Dayanjan Wijesinghe

Early Career Tenure-track Researcher, Virginia Commonwealth University

3 active projects

Predicting VOC in SCD

Vaso occlusive crisis (VOC) is a major cause of hospitalization for sickle cell disease (SCD) patients. An ability to predict the onset of a VOC event has the potential to enable proactive interventions to minimize/prevent event or event severity. The…

Scientific Questions Being Studied

Vaso occlusive crisis (VOC) is a major cause of hospitalization for sickle cell disease (SCD) patients. An ability to predict the onset of a VOC event has the potential to enable proactive interventions to minimize/prevent event or event severity. The question we are asking is, can you predict a VOC event by a combination of patients medical data?

Project Purpose(s)

  • Disease Focused Research (Sickle Cell Disease)

Scientific Approaches

We will initially begin by identifying SCD patients who have had at least a single VOC event. This will be our patient cohort. We will then employ a combination of classical statistical approaches as well as machine learning approaches to identify possible predictors of VOC events

Anticipated Findings

We hypothesize that a combination of EMR and other data will enable us to identify those patient that are hospitalized for a VOC events within one month. An algorithm that enables us to make this prediction with greatly help in the better management of SCD patients

Demographic Categories of Interest

  • Race / Ethnicity
  • Age

Research Team

Owner:

Collaborators:

  • Suad Alshammari - Graduate Trainee, Virginia Commonwealth University
  • Silas Contaifer - Graduate Trainee, Virginia Commonwealth University
  • Joshua Morriss - Graduate Trainee, Virginia Commonwealth University
  • Evan French - Project Personnel, Virginia Commonwealth University
  • VIRGINIA UNIVERSITY - Graduate Trainee, Virginia Commonwealth University
  • Daniel Contaifer Junior - Project Personnel, Virginia Commonwealth University

Predicting Major Adverse Cardiac Events in Heart Failure Patients with COVID-19

Aim 1: Determine the predictors of mortality and hospitalization for patients with acute or chronic heart failure (A/CHF) that had a diagnosis of COVID-19. Rapid onset of new or worsening heart failure symptoms are characteristic of AHF. Concomitance of COVID-19…

Scientific Questions Being Studied

Aim 1: Determine the predictors of mortality and hospitalization for patients with acute or chronic heart failure (A/CHF) that had a diagnosis of COVID-19. Rapid onset of new or worsening heart failure symptoms are characteristic of AHF. Concomitance of COVID-19 presents additional challenges towards treating A/CHF patients. Studies provide several candidate clinical and laboratory measures associated with worse clinical outcomes for patients with A/CHF and COVID-19. Identifying COVID-19 specific predictors of mortality and hospitalization for A/CHF patients would help explain the pathophysiology behind the progression of COVID-19 in A/CHF patients.

Aim 2: Stratify the risk for suboptimal guideline-directed medical therapy (GDMT) for A/CHF patients with COVID-19. COVID-19 obstructs A/CHF patients from reaching their optimal target doses. Assigning patients into different strata at risk of not achieving optimal GDMT targets may provide clinicians with more impactful treatment options.

Project Purpose(s)

  • Disease Focused Research (severe acute respiratory syndrome, acute on chronic heart failure)

Scientific Approaches

This retrospective study will include demographic characteristics and clinical features from the All of Us A/CHF and COVID-19 combined cohorts. Missing values will be imputed by multiple imputation. Dimensionality of the data will be reduced by supervised selection. Associations between demographic and clinical features will be made with the outcome of 1-year re-hospitalization with A/CHF as the primary diagnosis. Models generated will utilize standard regression, random forests, and gradient boosting, and will be evaluated by their predictive values, sensitivity, specificity, and c-statistics.

Combined clinical features at baseline will undergo k-means cluster analysis to subset groups. Features will undergo processing as described above. A predictive model will be developed, and a Cox proportional hazards regression analysis for re-hospitalization will be performed for each subgroup. All analyses are to be conducted on the All of Us workbench in the latest version of R and Python.

