Exploring convolutional neural networks and spatial video for on-the-ground mapping in informal settlements.

BACKGROUND: The health burden in developing world informal settlements often coincides with a lack of spatial data that could be used to guide intervention strategies. Spatial video (SV) has proven to be a useful tool to collect environmental and social data at a granular scale, though the effort required to turn these spatially encoded video frames into maps limits sustainability and scalability. In this paper we explore the use of convolution neural networks (CNN) to solve this problem by automatically identifying disease related environmental risks in a series of SV collected from Haiti. Our objective is to determine the potential of machine learning in health risk mapping for these environments by assessing the challenges faced in adequately training the required classification models. RESULTS: We show that SV can be a suitable source for automatically identifying and extracting health risk features using machine learning. While well-defined objects such as drains, buckets, tires and animals can be efficiently classified, more amorphous masses such as trash or standing water are difficult to classify. Our results further show that variations in the number of image frames selected, the image resolution, and combinations of these can be used to improve the overall model performance. CONCLUSION: Machine learning in combination with spatial video can be used to automatically identify environmental risks associated with common health problems in informal settlements, though there are likely to be variations in the type of data needed for training based on location. Success based on the risk type being identified are also likely to vary geographically. However, we are confident in identifying a series of best practices for data collection, model training and performance in these settings. We also discuss the next step of testing these findings in other environments, and how adding in the simultaneously collected geographic data could be used to create an automatic health risk mapping tool.

Socioeconomic Vulnerability and Sexually Transmitted Infection Among Pregnant Haitian Women.

BACKGROUND: Despite evidence that education and poverty act through distinct pathways to influence sexually transmitted infection (STI), few studies have examined the unique, independent associations of these socioeconomic vulnerabilities with sexual risk behaviors and STI among women. METHODS: From August to October 2013, women at an antenatal clinic in Gressier, Haiti, were interviewed and tested for chlamydial infection, gonorrhea, and trichomoniasis (N = 200). We measured low educational attainment as less than 9 years of schooling and currently living in poverty based on crowding, defined as more than 2 people sleeping in one room. We used logistic regression to estimate independent associations between each socioeconomic indicator and outcomes of sexual behaviors and STI. RESULTS: Approximately 29% of the sample had a current STI (chlamydia, 8.0%; gonorrhea, 3.0%; trichomoniasis, 20.5%), with 2.5% testing positive for more than 1 STI. Forty percent of the sample reported low educational attainment and 40% reported current poverty. Low educational attainment was associated with early risk behaviors, including twice the odds of earlier sexual debut (adjusted odds ratio [AOR], 2.09; 95% confidence interval [CI],: 1.14-3.84). Poverty was associated with reporting the current main sexual partner to be nonmonogamous (AOR, 2.01; 95% CI, 1.00-4.01) and current STI (AOR, 2.50; 95% CI, 1.26-4.98). CONCLUSIONS: Education and poverty seem to independently influence STI behaviors and infection, with low education associated with early sexual risk and poverty associated with current risk and infection. Improving women's educational attainment may be important in improving risk awareness, thereby reducing risky sexual behaviors and preventing a trajectory of STI risk.

Association of cardiac disease and alcohol use with the development of severe ciguatera.

OBJECTIVES: Ciguatera is a foodborne illness that causes severe gastrointestinal and neurologic symptoms, but the risk factors for illness are not well established. METHODS: In St Thomas, US Virgin Islands, we performed a case-control study for ciguatera, enrolling 47 patients primarily through the emergency department and 141 age- and sex-matched controls identified through an island-wide survey. We used conditional logistic regression to assess health-related risk factors for ciguatera, including comorbidities and alcohol and tobacco use. RESULTS: Cases were more likely than controls to report alcohol consumption at least weekly (odds ratio 5.7, 95% confidence interval 2.7-24.8). Cases also more frequently had a history of heart disease (odds ratio 6.3, 95% confidence interval 1.2-32.7). Previous ciguatera episodes, lower levels of education, and frequent fish consumption also were associated with illness. CONCLUSIONS: This study provides preliminary evidence that preexisting health status influences an individual's risk of developing severe ciguatera. There may be a benefit to educating high-risk individuals about preventing ciguatera.

