United States

Machine learning is an area of artificial intelligence that is responsible for advancements in a broad range of fields such as facial recognition, self-driving cars, fraud detection, and language translation. By utilizing the machine learning algorithms, it is possible to create personalized predictions for occurrence of a disease from a large set of data. The purpose of this study was to build a machine learning model from the Framingham Heart Study data set to predict the ten-year risk of coronary heart disease with a receiver operating characteristic area under the curve (ROC-AUC) of at least 0.70. The Framingham Heart Study is a long-term, cohort study with the purpose of determining the risk factors associated with cardiovascular diseases. We utilized a portion of these data that contained the records of 4,239 individuals. The R programming language was used to produce a series of machine learning models, including logistic regression, random forest, and support vector machine, and a stacked ensemble model composed of the previous three. The Framingham data set was split into training, validation, and testing sets. The training set was used to train the models, the validation set was used to tune the hyperparameters, and the testing set was used to assess the final performance of each model. The best performing model of the four was then selected as the final model. Each of the four models demonstrated similar performances with ROC-AUC scores ranging from 0.706 - 0.725. The logistic regression model, which was the simplest model of the four, was selected as the final model because it performed the best with a ROC-AUC of 0.725. The simplicity of this model makes it readily interpretable, thus increasing its value as a potential aid in the assessment of the risk for coronary heart disease. Our logistic regression model was then successfully incorporated into a Shiny app to make personalized predictions of the ten-year risk of coronary heart disease.

DSU Summer Research Symposium 2021

A Machine Learning Approach for Predicting the Ten-Year Risk of Coronary Heart Disease from the Framingham Heart Study Data

This annual scholarly program is designed to provide undergraduate students with a structured introduction to laboratory research within their home institution. Participants are matched with principal investigators and mentored by graduate students, enabling them to engage directly in ongoing research projects and develop foundational skills in scientific inquiry.

Indicium 2025

This annual scholarly program is designed to provide undergraduate students with a structured introduction to laboratory research within their home institution.
This year’s cohort focused on projects in the field of longevity science, exploring diverse topics such as bioinformatics, molecular biology, and cognitive science.

poster

Welcome to Delaware State University’s Annual Summer Research Symposium, and thank you for attending! Our symposium represents the culmination of students’ summer experiences in research at Delaware State University. This event showcases the work of college students working on faculty-led research projects, as well as middle and high school students experiencing research for the first time working in research groups. Whatever the case, this summer has been a unique experience for the students, and we hope a transformative one. Even for students who are not planning a career in scientific research, their summer experience has given them skills they will draw on for years to come. 
 Please join us on the mainstage and in Gather for the live event, and visit the recorded poster presentations any time.
 The student participants and organizers of the various summer programs at DSU would like to thank all of the research mentors who guided our summer program participants. There were many DSU faculty, graduate students, and other research and program staff who gave their time and energy to help our students have a fun and educational summer experience. Without their generosity and patience these summer programs would not be possible.
 For the research participants, we hope that you will maintain contact with your research mentors and colleagues. Science is a relatively small community made up of disciplinary “neighborhoods”. If you continue to work in a discipline, the other folks in that discipline will be like your neighbors that you run into regularly. Staying in contact with them will help you keep up with the news of the neighborhood, and open doors to educational and career opportunities. Whether you continue in science or take a different career path, for this summer you have been part of the world of research, and you will benefit from that perspective for the rest of your life.

Please register for the DSU Summer Research Symposium 2021 to be able to access the content.  
If you have already registered, please login with the same email you used to registered for the Conference.

Welcome to Delaware State University’s Annual Summer Research Symposium, and thank you for attending!

Human bodily functions, e.g., exhaled breath, perspiration, and physiological processes, exhibit several chemical and isotopic tracers that are potential disease signatures. Exhaled human breath analysis can be used for early disease detection and identify and discriminate disease biomarkers while being non-invasive. This project aims to develop rapid, real-time chemical and biological detection using mid-infrared laser systems. We used mid-infrared laser spectroscopy to detect the molecular absorption line transitions. In addition, we designed a long pathlength (30 m) Heriot Laser Optical Cavity to maximize the signal-to-noise ratio of experimental molecular absorption signals. This is a vital part of laser-based precision sensing. Furthermore, we also focused on a volatile organic compound and the detection of Nitrous Oxide (N2O). A rapid laser scan was performed at a frequency of 100 Hz with laser stabilized at a constant 27°C. This allowed us to estimate the current range, 50 mA to 72 mA, and the wavelength range of 4537 nm to 4540 nm of the mid-Infrared absorption from the molecular rotational-vibrational transitions. We also show that mid-infrared sensing is ideal for real-time precision detection of molecules. In addition, molecular density can be precisely estimated by the polynomial fitting of the absorption signal.

