Research Projects

Built Environment and Longitudinal Data Analysis

Social Determinants of Obesity Among Hispanic Adults: A Longitudinal Study (Active)

Hispanic individuals in the United States (US) face a disproportionate risk for obesity and related diseases compared to Non-Hispanic White individuals (NHW). Yet, their disease risk is lower than would be expected given the lower income and wealth among Hispanic communities compared to most other racial and ethnic groups. This has been called the “Hispanic health paradox”, and is especially strong among foreign-born Hispanics. However, longitudinal studies of these relationships in Hispanics across the U.S., are sparse to none. The primary aim of this study is to examine the longitudinal relationship between social support, social cohesion, and risk for objectively measured obesity for Hispanic individuals.

Evaluating differential exposure to food environment: the Multi-Ethnic Study of Atherosclerosis (Active)

Measures of the neighborhood food retail environment have been associated with dietary patterns and cardiovascular risk in epidemiologic studies. Since the implications for exposure assignment during this assessment in cohort studies are unclear, we aimed to examine if the characterization of the food environment for MESA participants differed by NETS and InfoUSA data sources for different exposures.

Fast-food and convenience outlets near private and public schools in California (Completed)

Food environments near schools (FENS) influence children's dietary habits and contribute to obesity. We compare the availability of fast-food restaurants (FFR) and convenience stores (CS) across indicators of schools' neighborhood socioeconomic characteristics: whether the school is a public vs. private school. We used diverse datasets, including locations of schools, FFR and CS, neighborhood urbanization, network buffer from schools' locations, and the names of the outlets. Count regression models were used to estimate the associations. We also used text-mining techniques to explore brand names and prevalent FFR and CS types. (Image created by AI)

Assessing Schools Racial Diversity of California: A Longitudinal Study (Active)

Research consistently shows that social determinants of health, including factors like racial/ethnic segregation, socioeconomic, and gender contributes to health disparities across geographic regions and social groups. In particular, racial/ethnic segregation is one social determinant. This study investigates the degree to which segregation has changed over approximately 20 years in California public schools, and to examine whether characteristics of school locations are associated with the patterns of change in segregation within schools. (Image created by AI)

Characterizing longitudinal mixtures of retail environment features and their associations with Health Outcome: the MESA (Active)

We propose to identify longitudinal neighborhood typologies for participants in MESA and evaluate associations between these typologies, incorporating multiple spatial scales and potential temporal transitions, with longitudinal changes in health outcomes related to obesity and cardiovascular risk and have been shown to be related to built environment exposures.

Image credit: Clinical Chemistry- https://www.eurekalert.org/news-releases/734690

Spatial Data Analysis and Bayesian Modeling

Spatial distribution of food outlets near private and public schools in California (Active)

Quantifying exposure to the food environment (FE) is challenging given its complexity. For example, multiple outlet types are available; metrics depend on the spatial scale used to create them; large differences across geographical regions result in multimodality and overdispersion; and even outlet counts or locations are similar, the composition of outlets brands can differ across places. We use a variety of methods to compare the FE near private vs. public schools in California, multivariate regressions for over dispersed count data to simultaneously examine the count of multiple types of food outlets and their spatial distribution.

Race/sex disparities in heart disease mortality (Completed)

We’re interested in investigating changes in disparities in heart disease mortality rates in North Carolina counties by both race and sex. In particular, we wish to assess the degree to which gaps between rates for whites and blacks have increased/decreased. We aimed to identify does rates for each race/sex combination changed over time? Is everyone experiencing a significant decline? Do the race/sex combinations exhibit common geographic trends Does there appear to be spatial clustering in these ratios?

Spatial Disparities of Georgia Colon Cancer (Completed)

Increase in colon cancer rates has been a concern in Georgia till date. We tried to summarize the standardized incidence ratio (SIR) to investigate the relationship between the SIR for CRC and the covariates provided through spatial temporal modeling.

Geographic and Racial Disparities of Obesity in California (Partially Completed)

Childhood obesity remains a significant public health concern in the United States. Data from 2017-2019 indicates a persistent prevalence of 19.7 percent among youth aged 2-19 years, highlighting the need for continued efforts to address this complex issue. While the topics of racial and geographic disparities are essential in their own right, it is crucial to study geographic trends in racial disparities, as this can shed light on regions in which all races/ethnicities also have similar health outcomes and regions where some racial/ethnic populations are being left behind. This study aims to quantify and investigate the geographic disparities of obesity among White and Hispanic racial groups. Explore the clusters of obesity among racial groups and the changes of clusters with time. Identify the temporal effect on geographic disparities.