Spatial Video and EpiExplorer: A Field Strategy to Contextualize Enteric Disease Risk in Slum Environments.

Disease risk associated with contaminated water, poor sanitation, and hygiene in informal settlement environments is conceptually well understood. From an analytical perspective, collecting data at a suitably fine scale spatial and temporal granularity is challenging. Novel mobile methodologies, such as spatial video (SV), can complement more traditional epidemiological field work to address this gap. However, this work then poses additional challenges in terms of analytical visualizations that can be used to both understand sub-neighborhood patterns of risk, and even provide an early warning system. In this paper, we use bespoke spatial programming to create a framework for flexible, fine-scale exploratory investigations of simultaneously-collected water quality and environmental surveys in three different informal settlements of Port-au-Prince, Haiti. We dynamically mine these spatio-temporal epidemiological and environmental data to provide insights not easily achievable using more traditional spatial software, such as Geographic Information System (GIS). The results include sub-neighborhood maps of localized risk that vary monthly. Most interestingly, some of these epidemiological variations might have previously been erroneously explained because of proximate environmental factors and/or meteorological conditions.

Long Term Memory Outcome of Repetitive, Low-Level Dietary Exposure to Domoic Acid in Native Americans.

Domoic acid (DA) is a marine-based neurotoxin that, if ingested via tainted shellfish, is associated with Amnesic Shellfish Poisoning (ASP). These acute effects of elevated DA exposure in humans have been well described. In contrast, the long-term impacts of lower level, repetitive, presumably safe doses of DA (less than 20 ppm) are minimally known. Since Native Americans (NA) residing in coastal communities of the Pacific NW United States are particularly vulnerable to DA exposure, this study focuses on the long-term, 8-year memory outcome associated with their repeated dietary consumption of the neurotoxin. Measures of razor clam consumption, memory, clerical speed and accuracy, and depression were administered over eight years to 500 randomly selected adult NA men and women ages 18-64. Data were analyzed using GEE analyses taking into consideration the year of study, demographic factors, and instrumentation in examining the association between dietary exposure and outcomes. Findings indicated a significant but small decline in total recall memory within the context of otherwise stable clerical speed and accuracy and depression scores. There is reason to believe that a continuum of memory difficulties may be associated with DA exposure, rather than a unitary ASP syndrome.

The Short- and Long-Term Impacts of Hurricane Irma on Florida Agricultural Leaders as Early Emergency Responders: The Importance of Workplace Stability.

The impacts of hurricane-related disasters in agricultural communities include extensive losses of fields, orchards, and livestock, the recovery of which could span many years. Agricultural Extension Agents (EAs) try to manage and mitigate these losses, while simultaneously overseeing emergency shelter operations. These non-professional emergency responders face numerous potential stressors, the outcomes of which are minimally known. This study examined the short- and long-term medical and behavioral outcome of 36 University of Florida Agricultural Extension Agents within two months and one year after Hurricane Irma, Florida, USA, taking into consideration personal/home and work-related hurricane impacts. Regression analyses indicated that combined home and work hurricane impacts were associated with greater anxiety, depression, and medical symptoms controlling for age and number of prior hurricane experiences within two months of landfall. One year later, depression symptoms increased as well as the use of negative disengagement coping strategies for which stability of the work environment was protective. The findings suggest that advanced training in emergency response, organization and time management skills, time off and temporary replacement for personally impacted EAs, and workplace stability, including enhanced continuity of operations plans, represent critical elements of health prevention and early intervention for this occupational group.

Isolation of Mayaro Virus from a Venezuelan Patient with Febrile Illness, Arthralgias, and Rash: Further Evidence of Regional Strain Circulation and Possible Long-Term Endemicity.

Fifty-two febrile patients living in Barquisimeto, Venezuela, were screened for arbovirus infection by virus culture during an outbreak of what was thought to be Zika virus infection. We report identification of Mayaro virus (MAYV) on culture of plasma from one patient, an 18-year-old woman with acute febrile illness, arthralgias, and psoriasiform rash. The strain was sequenced and was found to be most closely related to a 1999 strain from French Guiana, which, in turn, was related to two 2014 strains from Haiti. By contrast, previously reported outbreak-related MAYV strains from a sylvatic area approximately 80 miles from where the case patient lived were most closely related to Peruvian isolates. The two strain groups show evidence of having diverged genetically approximately 100 years ago.

