than a 60 minutes’ walk from a CCS, 4000 people are assigned to each CHW and more than a 60 minutes’ walk from a CCS, the maximum populations is 2500 per CHW in urban and metropolitan areas and 1000 in rural areas. Scenario C2 is similar to scenario C, except that the maximum population is 1000 in rural areas, whatever the distance to the closest CCS, and the 4000 threshold within a 60 minutes’ walk from a CCS is applied only for urban areas.
Comparison of the placement scenarios in the Grande-Anse department
1. The four CHW placement scenarios (A, B, C and C2). CHW positions are indicated with blue dots, community health centres (CCS) are indicated with red triangles. The colored surface indicates the predicted walking time to the closest CCS using the methodology by Weiss and colleagues [23, 24]: difficult-to-reach areas, located more than 60 minutes walk from the nearest CCS, are shown in orange/red and areas with easier access (less than 60 minutes walk) are shown in blue. 2. For interpretability: prediction of population density in 2020 per square kilometre [27, 29]. 3. For interpretability: walking time friction surface by Weiss et al. is shown as the time required to cross 1km [23]. The shapefile from the Centre National de l’Information Géo-Spatiale (CNIGS) was used [30] (available at https://data.humdata.org/dataset/hti-polbndl-adm1-cnigs-zip).
Comparison of the placement scenarios at the national level
1. Total number of CHW required in each scenario and proportion of CHWs affected to urban, rural and metropolitan areas. 2. Actual average number of inhabitants assigned per CHW and per scenario; the error bars represent the 5% and 95% quantiles of the distribution over all CHWs. In this panel, the metropolitan area is only defined for scenario A; in scenario B, the travel time defining areas close to a community health centre (CCS) is 30 minutes; in scenarios C and C2, it is 60 minutes (cf. Fig 1).
Comparison by department of the current number of CHWs according to the SPA survey [31] and the CHW mapping (“Cartography”, [13]), and the number of CHWs suggested in scenarios A, B, C and C2
Analysis of scenario C, where CHWs are positioned in the whole territory, accounting for the position of CCSs
1. Number of CHWs per section communale, according to scenario C. 2. Difference by section communale between the current number of CHWs in the SPA survey and the suggested number of CHWs under the scenario C. Negative values (signified in red) indicate a deficit, positive values (in blue) signify a surplus. The shapefile from the Centre National de l’Information Géo-Spatiale (CNIGS) was used [30] (available at https://data.humdata.org/dataset/hti-polbndl-adm1-cnigs-zip).
Improving access to care and community health in Haiti with optimized community health worker placement
  • Article
  • Full-text available

May 2022

·

95 Reads

·

5 Citations

·

Andrew Sunil Rajkumar

·

Paul Auxila

·

[...]

·

The national deployment of polyvalent community health workers (CHWs) is a constitutive part of the strategy initiated by the Ministry of Health to accelerate efforts towards universal health coverage in Haiti. Its implementation requires the planning of future recruitment and deployment activities for which mathematical modelling tools can provide useful support by exploring optimised placement scenarios based on access to care and population distribution. We combined existing gridded estimates of population and travel times with optimisation methods to derive theoretical CHW geographical placement scenarios including constraints on walking time and the number of people served per CHW. Four national-scale scenarios that align with total numbers of existing CHWs and that ensure that the walking time for each CHW does not exceed a predefined threshold are compared. The first scenario accounts for population distribution in rural and urban areas only, while the other three also incorporate in different ways the proximity of existing health centres. Comparing these scenarios to the current distribution, insufficient number of CHWs is systematically identified in several departments and gaps in access to health care are identified within all departments. These results highlight current suboptimal distribution of CHWs and emphasize the need to consider an optimal (re-)allocation.

