Spatiotemporal prediction of infectious diseases using structured Gaussian processes with application to Crimean-Congo hemorrhagic fever.

BACKGROUND: Infectious diseases are one of the primary healthcare problems worldwide, leading to millions of deaths annually. To develop effective control and prevention strategies, we need reliable computational tools to understand disease dynamics and to predict future cases. These computational tools can be used by policy makers to make more informed decisions. METHODOLOGY/PRINCIPAL FINDINGS: In this study, we developed a computational framework based on Gaussian processes to perform spatiotemporal prediction of infectious diseases and exploited the special structure of similarity matrices in our formulation to obtain a very efficient implementation. We then tested our framework on the problem of modeling Crimean-Congo hemorrhagic fever cases between years 2004 and 2015 in Turkey. CONCLUSIONS/SIGNIFICANCE: We showed that our Gaussian process formulation obtained better results than two frequently used standard machine learning algorithms (i.e., random forests and boosted regression trees) under temporal, spatial, and spatiotemporal prediction scenarios. These results showed that our framework has the potential to make an important contribution to public health policy makers.

Prevalence of rotavirus genotypes in children younger than 5 years of age before the introduction of a universal rotavirus vaccination program: report of rotavirus surveillance in Turkey.

BACKGROUND: Group A rotaviruses are the most common causative agent of acute gastroenteritis among children less than 5 years of age throughout the world. This sentinel surveillance study was aimed to obtain baseline data on the rotavirus G and P genotypes across Turkey before the introduction of a universal rotavirus vaccination program. METHODS: Rotavirus antigen-positive samples were collected from 2102 children less than 5 years of age who attended hospitals participating in the Turkish Rotavirus Surveillance Network. Rotavirus antigen was detected in the laboratories of participating hospitals by commercial serological tests such as latex agglutination, immunochromatographic test or enzyme immunoassay. Rotavirus G and P genotypes were determined by reverse transcription polymerase chain reaction (RT-PCR) using consensus primers detecting the VP7 and VP4 genes, followed by semi-nested type-specific multiplex PCR. RESULTS: RT-PCR found rotavirus RNA in 1644 (78.2%) of the samples tested. The highest rate of rotavirus positivity (38.7%) was observed among children in the 13 to 24 month age group, followed by children in the age group of 25 to 36 months (28.3%). A total of eight different G types, six different P types, and 42 different G-P combinations were obtained. Four common G types (G1, G2, G3, and G9) and two common P types (P[8] and P[4]) accounted for 95.1% and 98.8% of the strains, respectively. G9P[8] was the most common G/P combination found in 40.5% of the strains followed by G1P[8] (21.6%), G2P[8] (9.3%), G2P[4] (6.5%), G3P[8] (3.5%), and finally, G4P[8] (3.4%). These six common genotypes included 83.7% of the strains tested in this study. The rate of uncommon genotypes was 14%. CONCLUSION: The majority of the strains analyzed belonged to the G1-G4 and G9 genotypes, suggesting high coverage of current rotavirus vaccines. This study also demonstrates a dramatic increase in G9 genotype across the country.

PCR-based national bacterial meningitis surveillance in Turkey: years 2006 to 2009.

Polymerase chain reaction-based surveillance for bacterial meningitis including 841 children revealed 246 with bacterial DNA in cerebrospinal fluid samples of which 53% were Streptococcus pneumoniae, 19% Neisseria meningitidis, and 16% Haemophilus influenzae type b. The most common S. pneumoniae serotypes/serogroups were 1, 19F, 6A/6B, 23F, 5, 14, 18 and 19A. Among 47 meningococci, 86% were serogroup B, 6% serogroup C, 3% serogroup A, 3% serogroup X and 3% serogroup W.

Molecular characterization of measles viruses in Turkey (2010-2011): first report of genotype D9 involved in an outbreak in 2011.

Genetic characterization of measles viruses (MVs) combined with acquisition of epidemiologic information is essential for measles surveillance programs used in determining transmission pathways. This study describes the molecular characterization of 26 MV strains (3 from 2010, 23 from 2011) obtained from urine or throat swabs harvested from patients in Turkey. MV RNA samples (n = 26) were subjected to sequence analysis of 450 nucleotides comprising the most variable C-terminal region of the nucleoprotein (N) gene. Phylogenetic analysis revealed 20 strains from 2011 belonged to genotype D9, 3 to D4, 2 strains from 2010 to genotype D4 and 1 to genotype B3. This study represents the first report describing the involvement of MV genotype D9 in an outbreak in Turkey. The sequence of the majority of genotype D9 strains was identical to those identified in Russia, Malaysia, Japan, and the UK. Despite lack of sufficient epidemiologic information, the presence of variants observed following phylogenetic analysis suggested that exposure to genotype D9 might have occurred due to importation more than once. Phylogenetic analysis of five genotype D4 strains revealed the presence of four variants. Epidemiological information and phylogenetic analysis suggested that three genotype D4 strains and one genotype B3 strain were associated with importation. This study suggests the presence of pockets of unimmunized individuals making Turkey susceptible to outbreaks. Continuing molecular surveillance of measles strains in Turkey is essential as a means of acquiring epidemiologic information to define viral transmission patterns and determine the effectiveness of measles vaccination programs designed to eliminate this virus.