Micro-Space Complexity and Context in the Space-Time Variation in Enteric Disease Risk for Three Informal Settlements of Port au Prince, Haiti.

Diffusion of cholera and other diarrheal diseases in an informal settlement is a product of multiple behavioral, environmental and spatial risk factors. One of the most important components is the spatial interconnections among water points, drainage ditches, toilets and the intervening environment. This risk is also longitudinal and variable as water points fluctuate in relation to bacterial contamination. In this paper we consider part of this micro space complexity for three informal settlements in Port au Prince, Haiti. We expand on more typical epidemiological analysis of fecal coliforms at water points, drainage ditches and ocean sites by considering the importance of single point location fluctuation coupled with recording micro-space environmental conditions around each sample site. Results show that spatial variation in enteric disease risk occurs within neighborhoods, and that while certain trends are evident, the degree of individual site fluctuation should question the utility of both cross-sectional and more aggregate analysis. Various factors increase the counts of fecal coliform present, including the type of water point, how water was stored at that water point, and the proximity of the water point to local drainage. Some locations fluctuated considerably between being safe and unsafe on a monthly basis. Next steps to form a more comprehensive contextualized understanding of enteric disease risk in these environments should include the addition of behavioral factors and local insight.

Acute Respiratory Illness in Rural Haiti.

OBJECTIVES: Acute Respiratory Infection (ARI) is the most common cause of childhood morbidity and mortality in developing countries, including Haiti. Our objective was to detect pathogens found in children with ARI in rural Haiti to help develop evidence-based guidelines for treatment and prevention. METHODS: Retrospective study of students with ARI at four schools in rural Haiti. Viral and/or bacterial pathogens were identified by qPCR in 177 nasal swabs collected from April 2013 through November 2015. RESULTS: Most common viruses detected were Rhinovirus (36%), Influenza A (16%) and Adenovirus (7%), and bacteria were Streptococcus pneumoniae (58%) and Staphylococcus aureus (28%). Compared to older children, children aged 3-5 years had more Influenza A (28% vs. 9%, p=0.002) and Adenovirus detected (14% vs. 3%, p=0.01). Similarly, S. pneumoniae was greatest in children 3-5 years old (71% 3-5yrs; 58% 6-15 years; 25% 16-20 years; p=0.008). Children 3-10 years old presented with fever more than children 11-20 years old (22% vs 7%; p=0.02) and were more often diagnosed with pneumonia (28% vs 4%, p<0.001). CONCLUSIONS: Younger children had increased fever, pneumonia, and detection of Influenza A and S. pneumoniae. These data support the need for influenza and pneumococcus vaccination in early childhood in Haiti.

Lateral gene transfer of O1 serogroup encoding genes of Vibrio cholerae.

In Gram-negative bacteria, the O-antigen-encoding genes may be transferred between lineages, although mechanisms are not fully understood. To assess possible lateral gene transfer (LGT), 21 Argentinean Vibrio cholerae O-group 1 (O1) isolates were examined using multilocus sequence typing (MLST) to determine the genetic relatedness of housekeeping genes and genes from the O1 gene cluster. MSLT analysis revealed that 4.4% of the nucleotides in the seven housekeeping loci were variable, with six distinct genetic lineages identified among O1 isolates. In contrast, MLST analysis of the eight loci from the O1 serogroup region revealed that 0.24% of the 4943 nucleotides were variable. A putative breakpoint was identified in the JUMPstart sequence. Nine conserved nucleotides differed by a single nucleotide from a DNA uptake signal sequence (USS) also found in Pastuerellaceae. Our data indicate that genes in the O1 biogenesis region are closely related even in distinct genetic lineages, indicative of LGT, with a putative DNA USS identified at the defined boundary for the DNA exchange.

A new "American" subgroup of African-lineage Chikungunya virus detected in and isolated from mosquitoes collected in Haiti, 2016.

