Research Projects Directory

Scientific Approaches

We will create at least one cohort of patients with hypertension based on a data type indicative of hypertension. We will then compare the count and demographics of the patients in this cohort with other cohorts of patients with hypertension created with different data types indicative of hypertension than the one we used.

Scientific Approaches

I plan to use survival analysis for this study. Additionally, we want to study how to use data over time to look at chronic disease.

Scientific Approaches

We will primarily use All of Us as a validation dataset for machine learning models trained using data from (1) a large New York hospital system and (2) the UK Biobank. Our models are trained using gradient boosting methods and include demographics, common laboratory measurements, vitals, past diagnoses, and medication usage. Model performance is evaluated using both sensitivity-specificity and precision-recall metrics. We will also specifically examine disparities in performance between races and other demographic subsets.

Scientific Approaches

We plan on using traditional Cox regression and adding time-dependent covariates as becomes appropriate. We have conducted a preliminary analysis using local EHR data and are hoping to replicate that study in AOU.

Scientific Approaches

This is a retrospective observational research
1. Cohorts: 1) relatively healthy elders, check with age groups [ 70-80)[80,90),[90+) 2) elders with serious disease, as cancer, sever chronic diseases
2. Extract Social Determinants of Health elements for these two cohorts
3. Using statistics description, statistical inference, and Machine learning algorithms to identify the elements that will influence elders health.
4. Analytic tool: AoU workbench, jupyter notebook-python environment.

Scientific Approaches

We plan to use the All of Us database to examine the correlation between a participant's PTC haplotype and their BMI. In order to do so, we will use whole genome sequencing information to look for specific mutations in the PTC gene that affect tasting ability. We will use an R notebook and Python to extract and analyze the genetic and medical information of the cohort. Our cohort will only consist of individuals ages 18-27 with no further restrictions placed on the composition of the cohort.

Scientific Approaches

Initial focus Project focus area: The activity monitoring, Heart rate (HR) measurements during rest and exercise using Fitbit data and its relation with EHR.

Research focus :
Activity monitoring
Cardiovascular disease.
Obesity and medical condition
Population Health
Social / Behavioral

Scientific Approaches

The scientific approaches used for this study is to look at certain datasets ranging from ages 28-37 in order to investigate to see if certain individuals are either super-tasters, tasters, or non-tasters. Also to see how BMI is affected if an individual is a super-taster, taster, or non-taster. The R-program and Python will be used to generate graphs.

Scientific Approaches

We are a subgroup of a much larger team working on this project. Specifically, we are tasked with researching individuals aged 38-47. We plan to utilize and credit the All of Us Research Database's file on individuals aged 38-47 on whether or not they are PTC tasters and their BMI. Additionally, in our class we extracted DNA samples, ran PCR, and gel electrophoresis to determine whether we were tasters/non-tasters,

Scientific Approaches

Data on a cohort of racial-ethnic minority patients will be collected from the All of Us COPE survey database and reviewed for baseline and sociodemographic information. Participants eligible for inclusion in this study are those who self-report their race or ethnicity as Hispanic and/or nonwhite. The period of time from which data will be collected will be May 2020 through July 2020. Medication misuse will be defined as affirmative responses to the query “did you use prescription opioids/stimulants/sedatives in any way a doctor did not direct you to use it?”. Psychological distress will be classified using questions assessing mood and anxiety based on the Patient Health Questionnaire (PHQ-9) depression module and the Generalized anxiety disorder (GAD-7) mental health rating scales. Both screening tools are based on the DSM-IV criteria for major depression and generalized anxiety disorder and have well-documented reliability and validity in literature. (8,9)

Scientific Approaches

Utilize survey data from adults with chronic conditions to determine symptom clusters. Use Cohort Builder to identify eligible adults diagnosed with certain conditions. Use participant responses to survey questions to generate symptom phenotypes and analyze potential predictors of symptomatic phenotypes.

Scientific Approaches

Algorithm 1:
https://phekb.org/phenotype/rheumatoid-arthritis-ra
This algorithm uses ICD9, ICD10, and laboratory. Use the code that the authors provided to extract the data

Algorithm 2:
https://phekb.org/phenotype/rheumatoid-arthritis-demonstration-project
This algorithm uses ICD9, medications and natural language processing. You just need to use the ICD codes and medications only.

Scientific Approaches

We will use Deep Deep Reinforcement Learning with subjects in the AllOfUs dataset who have urine culture reports while being hospitalized. We will test a locally developed algorithm with AllOfUs data.

Scientific Approaches

Not applicable - these notebooks demonstrate example analysis how to use Hail and PLINK to perform genome-wide association studies using the All of Us genomic data and phenotypic data.