Laser-based precision sensing for biomedical applications

Dandelion (Taraxacum officinale) root extract (DRE) is known to have anti- cancer activity against breast cancer, leukemia, prostate cancer, colorectal cancer and melanoma cells (1-5). Even though this plant is sometime seen as a pesky weed in our culture, in other cultures such as Chinese, Middle Eastern, Native Americans, and East Indians, it is appreciated for its medicinal effects against cancer and other diseases ranging from indigestion, heartburn and kidney disease. Based on the reported anti-cancer cell activities of DRE, particularly on aggressive cancer cells but not on normal cells, we decided to study its activity on two breast cancer cell lines, one expressing the cell membrane receptor CD44 (i.e., MCF-7/CD44) and a control cell line negative for CD44 expression (i.e., MCF-7/pLXIN) (3, 6). The MCF-7/CD44 cell line is an aggressive multidrug resistant, very migratory and in vitro invasive while the MCF-7/pLXIN cell line is not. We hypothesize that aqueous DRE will have differential anti-cancer cell activity on these two cell lines based on their CD44 expression status. A cell viability assay (MTS) carried out after 48 hrs of DRE treatment showed that MCF-7/CD44 cells were more sensitive to DRE (when extracted in boiled PBS) compared to MCF-7/pLXIN cells. Treatment of these cell lines with non-boiled PBS extracted DRE showed no cell toxicity effect. This data validated our hypothesis and suggest that DRE might be effective at targeting CD44-expressing aggressive breast cancer cells.

Effects of Dandelion Root Extract (DRE) on CD44(+) and CD44(-) for Cell Viability

We report Raman spectroscopy measurements on samples of Ficoll polysaccharide dissolved in water at different concentrations (up to 1000 mg/ml) and collected at ambient temperature. The goal is to identify and assess Raman spectral signatures associated with the molecular structure of Ficoll PM70 and the concentration effect on these signatures. Raman spectroscopy is used to determine molecular vibrational modes, providing spectral peaks that identify the molecules (e.g Ficoll), especially in crowded and multicomponent matrices. Here, an excitation laser beam is directed onto the solution sample and interact with the structural vibrational modes of the macromolecules, resulting in energy shift of the laser photons that can be readily measured. Ficoll is accepted as a water-soluble, neutral, highly branched polymer and is commonly applied as a macromolecular crowding agent in diverse studies of protein folding and stability as well as cell volume signaling. Further, it is also used to separate blood into its components such as leukocytes and erythrocytes. In our case, we have been using Ficoll as a model crowded system to investigate the complex behavior of diverse nanoparticles. We used a HORIBA MacroRAM spectrometer to measure the Raman spectra of Ficoll solutions. In addition to the commonly reported Raman peaks of water (~1640 cm-1 HOH bend; ~3200 cm-1 OH stretching band) and possibly dissolved O2 (~1556 cm-1) and N2 (~2331 cm-1) we have identified four distinct spectral peaks (~1270 cm-1 ,~1350 cm-1,~1460 cm-1, ~2900 cm-1) likely associated with Ficoll in the 1000 to 3500 cm-1 wavenumber range, a range set to avoid possible spectral contamination from the quartz sample holder. Increase of the Ficoll concentration shows systematic increase of the intensity of the observed Ficoll peaks and systematic decrease in intensity of the two peaks associated with water, the latter effect being due likely to reduction of water content. These observations and a detailed analysis and interpretation of the measured spectral information enclosed in the Raman emission of Ficoll macromolecules may lead to a better understanding of the complexity in chemical analysis of crowded biopolymeric samples.

Raman Spectra of Concentrated Ficoll Solutions

Parkinson’s Diseases (PD) is the second most common neurodegenerative disorder and is characterized by a preferential loss of dopaminergic neurons of the substantia nigra pars compacta. The cause of the disease remains unknown; however, it is thought that environmental factors may contribute to its development. Rotenone is a pesticide that blocks mitochondrial respiration in the Electron Transport Chain by inhibiting Complex I and has been identified as an environmental risk factor for sporadic PD. Here our goal is to develop strategies that can lead to the development of potential preventative measures against PD. Dacarbazine is a potential anti-PD drug that is known for its use in chemotherapy for skin cancer and was identified in a previous screen for neuroprotective agents against the disease. We hypothesize that Dacarbazine can relieve or reverse the effects of rotenone in Drosophila. We set up a study to analyze the effect of dacarbazine against rotenone-induced neurotoxicity using locomotion behavior as a measure of neuronal toxicity. We set up four conditions: rotenone only; rotenone combined with dacarbazine; dacarbazine only; and DMSO only as vehicle only control. Here we present our results in two paradigms: a first one in which we performed a brief chronic exposure to rotenone and/or dacarbazine for three weeks and a second paradigm in which the chemical treated flies were allowed to recover on normal food for two additional weeks. Together, these studies could help determine whether dacarbazine confers protection against rotenone-induced neurotoxicity and could inform future efforts to establish a neuroprotective strategy against sporadic PD.