As part of on-going arboviral surveillance activity in a semi-rural region in Haiti, Chikungunya virus (CHIKV)-positive mosquito pools were identified in 2014 (the peak of the Caribbean Asian-clade epidemic), and again in 2016 by RT-PCR. In 2014, CHIKV was only identified in Aedes aegypti (11 positive pools/124 screened). In contrast, in sampling in 2016, CHIKV was not identified in Ae. aegypti, but, rather, in (a) a female Aedes albopictus pool, and (b) a female Culex quinquefasciatus pool. Genomic sequence analyses indicated that the CHIKV viruses in the 2016 mosquito pools were from the East-Central-South African (ECSA) lineage, rather than the Asian lineage. In phylogenetic studies, these ECSA lineage strains form a new ECSA subgroup (subgroup IIa) together with Brazilian ECSA lineage strains from an isolated human outbreak in 2014, and a mosquito pool in 2016. Additional analyses date the most recent common ancestor of the ECSA IIa subgroup around May 2007, and the 2016 Haitian CHIKV genomes around December 2015. Known CHIKV mutations associated with improved Ae. albopictus vector competence were not identified. Isolation of this newly identified lineage from Ae. albopictus is of concern, as this vector has a broader geographic range than Ae. aegypti, especially in temperate areas of North America, and stresses the importance for continued vector surveillance.

Controlling cholera in the Ouest Department of Haiti using oral vaccines.

Following the 2010 cholera outbreak in Haiti, a plan was initiated to provide massive improvements to the sanitation and drinking water infrastructure in order to eliminate cholera from the island of Hispaniola by 2023. Six years and a half billion dollars later, there is little evidence that any substantial improvements have been implemented; with increasing evidence that cholera has become endemic. Thus, it is time to explore strategies to control cholera in Haiti using oral cholera vaccines (OCVs). The potential effects of mass administration of OCVs on cholera transmission were assessed using dynamic compartment models fit to cholera incidence data from the Ouest Department of Haiti. The results indicated that interventions using an OCV that was 60% effective could have eliminated cholera transmission by August 2012 if started five weeks after the initial outbreak. A range of analyses on the ability of OCV interventions started January 1, 2017 to eliminate cholera transmission by 2023 were performed by considering different combinations of vaccine efficacies, vaccine administration rates, and durations of protective immunity. With an average of 50 weeks for the waiting time to vaccination and an average duration of three years for the vaccine-induced immunity, all campaigns that used an OCV with a vaccine efficacy of at least 60% successfully eliminated cholera transmission by 2023. The results of this study suggest that even with a relatively wide range of vaccine efficacies, administration rates, and durations of protective immunity, future epidemics could be controlled at a relatively low cost using mass administration of OCVs in Haiti.

Cultural influences behind cholera transmission in the Far North Region, Republic of Cameroon: a field experience and implications for operational level planning of interventions.

INTRODUCTION: In recent years, the Far North Region of Cameroon has experienced serious and recurrent cholera outbreaks. Yet, understanding of cultural influences on outbreaks and spread remain poorly understood. This qualitative study explored cultural influences on cholera exposure in this region. METHODS: Interviews and group discussions were conducted in two phases. Phase I involved key informants and phase II included focus group and household discussions. Thematic techniques including word repetition, key-indigenous-terms, and key-words-in-context were used for qualitative data analysis. RESULTS: Key informants attributed cholera etiology to dirt and spread through water (caneri) and food (group eating or faire-un-rond) while group discussions attributed it to a reprimand from god and transmission through the air. Participants suggested that funerals, weddings, open defecation, and mountaintop burial might influence cholera exposure and facilitate its spread. Hospital avoidance and non-adherence with cholera treatment regimens were linked to favorable beliefs about traditional medicine (rural-urban mentality confrontation). Furthermore, a multiplicity of ethnic languages, mistrust of message sources, culture of dependency and sentimental animal husbandry were barriers to the reception of public health messages. CONCLUSION: Many participants had limited scientific knowledge about cholera etiology and transmission. The cultural practice of mountain burial seemed to explain the high cholera attack rate in the mountainous terrain compared to the floodplains. Cultural factors are likely to play important roles in the exposure to and spread of cholera. Understanding cultural context, individual and community perceptions of risk and disease may help public health agencies in response to outbreak prevention and control.

