Evaluation of an accelerometer-based monitor for detecting bed net use and human entry/exit using a machine learning algorithm.

BACKGROUND: Distribution of long-lasting insecticidal bed nets (LLINs) is one of the main control strategies for malaria. Improving malaria prevention programmes requires understanding usage patterns in households receiving LLINs, but there are limits to what standard cross-sectional surveys of self-reported LLIN use can provide. This study was designed to assess the performance of an accelerometer-based approach for measuring a range of LLIN use behaviours as a proof of concept for more granular LLIN-use monitoring over longer time periods. METHODS: This study was carried out under controlled conditions from May to July 2018 in Liverpool, UK. A single accelerometer was affixed to the side panel of an LLIN and participants carried out five LLIN use behaviours: (1) unfurling a net; (2) entering an unfurled net; (3) lying still as if sleeping; (4) exiting from under a net; and, (5) folding up a net. The randomForest package in R, a supervised non-linear classification algorithm, was used to train models on 20-s epochs of tagged accelerometer data. Models were compared in a validation dataset using overall accuracy, sensitivity and specificity, receiver operating curves and the area under the curve (AUC). RESULTS: The five-category model had overall accuracy of 82.9% in the validation dataset, a sensitivity of 0.681 for entering a net, 0.632 for exiting, 0.733 for net down, and 0.800 for net up. A simplified four-category model, combining entering/exiting a net into one category had accuracy of 94.8%, and increased sensitivity for net down (0.756) and net up (0.829). A further simplified three-category model, identifying sleeping, net up, and a combined net down/enter/exit category had accuracy of 96.2% (483/502), with an AUC of 0.997 for net down and 0.987 for net up. Models for detecting entering/exiting by adults were significantly more accurate than for children (87.8% vs 70.0%; p < 0.001) and had a higher AUC (p = 0.03). CONCLUSIONS: Understanding how LLINs are used is crucial for planning malaria prevention programmes. Accelerometer-based systems provide a promising new methodology for studying LLIN use. Further work exploring accelerometer placement, frequency of measurements and other machine learning approaches could make these methods even more accurate in the future.

Associating the scale-up of insecticide-treated nets and use with the decline in all-cause child mortality in the Democratic Republic of Congo from 2005 to 2014.

BACKGROUND: To reduce the malaria burden and improve the socioeconomic status of its citizens, the Democratic Republic of Congo scaled up key malaria control interventions, especially insecticide-treated nets (ITNs), between 2005 and 2014. Since then, the effects of these interventions on malaria mortality and morbidity have not been assessed. This study aimed to measure the impact of the National Malaria Control Programme's efforts and to inform future control strategies. METHODS: The authors used data from the Demographic and Health Surveys 2007 and 2013-2014 to assess trends in all-cause childhood mortality (ACCM) against trends in coverage of malaria interventions at national and subnational levels. The authors used the plausibility argument to assess the impact of the malaria control interventions and used Kaplan-Meier survival probability and Cox proportional hazard models to examine the effect of ITN ownership on child survival. Contextual factor trends affecting child survival were also considered. RESULTS: Countrywide, household ownership of at least one ITN increased, from 9% in 2007 to 70% in 2013-2014. All provinces experienced similar increases, with some greater than the national level. ITN use increased between 2007 and 2013-2014 among children under five (6% to 55%). Severe anaemia (haemoglobin concentration < 8 g/dl) prevalence among children aged 6-59 months significantly decreased, from 11% (95% confidence interval [CI] 9-13%) in 2007 to 6% (95% CI 5-7%) in 2013-2014. During the same period, ACCM declined, from 148 (95% CI 132-163) to 104 (95% CI 97-112) deaths per 1000 live births. The decline in ACCM was greater among children aged 6-23 months (relative reduction of 36%), compared to children aged 24-59 months (relative reduction of 12%). Cox regression indicated that household ownership of at least one ITN reduced the risk of mortality by 24% among children under five (risk ratio = 0.76, 95% CI 0.64-0.90). Contextual factor analysis revealed marginal improvements in socioeconomic indicators and other health interventions. CONCLUSIONS: Given the patterns of the coverage of malaria control interventions, patterns in ACCM by province, and marginal improvements in contextual factors, the authors conclude that the malaria control interventions have plausibly contributed to the decrease in ACCM in the Democratic Republic of Congo from 2005 to 2014.

Using community-based reporting of vital events to monitor child mortality: Lessons from rural Ghana.

BACKGROUND: Reducing neonatal and child mortality is a key component of the health-related sustainable development goal (SDG), but most low and middle income countries lack data to monitor child mortality on an annual basis. We tested a mortality monitoring system based on the continuous recording of pregnancies, births and deaths by trained community-based volunteers (CBV). METHODS AND FINDINGS: This project was implemented in 96 clusters located in three districts of the Northern Region of Ghana. Community-based volunteers (CBVs) were selected from these clusters and were trained in recording all pregnancies, births, and deaths among children under 5 in their catchment areas. Data collection lasted from January 2012 through September 2013. All CBVs transmitted tallies of recorded births and deaths to the Ghana Birth and deaths registry each month, except in one of the study districts (approximately 80% reporting). Some events were reported only several months after they had occurred. We assessed the completeness and accuracy of CBV data by comparing them to retrospective full pregnancy histories (FPH) collected during a census of the same clusters conducted in October-December 2013. We conducted all analyses separately by district, as well as for the combined sample of all districts. During the 21-month implementation period, the CBVs reported a total of 2,819 births and 137 under-five deaths. Among the latter, there were 84 infant deaths (55 neonatal deaths and 29 post-neonatal deaths). Comparison of the CBV data with FPH data suggested that CBVs significantly under-estimated child mortality: the estimated under-5 mortality rate according to CBV data was only 2/3 of the rate estimated from FPH data (95% Confidence Interval for the ratio of the two rates = 51.7 to 81.4). The discrepancies between the CBV and FPH estimates of infant and neonatal mortality were more limited, but varied significantly across districts. CONCLUSIONS: In northern Ghana, a community-based data collection systems relying on volunteers did not yield accurate estimates of child mortality rates. Additional implementation research is needed to improve the timeliness, completeness and accuracy of such systems. Enhancing pregnancy monitoring, in particular, may be an essential step to improve the measurement of neonatal mortality.