Download

Beyond national indicators: adapting the Demographic and Health Surveys’ sampling strategies and questions to better inform subnational malaria intervention policy

March 2021

·

80 Reads

·

17 Citations

Malaria Journal

In malaria-endemic countries, prioritizing intervention deployment to areas that need the most attention is crucial to ensure continued progress. Global and national policy makers increasingly rely on epidemiological data and mathematical modelling to help optimize health decisions at the sub-national level. The Demographic and Health Surveys (DHS) Program is a critical data source for understanding subnational malaria prevalence and intervention coverage, which are used for parameterizing country-specific models of malaria transmission. However, data to estimate indicators at finer resolutions are limited, and surveys questions have a narrow scope. Examples from the Nigeria DHS are used to highlight gaps in the current survey design. Proposals are then made for additional questions and expansions to the DHS and Malaria Indicator Survey sampling strategy that would advance the data analyses and modelled estimates that inform national policy recommendations. Collaboration between the DHS Program, national malaria control programmes, the malaria modelling community, and funders is needed to address the highlighted data challenges.


Maps and Metrics of Insecticide-Treated Net Coverage in Africa: Access, Use, and Nets-Per-Capita, 2000-2020

February 2021

·

171 Reads

Insecticide-treated nets (ITNs) are one of the most widespread and impactful malaria interventions in Africa, yet a spatially-resolved time series of ITN coverage has never been published. Using data from multiple sources, we generate high-resolution maps of ITN access, use, and nets-per-capita annually from 2000 to 2020 across the 40 highest-burden African countries. Our findings support several existing hypotheses: that use is high among those with access, that nets are discarded more quickly than official policy presumes, and that effectively distributing nets grows more difficult as coverage increases. These results can inform both policy decisions and downstream malaria analyses.


Association between the proportion of Plasmodium falciparum and Plasmodium vivax infections detected by passive surveillance and the magnitude of the asymptomatic reservoir in the community: a pooled analysis of paired health facility and community data

April 2020

·

231 Reads

·

17 Citations

The Lancet Infectious Diseases

Background Passively collected malaria case data are the foundation for public health decision making. However, because of population-level immunity, infections might not always be sufficiently symptomatic to prompt individuals to seek care. Understanding the proportion of all Plasmodium spp infections expected to be detected by the health system becomes particularly paramount in elimination settings. The aim of this study was to determine the association between the proportion of infections detected and transmission intensity for Plasmodium falciparum and Plasmodium vivax in several global endemic settings. Methods The proportion of infections detected in routine malaria data, P(Detect), was derived from paired household cross-sectional survey and routinely collected malaria data within health facilities. P(Detect) was estimated using a Bayesian model in 431 clusters spanning the Americas, Africa, and Asia. The association between P(Detect) and malaria prevalence was assessed using log-linear regression models. Changes in P(Detect) over time were evaluated using data from 13 timepoints over 2 years from The Gambia. Findings The median estimated P(Detect) across all clusters was 12·5% (IQR 5·3–25·0) for P falciparum and 10·1% (5·0–18·3) for P vivax and decreased as the estimated log-PCR community prevalence increased (adjusted odds ratio [OR] for P falciparum 0·63, 95% CI 0·57–0·69; adjusted OR for P vivax 0·52, 0·47–0·57). Factors associated with increasing P(Detect) included smaller catchment population size, high transmission season, improved care-seeking behaviour by infected individuals, and recent increases (within the previous year) in transmission intensity. Interpretation The proportion of all infections detected within health systems increases once transmission intensity is sufficiently low. The likely explanation for P falciparum is that reduced exposure to infection leads to lower levels of protective immunity in the population, increasing the likelihood that infected individuals will become symptomatic and seek care. These factors might also be true for P vivax but a better understanding of the transmission biology is needed to attribute likely reasons for the observed trend. In low transmission and pre-elimination settings, enhancing access to care and improvements in care-seeking behaviour of infected individuals will lead to an increased proportion of infections detected in the community and might contribute to accelerating the interruption of transmission. Funding Wellcome Trust.




