Serdar Bozdag
Mid-career Tenured Researcher, University of North Texas
4 active projects
Disease Risk Prediction
Scientific Questions Being Studied
Predicting disease diagnosis and susceptibility to a disease is a challenging and important task. Several diseases could be prevented by careful interventions such as changes in diet, lifestyle, and other forms of preventive care, which is also called preventive medicine. Developments in preventive medicine would help reduce the burden on health care as much as $45 billion per year based on a 2010 report. The objective of this project is to develop a machine learning model that predicts future disease risk of individuals by utilizing the large biomedical datasets available in the All of Us (AoU) Research Program database.
Here, we aim to build a computational method to integrate multiple data modalities available in the AoU Research Program database to predict current and future diagnoses of individuals.
Project Purpose(s)
- Methods Development
- Ancestry
Scientific Approaches
We will build a machine learning model utilizing different data modalities in AoU database (e.g., FitBit, Physical measurement, Survey, genetic, etc.). We will generate an AoU workspace with a large and diverse cohort by collecting relevant available features for all individuals from all demographic populations based on factors such as age, gender, sex at birth, race and ethnicity.
To detect outlier values in the dataset, we will employ an outlier detection approach for each feature and mark the extreme values as missing. We will filter out the features completely if they have missing values for a large proportion of the cohort. For other features, we will employ data imputation methods.
To ensure that the values are in similar scale, we will normalize each feature. We will employ feature selection methods to select the most relevant features.
We will train a machine learning model and evaluate it on a held out dataset.
Anticipated Findings
This research will allow us to interpret existing biomedical datasets and decipher their contributions as risk factors for various diseases in different demographic populations. Understanding what lifestyle and biological data points could be associated with certain diseases will help researchers and clinicians better understand the etiology of diseases. By integrating genomic and life-style related datasets, the research will also shed more light on the interaction between genetic and environmental factors.
Demographic Categories of Interest
This study will not center on underrepresented populations.
Data Set Used
Controlled TierResearch Team
Owner:
- Serdar Bozdag - Mid-career Tenured Researcher, University of North Texas
Collaborators:
- Yashu Vashishath - Graduate Trainee, University of North Texas
- Ethan Rebello - Undergraduate Student, University of North Texas
Type 2 Diabetes Subtyping
Scientific Questions Being Studied
The effects of different blood components on glucose levels in hyperglycemic patients. We want to create a predictive model which can estimate change in level of glucose by changing blood component concentration, such as RBC counts, electrolyte concentration, etc. , in Type 2 diabetes patients.
Project Purpose(s)
- Disease Focused Research (hyperglycemia)
Scientific Approaches
We will analyze the lab results of 150,000 plus individuals in the All Of Us Research workbench. We will try to find a pattern between glucose levels compared to blood components levels in normal and diabetic patients in Python and determine which blood component is associated with change in blood glucose levels. Once we determine these components, we want to create a predictive model which will determine which blood component can assist in blood glucose maintenance.
Anticipated Findings
The results of our study, once published, will assist doctors in making better decisions to regulate glucose levels with drugs, dietary, and lifestyle changes as well as if any medication changes levels of any blood components, our model can predict how it will affect glucose levels in the patients.
Demographic Categories of Interest
This study will not center on underrepresented populations.
Data Set Used
Registered TierResearch Team
Owner:
- Yashu Vashishath - Graduate Trainee, University of North Texas
- Serdar Bozdag - Mid-career Tenured Researcher, University of North Texas
- Islam Ebeid - Research Fellow, University of North Texas
- Mohammad Al Olaimat - Graduate Trainee, University of North Texas
Tier 5 - Type 2 Diabetes Subtyping
Scientific Questions Being Studied
The effects of different blood components on glucose levels in hyperglycemic patients. We want to create a predictive model which can estimate change in level of glucose by changing blood component concentration, such as RBC counts, electrolyte concentration, etc. , in Type 2 diabetes patients.
Project Purpose(s)
- Disease Focused Research (hyperglycemia)
Scientific Approaches
We will analyze the lab results of 150,000 plus individuals in the All Of Us Research workbench. We will try to find a pattern between glucose levels compared to blood components levels in normal and diabetic patients in Python and determine which blood component is associated with change in blood glucose levels. Once we determine these components, we want to create a predictive model which will determine which blood component can assist in blood glucose maintenance.
Anticipated Findings
The results of our study, once published, will assist doctors in making better decisions to regulate glucose levels with drugs, dietary, and lifestyle changes as well as if any medication changes levels of any blood components, our model can predict how it will affect glucose levels in the patients.
Demographic Categories of Interest
This study will not center on underrepresented populations.
Data Set Used
Registered TierResearch Team
Owner:
- Yashu Vashishath - Graduate Trainee, University of North Texas
- Serdar Bozdag - Mid-career Tenured Researcher, University of North Texas
- Sarah Beaver - Graduate Trainee, University of North Texas
- Islam Ebeid - Research Fellow, University of North Texas
- Mohammad Al Olaimat - Graduate Trainee, University of North Texas
A1C vs RBC count analysis
Scientific Questions Being Studied
We want to explore data to check if changes in RBC count or Hemoglobin conc. have any effect on A1C in normal as well as diabetic patients. The analysis will help understand if RBC count and Hb conc. can play a role in miss diagnosis of Diabetic patients because of A1C levels.
Project Purpose(s)
- Educational
Scientific Approaches
We are trying to check if the hypothesis is true or not before we conduct any wet lab experiments to confirm our hypothesis.
Anticipated Findings
We anticipate that as the RBC count and Hb conc. increases, A1C levels would go down.
Demographic Categories of Interest
This study will not center on underrepresented populations.
Data Set Used
Registered TierResearch Team
Owner:
- Yashu Vashishath - Graduate Trainee, University of North Texas
- Serdar Bozdag - Mid-career Tenured Researcher, University of North Texas
You can request that the All of Us Resource Access Board (RAB) review a research purpose description if you have concerns that this research project may stigmatize All of Us participants or violate the Data User Code of Conduct in some other way. To request a review, you must fill in a form, which you can access by selecting ‘request a review’ below.