Scientific Approaches

For Educational purpose to show best practices when using jupyter notebooks for data access, storage, data manipulations - transformations, conversions, cleaning, optimization and other research support related issues that is useful for multiple AoU researchers.

Scientific Approaches

Not applicable - these notebooks demonstrate example analysis how to use Hail and PLINK to perform genome-wide association studies using the All of Us genomic data and phenotypic data.

Scientific Approaches

To generate PRSs for diverse ancestries, we will use data from the eMERGE consortium, Million Veteran’s Program (MVP), the All of Us (AoU) program, dbGaP, PRIMED consortium sites, the UK Biobank, and collaborations with several international groups representing Middle Eastern, South Asian, and East Asian cohorts. Our specific aims are: Aim 1. Integrate and harmonize data from heterogeneous sources to enable cross platform phenotyping and generation of PRSs for common diseases in diverse ancestry groups. Aim 2. Develop PRSs for CHD and its major risk factors (hypertension, diabetes, obesity, hypercholesterolemia) in populations of diverse ancestry. Aim 3. Develop novel statistical and computational methods to account for diverse genetic ancestry and admixture in models of polygenic risk. Aim 4. Develop ‘clinic ready’ PRSs for diverse ancestry groups by creating reference distributions of a CHD PRS and integrate it with clinical information to compute absolute risk estimates.

Scientific Approaches

Using the ALL of US full cohort, we can help determine the associations of some of these factors which would have been modifiable. Such modifiable risk factors would include: HIV without any or consistent HIV (ART) treatment, use of prednisone or common immunosuppressive drugs before the onset of TB (note TB treatment can include prednisone which should not be considered a risk), poorly controlled diabetes (stratified by hemoglobin a1c or in those insulin dependent without Hga1c recorded), known latent TB (quantiferon, T-spot, or Mantoux positivity) without latent TB treatment. We will also assess the timeframe of who has been assessed for latent TB before or with TB diagnosis and how many are not treated for latent TB, especially in those with another risk factor.

Scientific Approaches

Dataset: All of Us dataset
Research methods: General Linear Model (GLM) statistical test to assess the data. Additional statistical tests will be used based on the GLM. We will conduct our analysis in R coding language.

Scientific Approaches

We will use the cohort builder to find US participants aged 18 years and older who have been diagnosed with chronic diastolic or systolic heart failure. We will use R to retrieve heart failure medications, antihypertensive medications, and diabetes medications. For each type of heart failure, we will categorize the sample into 4 groups: optimized group (those who are on all four recommended medications), 3 reference groups (those who are on 3 medications vs. 2 medications vs. 1 medication). We will use the Chi-Square test to compare the extent of adherence to recommended medications to demographic characteristics. Multiple logistic regression will be used to examine the association of optimization of guideline directed therapy for diastolic and systolic heart failure and the SDOH mentioned above. One limitation is that we can only include a specific group of participants with diagnoses that belong to either chronic diastolic or systolic heart failure, so our population is limited.

Scientific Approaches

Our methodology consists of compiling and analyzing the relevant data on the potential genetic, lifestyle, and social/environmental risk factors associated with WKS. We hope to pull raw genomic and environmental data from All of Us participants with WKS (or perhaps, a high risk of developing the disease based on family history or known health behaviors) to ascertain any outstanding genetic markers and measure the prevalence of certain social and behavioral determinants (including low income/socioeconomic status, employment, frequent alcohol use, poor diet/nutrition, etc.). Our project is purely correlative and will provide insight into how WKS may present as a complex disease, which is crucial to recognize to effectively devise WKS prevention programs.

Scientific Approaches

For this study, we will descriptive and inferential statistical methods to explore potential patterns around engagement in the All of Us Research program among the presently diverse cohort of participants of African ancestry or descent.

Scientific Approaches

We will study adults with a history of cancer who answered the health care access and utilization survey. We will describe the cohort's demographic characteristics and use propensity matching to identify similar adults without a prior history of cancer. We will compare the prevalence of delayed/forgone care among these groups. We will apply generalized linear models to identify risk factors associated with delays in care due to patient-clinician identity discordance.

Scientific Approaches

We hope to collect survey data from participants with specific dermatologic diseases. Multivariable regression will be used to analyze the relationship between race, ethnicity, and barriers to care.

Scientific Approaches

In this workspace, we are seeking to understand how the data are structured, approaches to setting up cohorts, concepts, and datasets for analysis. The objective of this workspace is to become more familiar with the All of Us dataset so we can conduct a more in-depth analysis examining the association between access to healthcare services and social determinants of health.