Figure 1: Observed data against predictions for cross-validation hold-out samples on a square root transformed scale. a) Six-fold random cross-validation. b) Three-fold spatial cross-validation with folds indicated by colour.
Table 1 : Pearson correlations between observed and predicted values.
Figure 2: Left: Observed data for Colombia (grey for zero incidence). Right: Out-of-sample predictions for the random cross-validation, machine learning only model. For each cross-validation fold, predictions are made for the held out data which are then combined to make a single surface.
Machine learning model results and means of fitted parameters (i.e. model weights) across cross-validation folds of the machine learning predictions only model.
Model ensembles with different response variables for base and meta models: malaria disaggregation regression combining prevalence and incidence data

February 2019

·

115 Reads

·

2 Citations

Maps of infection risk are a vital tool for the elimination of malaria. Routine surveillance data of malaria case counts, often aggregated over administrative regions, is becoming more widely available and can better measure low malaria risk than prevalence surveys. However, aggregation of case counts over large, heterogeneous areas means that these data are often underpowered for learning relationships between the environment and malaria risk. A model that combines point surveys and aggregated surveillance data could have the benefits of both but must be able to account for the fact that these two data types are different malariometric units. Here, we train multiple machine learning models on point surveys and then combine the predictions from these with a geostatistical disaggregation model that uses routine surveillance data. We find that, in tests using data from Colombia and Madagascar, using a disaggregation regression model to combine predictions from machine learning models trained on point surveys improves model accuracy relative to using the environmental covariates directly.


Variational Learning on Aggregate Outputs with Gaussian Processes

May 2018

·

26 Reads

While a typical supervised learning framework assumes that the inputs and the outputs are measured at the same levels of granularity, many applications, including global mapping of disease, only have access to outputs at a much coarser level than that of the inputs. Aggregation of outputs makes generalization to new inputs much more difficult. We consider an approach to this problem based on variational learning with a model of output aggregation and Gaussian processes, where aggregation leads to intractability of the standard evidence lower bounds. We propose new bounds and tractable approximations, leading to improved prediction accuracy and scalability to large datasets, while explicitly taking uncertainty into account. We develop a framework which extends to several types of likelihoods, including the Poisson model for aggregated count data. We apply our framework to a challenging and important problem, the fine-scale spatial modelling of malaria incidence, with over 1 million observations.

Citations (4)


... Since then, the concept has had mixed fortunes, with results that have sometimes been severely criticised [3], ranging from differing understandings among experts to incomplete implementation influenced by other agendas driven by some international institutions [4]. However, on balance, investment in strong primary care has been widely described as one of the most cost-effective and equitable ways [5] of moving towards universal health coverage [3,6]. Several countries, including Burkina Faso, then adopted and began implementing PHC and tried to operationalise it through several initiatives. ...

Reference:

Factors Associated with Communities’ Satisfaction with Receiving Curative Care Administered by Community Health Workers in the Health Districts of Bousse and Boussouma in Burkina Faso, 2024
Improving access to care and community health in Haiti with optimized community health worker placement

... More recent guidance from the WHO suggests that communities can be excluded from receiving bed nets during net distribution campaigns based on current and historical data on malaria prevalence rate [35]. However, fine-scale malaria risk data at the smallest administrative levels are often unavailable for urban areas in malaria-endemic countries [36][37][38]. ...

Beyond national indicators: adapting the Demographic and Health Surveys’ sampling strategies and questions to better inform subnational malaria intervention policy

Malaria Journal

... Furthermore, there is likely ascertainment bias associated with mixed infections in areas with co-circulating parasite strains, as efforts might be biased towards P. falciparum detection [130]. This is likely to be particularly common during episodes of clinical malaria when parasitaemia of one species greatly exceeds the other, and the innate host immune response may suppress both infections. ...

Association between the proportion of Plasmodium falciparum and Plasmodium vivax infections detected by passive surveillance and the magnitude of the asymptomatic reservoir in the community: a pooled analysis of paired health facility and community data

The Lancet Infectious Diseases

... Furthermore, the differential sizes of municipalities, with some being considerably large, pose challenges in representing their climatic spatial variation with a single metric, consequently diminishing their significant impact on malaria risk. Although microscale environmental characteristics are lost when we aggregate data for a year at the municipality level (such as the differences in the land use temperature), biome-scale models are a fundamental tool because epidemiological data are frequently reported at the administrative regions [62]. In addition, the primary socio-ecological processes that lead to the rise and resurgence of zoonotic diseases may take place on a biome-scale level [63][64][65]. ...

Model ensembles with different response variables for base and meta models: malaria disaggregation regression combining prevalence and incidence data