Ameliorating the effects of an environmental toxin in a Drosophila model of Parkinson’s Disease

Parkinson’s Disease (PD) is a common neurodegenerative disease. Although not fully understood, we do know that the alpha-synuclein protein plays a major role in causing PD. α-syn Drosophila flies overexpress the α-syn protein which it normally does not express. This allows it to be used as a model for studying Parkinson’s Disease. Gal4 is a driver that allows for the expression of α-syn. For this project, a cross of GAL4 Drosophila and α-syn Drosophila was compared to two control groups to develop a model. One control group was Gal4 crossed into wild-type background. The second control group was UAS-aSyn crossed into wild-type background. To develop the different variations, female drosophila were collected and crossed with adult male drosophila. Over a period of 8 days, the first generation offspring from the three groups were collected and divided into flies separated by sex. We ran locomotion assays using MWT (Multi-Worm Tracker) to track the movement of the flies. The males appeared to be moving at a slower rate than the females. We recorded survival, tracking the number of flies in each vial twice a week. On the 6th week of survival, there was a dramatic drop in the number of flies. Also, both female and male GAL4 x A-SYN Drosophila had similar survival rates. Once we develop a model we will use the model to run an experiment using PFF (Preformed Fibrils) to essentially test whether or not the overexpression of the α-syn gene protects the drosophila from getting Parkinson’s.

The Development of a Model to Study Genetic Parkinson’s Disease using Alpha-synuclein overexpressing Drosophila

Parkinson’s disease (PD) is the most prevalent neurodegenerative motor disease, mostly occurring in people over the ages of 65 years. Protein aggregation known as Lewy body is the hallmark for dopaminergic neuronal death in Parkinson’s disease pathology(1). PD patients initially experience motor deficits such as tremors and postural instability. As the disease progresses, patients lose complete control of physiological functions, which ultimately causes death. The current drugs available for the treatment of PD only alleviates the symptoms by boosting dopamine levels, but do not halt neurodegeneration. The main objective of this study is to evaluate the efficacy of a novel compound-6 by AptaBio as a potential therapeutic for Parkinson’s disease treatment. We analyzed whether this compound was efficacious in reducing PFF (Preformed fibrils of alpha-synuclein)-induced toxicity in N27 dopaminergic cells(3). The compound-6 not only enabled to increase cell viability and reduce cytotoxicity in vitro, but also reduced oxidative stress and minimized protein aggregation. We first tested the compound using a transgenic mouse model known as line 61(3). We demonstrate that the compound-6 successfully decreased motor deficits with stabilizing the levels of dopamine. However, we wanted to use an in vivo model that was more suitable to showcase the compound’s capability of halting neurodegeneration. Hence, we chose to use the PFF injected mouse model. Sonicated PFF was stereotaxically injected to the striatum of C57BL/6 male and female mice bilaterally (aged around 12 months). The PFF injection increases the phosphorylated alpha-synuclein and causes protein aggregation(2). After 4 to 5 months post injection, mice started showing Parkinsonian symptoms in behavior. At this stage, we gavaged the mice with the AptaBio compound-6 at two different doses of 5mg/kg and 25 mg/kg for 5 weeks. After gavaging we performed multiple behavioral tests such as hindlimb clasping, pole test, rotarod, nesting and grooming to measure whether motor function improved with the treatment as compared to PFF injected group that wasn’t treated. There was another group that was only injected with PBS, which was considered as positive controls since they do not show any motor deficits. Currently, we have finished all the behavioral tests and are finishing up data analysis for next 2 weeks. Taken together, our results are promising to show that the AptaBio compound-6 can be a novel and effective treatment for PD by halting neurodegeneration and promoting neuronal recovery.