Oncology Research and Survival Analysis Application

Effect of Chemo and Hormone Therapies on Tumor Recurrence in Post-Surgery Breast Cancer Patients (Completed)

To determine survival prognosis for primary breast cancer patients who received surgery and compare survival of patients who received chemo or hormone therapy (CHT) in addition to surgery vs. those who only had surgery, adjusting for age, menopause status, size of tumor, and grade of tumor. We used Martingale score plots to test the violation of the PH assumption of considered variables before fitting an extended Cox model. We also fit several parametric models and tested the AFT assumption for our primary variable of interest. After determining the appropriate model, we assessed the impacts of predictors on survival time. We found that the extended Cox approach was the most appropriate model. The model showed significant effects for CHT in addition to surgery, reducing the hazard of breast cancer recurrence before 550 days and increasing the hazard after 550 days. Semi-parametric methods using a stepwise approach were used to examine the data. Significant differences were found between groups before and after 550 days post-surgery. Strengths and limitations of our approach are discussed, as well as directions for future research.

Author: Gianni Anfuso, Md. Karimuzzaman, Katherine Ardeleanu, Darius McDaniel

Machine Learning Application

Predicting School Racial Diversity by Spatial Generalized Least Square Random Forest (Completed)

Machine learning algorithms often struggle to effectively explain and identify patterns in spatially correlated data. Recent literature has highlighted this challenge, particularly in the context of random forest models, which do not inherently account for spatial dependency. To address this limitation, Saha, Basu, and Datta (2020) have proposed a novel approach known as a generalized least square (GLS) based random forest (RF-GLS), which considers a location’s nearest neighbors. This study aims to compare the performance of the conventional random forest model with the newly proposed RF-GLS method in predicting the racial diversity of California public schools. The approach involves incorporating spatial characteristics of school locations through kriging-based prediction of spatial responses at new locations.

Text Mining and Sentiment Analysis of Newspaper Headlines (Completed)

The purpose of this paper is to examine the pattern of words that appeared on the front page of a well-known daily English newspaper in Bangladesh, The Daily Star, in 2018 and 2019. The elucidation of that era’s possible social and political context was also attempted using word patterns. The study employs three widely used and contemporary text mining techniques: word clouds, sentiment analysis, and cluster analysis. The word cloud reveals that election, kill, cricket, and Rohingya-related terms appeared more than 60 times in 2018, whereas BNP, poll, kill, AL, and Khaleda appeared more than 80 times in 2019. These indicated the country’s passion for cricket, political turmoil, and Rohingya-related issues. Furthermore, sentiment analysis reveals that words of fear and negative emotions appeared more than 600 times, whereas anger, anticipation, sadness, trust, and positive-type emotions came up more than 400 times in both years.

Finally, the clustering method demonstrates that election, politics, deaths, digital security act, Rohingya, and cricket-related words exhibit similarity and belong to a similar group in 2019, whereas rape, deaths, road, and fire-related words clustered in 2018 alongside a similar-appearing group. In general, this analysis demonstrates how vividly the text mining approach depicts Bangladesh’s social, political, and law-and-order situation, particularly during election season and the country’s cricket craze, and also validates the significance of the text mining approach to understanding the overall view of a country during a particular time in an efficient manner.

COVID-19 and Global Health Research

Preventive Actions and Healthcare Factors in Controlling COVID-19 Pandemic (Completed)

With the insurgence of the COVID-19 pandemic, many people died in the past several months, and the situation is ongoing with increasing health, social, and economic panic and vulnerability. This study generates evidence of taking the most impactful actions to combat COVID-19. In order to generate community-based scientific evidence, this study analyzed the outcome of COVID-19 in response to different control measures, healthcare facilities, life expectancy, and prevalent diseases.

We used more than a hundred countries’ data collected from different databases. The reduction of COVID-19 cases is strongly correlated with the earliness of preventive initiation. The apathy of taking nationwide immediate precaution measures has been identified as one of the critical reasons to make the circumstances worse. There is significant non-linear relationship between COVID-19 case fatality and number of physicians, nurses and midwives, hospital beds, life expectancy of both sexes, life expectancy of female, and life expectancy of male. COVID-19 deaths were found to be reduced with increased medical personnel and hospital beds. No association between the comorbidities and severity of COVID-19 was found excluding asthma, cancer, Alzheimer’s, and smoking. Enhancing healthcare facilities and early imposing the control measures could be valuable to prevent the COVID-19 pandemic. No association between COVID-19 and other comorbidities warranted further investigation at the pathobiological level.