Detection of Alphacoronavirus vRNA in the Feces of Brazilian Free-Tailed Bats (Tadarida brasiliensis) from a Colony in Florida, USA.

Bats are natural reservoirs of coronaviruses and other viruses with zoonotic potential. Florida has indigenous non-migratory populations of Brazilian free-tailed bats (Tadarida brasiliensis) that mostly roost in colonies in artificial structures. Unlike their counterparts in Brazil and Mexico, the viruses harbored by the Florida bats have been underexplored. We report the detection of an alphacoronavirus RNA-dependent RNA polymerase (RdRp) gene sequence in the feces of two of 19 different T. brasiliensis that were capture/release bats that had been evaluated for overall health. The RdRp sequence is similar but not identical to previously detected sequences in the feces of two different species of bats (T. brasiliensis and Molossus molossus) in Brazil. In common with the experience of others doing similar work, attempts to isolate the virus in cell cultures were unsuccessful. We surmise that this and highly related alphacoronavirus are carried by Brazilian free-tailed bats living in a wide eco-spatial region. As various coronaviruses (CoVs) that affect humans emerged from bats, our study raises the question whether CoVs such as the one detected in our work are yet-to-be-detected pathogens of humans and animals other than bats.

Cholera public health surveillance in the Republic of Cameroon-opportunities and challenges.

INTRODUCTION: In Cameroon, cholera has periodically resurfaced since it was first reported in 1971. In 2003, Cameroon adapted the Integrated Disease Surveillance and Response (IDSR) strategy to strengthen surveillance in the country. This study was an in-depth description and assessment of the structure, core and support functions, and attributes of the current cholera surveillance system in Cameroon. It also discussed its strengths and challenges with hope that lessons learned could improve the system in Cameroon and in other countries in Africa implementing the IDSR strategy. METHODS: Semi-structured key informant interviews, peer reviewed articles, and government record review were conducted in the Far North and Centre Regions of Cameroon. We used the matrix and conceptual framework from the World Health Organization (WHO) and Centers for Disease Control and Prevention, WHO Regional Office for Africa Technical Guidelines to frame the study. Site visits included the WHO country office, the ministry of public health (MoPH), two Regional Public Health Delegations (RPHDs), eight health districts (HDs) and health facilities (HFs) including two labs. RESULTS: Cholera surveillance is passive but turns active during outbreaks and follows a hierarchical structure. Cholera data are collected at HFs and sent to HDs where data are compiled and sent to the RPHD in paper format. RPHDs de-identify, digitalize, and send the data to the MoPH via internet and from there to the WHO. The case definition was officially changed in 2010 but the outdated definition was still in use in 2013. Nationally, there are 3 laboratories that have the ability to confirm cholera cases; the lack of laboratory capacity at HFs hampers case and outbreak confirmation. The absence of structured data analysis at the RPHD, HD, and HF further compounds the situation, making the goal of IDSR of data analysis and rapid response at the HD very challenging. Feedback is strongest at the central level (MoPH) and non-existent at the levels below it, with only minimal training and supervision of staff. In 2012, mobile phone coverage expanded to all 183 HDs and to HFs in 2014 in the Far North and North Regions. The phones improved immediate reporting and outbreak control. Further, the creation of cholera command and control centers, and introduction of laptops at all RPHDs are major strengths in the surveillance system. Completeness and timeliness of reporting varied considerably among levels. CONCLUSION: Significant milestones in the hierarchical structure towards integration and achieving early detection and rapid response in cholera surveillance are in effective use; however, some challenges exist. The surveillance system lack labs at HFs and there is no data analysis at HD level. Thus, the goal of IDSR-strategy of early detection, data analysis, and rapid response at the HD level is a challenge. Both human and material resources are needed at the HD level to achieve this goal.

Cholera in Cameroon, 2000-2012: Spatial and Temporal Analysis at the Operational (Health District) and Sub Climate Levels.