Seroprevalance of pertussis antibodies in maternal and cord blood of preterm and term infants.

BACKGROUND: The resurgence of pertussis has resulted in an increased morbidity and mortality, especially among young infants. The aim of our study was to determine the antibody concentrations against pertussis antigens in cord and maternal blood in both preterm and term infant-mother pairs and to evaluate the efficacy of transplacental antibody transfer. METHODS: Antibodies to pertussis toxin (PT) and filamentous hemagglutinin (FHA) in maternal and cord blood samples were measured by in-house enzyme linked immunosorbent assay (ELISA) in 100 preterm infant-mother and 100 term infant-mother pairs. Geometric mean concentrations (GMCs) of pertussis antibodies and cord:maternal GMC ratios were calculated. RESULTS: Cord GMCs for anti-PT and anti-FHA in the preterm group were 13.15 and 14.55 ELISA U/ml (EU/ml), respectively. Cord GMCs for anti-PT and anti-FHA in the term group were 19.46 and 19.18 EU/ml, respectively. Cord anti-PT GMC was significanlty lower in the preterm group (p=0.037). There were no differences between the groups with regard to maternal anti-PT and anti-FHA GMC. Placental transfer ratios for anti-PT and anti-FHA in preterms were 68% and 72%, respectively. The same ratios in terms were 107% and 120%, respectively and were significantly higher than those of preterms (p<0.001). Placental transfer ratios were even lower in preterms <32 weeks when compared to preterms ≥32 weeks and terms. There was a strong correlation between maternal and cord anti-pertussis antibody levels both in preterm and term infants. CONCLUSIONS: Anti-pertussis antibody levels were generally low in infant-mother pairs and would not be adequate to confer protection until the onset of primary immunization series. Transplacental anti-pertussis antibody transfers and antibody levels were lower in the cord blood of preterm infants, especially in those <32 weeks. These findings support the rationale for maternal immunization, which in combination with cocooning, could be a better option for preterm infants.

The epidemiology of Crimean-Congo hemorrhagic fever in Turkey, 2002-2007.

BACKGROUND: Crimean-Congo hemorrhagic fever (CCHF) is a serious disease caused by the CCHF virus of the Bunyaviridae family. The disease has been reported in 30 countries in Africa, Asia, Eastern Europe, and the Middle East. It has been present in Turkey since 2002. In this study we present and discuss the epidemiological features, clinical and laboratory findings, treatment, and outcome of cases diagnosed with CCHF between 2002 and 2007 from the surveillance results of the Turkish Ministry of Health (MoH). METHODS: According to the surveillance system of the MoH, data for patients with clinical, laboratory, and epidemiological findings compatible with CCHF are recorded on case reporting forms. These forms are submitted to the General Directorate of Primary Health Care of the MoH by the city health directorates. All the surveillance data regarding CCHF were recorded on a database (SSPS 11.0) established in the Communicable Diseases Department of the MoH. RESULTS: According to the surveillance reports of the Turkish MoH, between 2002 and 2007, 1820 CCHF cases occurred (150 in 2002-2003, 249 in 2004, 266 in 2005, 438 in 2006, and 717 in 2007). The crude fatality rate was calculated to be 5% (92/1820). Two thirds of the CCHF cases were reported from five cities located in the Mid-Eastern Anatolia region; 69.4% of the cases were from rural areas. The male to female ratio was 1.13:1. Of all the reported cases, 68.9% had a history of tick-bite or tick contact and 84.1% were seen in the months of May, June, and July. Of 1820 CCHF cases, three (0.16%) were nosocomial infections. CONCLUSIONS: CCHF appears to be a seasonal problem in the Mid-Eastern Anatolia region of Turkey. The possible risk factors for transmission and the clinical and laboratory findings of patients with a diagnosis of CCHF were found to be similar to those reported in the literature. The mean fatality rate for Turkey is lower than the rate reported for other series from other parts of the world.