Routine Vaccination Disruption in Low-Income Countries Due to COVID-19 (Completed)

The long-term impact of this pandemic will continue in almost all sectors of a country, such as health, economic situations, education, mental health, and violence. The authors intended to discuss the prolonged effect of COVID-19 on Bangladesh's health, well-being, education, and economy through a mixed approach. It is realized that healthcare services and resources are always essential for predominant health conditions and fatal disease patients. Mental health has also been impacted a lot during this pandemic. The country's youth unemployment is projected to intensify due to the economic effects, which would promote antisocial behavior and cause social discontent among young people. Therefore, many social services systems need strategic backup resources at community, national, and global levels if any basic system may collapse due to COVID-19 and socio-economic and geopolitical negligence in handling post-pandemic challenges.

Maternal and Child Health Research

Trends and long-term variation explaining nutritional determinants of child linear growth (Completed)

To examine the height-for-age z-score (HAZ) of 0–35 months’ children along with stunting prevalence to identify trends, changes, and available nutrition-sensitive and specific determinants that could help explain the long-term variation in child linear growth using successive Bangladesh Demographic and Health Surveys (BDHS) data from 1996 to 2018.

The mean HAZ increased by 0·91(±1·53) with 0·041 annual average change, while the percentages of stunting (–26·63 ± 0·54) and severe stunting (–21·12 ± 0·48) showed a reduction with 1·21 and 0·96 average annual changes, respectively. The nutrition-sensitive and specific factors identified through regression decomposition describing long-term variation in child linear growth should be focused further to attain sustainable development goals.

Association of dietary diversity of 6-23 months aged children with prenatal and postnatal obstetric care (Completed)

Dietary diversity is a key determinant of infant and young child eating patterns for a variety of food groups taken by children between the ages of 6–23 months. The study aimed to examine the association between prenatal and postnatal obstetric care factors of mother and child’s dietary diversity, and specific food practices.

Findings revealed that ≥ 4 antenatal care (ANC) visits care visits increased the DDS, increased the likelihood of MDD, and ISSSF, consuming eggs, and vitamin A vegetables and fruits. Moreover, DDS and MDD are linked to childbirth in a medical facility. The C-section delivery influences the DDS, MDD, and ISSSF. Besides, postnatal visits within 48h of delivery linked to MDD and ISSSF, and physicians or professionals providing postnatal checkups were significantly associated with DDS, MDD, and ISSSF. Knowledge of child nutritional feeding should emphasize during prenatal and postnatal obstetric care of mother, particularly during antenatal and postnatal visits, C-section delivery, and birth in a healthcare facility to eradicate malnutrition and establish healthy child feeding practices.

Likelihood of Infectious Diseases (Diarrhea/ARI) Due to Lack of Exclusive Breastfeeding of Infants (0-6 Months) (Completed)

Infectious diseases are leading causes of child mortality, and lack of exclusive breastfeeding (EBF) among infants aged 0–6 months increases child morbidity and mortality from various infectious diseases in developing countries. However, as per existing literature, no study has been conducted yet to determine the lack of EBF practice effect on child mortality in Bangladesh. With this backdrop, the authors intend to measure the likelihood of infectious diseases due to the lack of EBF in infants aged 0–6 months in Bangladesh.

The findings of this study emphasize the importance of EBF up to six months of age of infants against diarrhea and ARI-specific morbidity and mortality. Our results also agreed with the recommendation of the World Health Organization (WHO), United Nations International Children’s Emergency Fund (UNICEF), American Academy of Pediatrics (AAP), American Academy of Family Physicians (AAFP), and National Nutrition Programme of Ethiopia (NNPE) that the EBF practice for the first six months of age could be a best, cost-effective, long-lasting natural preventive way to reduce the child morbidity and mortality due to infectious diseases in developing countries.

Influencing factors associated with maternal delivery at home in urban areas (Completed)

The associated factors and patterns of giving birth in home settings of rural areas have been extensively studied in Bangladeshi literature. However, urban areas still need to be explored, particularly with recent data. Therefore, the authors aimed to investigate the influential determinants of delivery at home in urban areas of Bangladesh.

Despite significant progress in women and reproductive health in Bangladesh, the proportion of delivery in the home in urban areas is alarming and should be emphasized more. The authors believe the identified factors will help design interventions and policy development on this issue.