INTRODUCTION: Recurrent cholera outbreaks have been reported in Cameroon since 1971. However, case fatality ratios remain high, and we do not have an optimal understanding of the epidemiology of the disease, due in part to the diversity of Cameroon's climate subzones and a lack of comprehensive data at the health district level. METHODS/FINDINGS: A unique health district level dataset of reported cholera case numbers and related deaths from 2000-2012, obtained from the Ministry of Public Health of Cameroon and World Health Organization (WHO) country office, served as the basis for the analysis. During this time period, 43,474 cholera cases were reported: 1748 were fatal (mean annual case fatality ratio of 7.9%), with an attack rate of 17.9 reported cases per 100,000 inhabitants per year. Outbreaks occurred in three waves during the 13-year time period, with the highest case fatality ratios at the beginning of each wave. Seasonal patterns of illness differed strikingly between climate subzones (Sudano-Sahelian, Tropical Humid, Guinea Equatorial, and Equatorial Monsoon). In the northern Sudano-Sahelian subzone, highest number of cases tended to occur during the rainy season (July-September). The southern Equatorial Monsoon subzone reported cases year-round, with the lowest numbers during peak rainfall (July-September). A spatial clustering analysis identified multiple clusters of high incidence health districts during 2010 and 2011, which were the 2 years with the highest annual attack rates. A spatiotemporal autoregressive Poisson regression model fit to the 2010-2011 data identified significant associations between the risk of transmission and several factors, including the presence of major waterbody or highway, as well as the average daily maximum temperature and the precipitation levels over the preceding two weeks. The direction and/or magnitude of these associations differed between climate subzones, which, in turn, differed from national estimates that ignored subzones differences in climate variables. CONCLUSIONS/SIGNIFICANCE: The epidemiology of cholera in Cameroon differs substantially between climate subzones. Development of an optimal comprehensive country-wide control strategy for cholera requires an understanding of the impact of the natural and built environment on transmission patterns at the local level, particularly in the setting of ongoing climate change.

Epidemiology of Ciguatera in Florida.

Ciguatera is the most commonly reported marine food-borne illness worldwide. Because there is a biological plausibility that ciguatera may be impacted by long-term climate variability and Florida is on the northern border of the geographic distribution of ciguatera, it is important to update our understanding of its epidemiology in Florida. We performed an analysis of 291 reports in Florida from 2000 to 2011 and an e-mail survey of 5,352 recreational fishers to estimate incidence and underreporting and identify high risk demographic groups, fish types, and catch locations. Incidence was 5.6 per 100,000 adjusted for underreporting. Hispanics had the highest incidence rate (relative risk [RR] = 3.4) and were more likely to eat barracuda than non-Hispanics. The most common catch locations for ciguatera-causing fish were the Bahamas and Florida Keys. Cases caused by fish from northern Florida were infrequent. These results indicate that ciguatera incidence is higher than estimated from public health reports alone. There is little evidence that incidence or geographic range has increased because of increased seawater temperatures since earlier studies.

Increased isolation frequency of toxigenic Vibrio cholerae O1 from environmental monitoring sites in Haiti.

Since the identification of the first cholera case in 2010, the disease has spread in epidemic form throughout the island nation of Haiti; as of 2014, about 700,000 cholera cases have been reported, with over 8,000 deaths. While case numbers have declined, the more fundamental question of whether the causative bacterium, Vibrio cholerae has established an environmental reservoir in the surface waters of Haiti remains to be elucidated. In a previous study conducted between April 2012 and March 2013, we reported the isolation of toxigenic V. cholerae O1 from surface waters in the Ouest Department. After a second year of surveillance (April 2013 to March 2014) using identical methodology, we observed a more than five-fold increase in the number of water samples containing culturable V. cholerae O1 compared to the previous year (1.7% vs 8.6%), with double the number of sites having at least one positive sample (58% vs 20%). Both seasonal water temperatures and precipitation were significantly related to the frequency of isolation. Our data suggest that toxigenic V. cholerae O1 are becoming more common in surface waters in Haiti; while the basis for this increase is uncertain, our findings raise concerns that environmental reservoirs are being established.