Anticipated Findings

We may expect to find clinical features and laboratory parameters associated with elevated systemic inflammation, endothelial dysfunction, and hypercoagulation to be strong predictors of adverse outcomes for A/CHF patients who has contracted COVID-19. Clinical features like carbon dioxide and oxygen partial pressures in arterial blood may serve as correlates of worse outcomes. Predictive laboratory features may include high-sensitivity C-reactive protein (hs-CRP), brain and atrial natriuretic peptides (BNP/ANP), ferritin, interleukins, neutrophils, complete blood count and d-dimer quantities among others.

In stratifying patients at-risk of not adhering to GDMT, stratification we expect that data pertaining to a patient’s health care access and utilization, as well as the severity of their COVID-19 infection, may put them at greater risk of non-adherence. Severity of COVID-19 infection may be understood as a profile of high inflammation like elevated levels of hs-CRP or interleukins.

Demographic Categories of Interest

This study will not center on underrepresented populations.

Research Team

Owner:

Collaborators:

  • Suad Alshammari - Graduate Trainee, Virginia Commonwealth University
  • Silas Contaifer - Graduate Trainee, Virginia Commonwealth University
  • Daniel Contaifer Junior - Project Personnel, Virginia Commonwealth University
  • VIRGINIA UNIVERSITY - Graduate Trainee, Virginia Commonwealth University
  • Kevin Ledezma - Graduate Trainee, Boston University

Improving the estimation of CVD risk among CKD patients

It is well known that there is a strong association between Chronic Kidney Disease (CKD), and Cardiovascular Diseases (CVD). Although there are several CVD prediction scores available to predict the probability of CVD in the general population, none of them…

Scientific Questions Being Studied

It is well known that there is a strong association between Chronic Kidney Disease (CKD), and Cardiovascular Diseases (CVD). Although there are several CVD prediction scores available to predict the probability of CVD in the general population, none of them are accurate enough to estimate the extent of CVD risk in CKD patients. For example, the Framingham score has been questioned as it was created based on a homogenous, geographically restricted and predominantly white population. Additionally, there are other factors (non-traditional risk factors of CVD that are specific to CKD such as albuminuria, anemia, fluid overload etc.) which are not included in the Framingham score which may play an essential role in estimating ischemic heart disease in patients with CKD. As a result, it may overestimate the risk and have poor discriminatory power in the CKD population. All these indicate the need for the development of more appropriate risk factors to assess the CVD risk among CKD patients.

Project Purpose(s)

  • Disease Focused Research (chronic kidney failure)

Scientific Approaches

• A retrospective cohort study will be conducted to create a model capable of predicting CVD using traditional and non-traditional risk factors of CVD in patients with CKD. The study will include CVD patients with CKD: age 18-80, CKD stage 1-5(including patients on dialysis), in accordance with the Kidney Disease Outcomes Quality Initiative (KDOQI).
• The baseline data collection will include but not limited to age, sex, CKD stage, CVD’s history, medications history as well as traditional CVD risk factors.
• Baseline renal functions and other parameters related to CVD will be assessed and those include: glomerular filtration rate (GFR), serum creatinine, serum cystatin C, low-density cholesterol (LDL), high-density cholesterol (HDL), total cholesterol and high-sensitivity C-reactive protein (CRP).
• To investigate the association between CVD in CKD patients and potential risk factors, Cox Proportional-Hazards model will be used for the statistical analysis.

Anticipated Findings

• Creating a model using traditional risk factors (hypertension, diabetes at el) and non-traditional risk factors (albuminuria, anemia at el) can estimate the risk of CVD in patients with CKD. This model would help to adequately treat CVD and prevent worsening of CVD in patients with CKD

Demographic Categories of Interest

  • Race / Ethnicity
  • Age

Research Team

Owner:

Collaborators:

  • Suad Alshammari - Graduate Trainee, Virginia Commonwealth University
  • Joshua Morriss - Graduate Trainee, Virginia Commonwealth University
  • Daniel Contaifer Junior - Project Personnel, Virginia Commonwealth University
  • VIRGINIA UNIVERSITY - Graduate Trainee, Virginia Commonwealth University
1 - 3 of 3
<
>
Request a Review of this Research Project

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.