AptaBio Compound-6 As a Potential Therapeutic for Parkinson’s disease models

Recognition of an α-helix and a complementary cleft are prevelant in many disease-relevant Protein-Protein Interactions (PPIs), and disruption of these PPIs is the focus of many drug discovery efforts. In the first part of our study, we used computational hot-spot analysis to illuminate critical residues in established PPIs involving an α-helix and a complementary cleft. In the second part of this research, we designed proteins to inhibit α-helix / complementary cleft PPIs involved in two distinct steps of the SARS-CoV-2 entry pathway. Previously, we have shown that helix-grafted proteins designed to mimic the HIV-1 gp41 C-peptide, inhibit an interaction involving HIV-1 gp41 N-peptide trimer, and gp41 C-peptide, resulting in the inhibition of HIV-1 entry in a live virus assay. Here, we used helix-grafted display to create novel proteins designed to inhibit two distinct steps in the SARS-CoV-2 entry pathway, which both rely on PPIs involving an α-helix and a complementary cleft. The design of these proteins, t

Targeting α-Helix-Dependent Protein-Protein Interactions In Two Steps of the SARS-CoV-2 Entry Pathway

COVID-19 is a new strain of Coronavirus that had not been discovered in humans previously. It is a disease brought about by SARS-CoV-2 that can trigger a respiratory tract infection. The pandemic has affected people’s lives negatively across the U.S. It was suggested that racial/ethnic minorities were mostly affected by the outburst of this pandemic according to data from the survey people took. The pandemic has affected employment according to the finding from the data filled out by the individuals. More people have lost their jobs because it was not considered essential. As a result, the poverty rate went up. This research will be carried out with community partner organizations, who will recruit participants from communities to take a structured survey and serology test. The purpose of this research is to explore the economic impact of COVID-19 in Delaware’s underserved communities. Specifically, this study will investigate unemployment and household income during the COVID-19 pandemic. The study will also examine whether there are differentials in unemployment and household income by race, gender, and educational level. The significance of this research is to understand the economic impact of the COVID-19 pandemic in underserved communities and bring awareness.

Economic impact of COVID-19 in Delware underserved communities.

Different studies show that TDP-43 plays a major role in association with other neurological diseases, however, we want to test why TDP-43 is a major factor. We injected a mouse with TDP-43 and injected another mouse with Alzheimer's Disease. The two mice were watched for more than over a month now. We used the mice's pellets to conduct the different tests we ran. We sonicatied the different mice pellets for a 1 min 20 minutes with a 5 sec pulse and 5 sec off cycle. Once the sonication was done, we transferred the sonicated pellets to the centrifuge machine which ran for about 35 minutes. Other graduate students began to compare results from these batches of pellets to the others that they tested to see the difference or similarities. These results showed us that the Alzheimer's disease mouse model that we developed over expresses pathological tau and TDP-43.

TDP-43 Mice Model

Alzheimer’s disease (AD) is the most common form of dementia, accounting for 70 to 80% of all cases. The greatest risk factor for AD is aging; an estimated 30 million people are living with AD and over the next 30 years it is projected to increase to over 130 million worldwide. In addition, one in three seniors have died of AD or another dementing disease, and since 2000, deaths from Alzheimer’s disease have increased 89%. TDP-43 is the major pathological protein in motor neuron disease and frontotemporal dementia. Previously, TDP-43 pathology has been reported that in up to 50% of late-onset, sporadic cases of Alzheimer’s disease, the TDP-43 protein has been reported as an age-related co-pathology linked to dementia severity. TDP-43 proteinopathy in those over 80 has recently been recognized as an independent disease entity, limbic- predominant age-related TDP-43 encephalopathy (LATE). There are over one million people living with LATE in the United States. As LATE shares some of the clinical features of dementia seen in AD in addition to co-existing with AD, understanding mechanisms leading to the pathogenesis of LATE and co-morbid TDP-43 pathology could lead to the development of targeted therapeutics. We have developed an inducible cell-specific pathological model by introducing three familial ALS mutations in the human TDP-43 gene(3XTDP43). Utilizing the tetracycline inducible system, we drove 3XTDP43 specifically in the hippocampus (Camk2a-tTA). Ten-month old mice were used in this study where we performed fMRI, histology, Western blot, and solubility fractionations. Results show phosphorylated TDP-43 pathology and a distinct decrease in solubility. In addition, 3XTDP43 overexpression increases phosphorylation of endogenous tau which may indicate aggregation. Though the mice were imaged (fMRI) we have not evaluated those results as of yet. This model may provide a better understanding of the pathogenesis of TDP-43 proteinopathies.

Downloads

Next from DSU Summer Research Symposium 2021

Laser-based precision sensing for biomedical applications

Similar lecture

Evaluating Supervised Machine Learning Models for Heart Disease Prediction