QuantMed℠ Analytics

ABOUT

Photo by Logan Voss on Unsplash

Hello, and welcome. My name is Aaron Tien. I am a statistical consultant specializing in biostatistics and clinical research. I hold both a Bachelor of Science and a Master of Arts in Statistics from the University of Pittsburgh. Over the past several years, I have collaborated with clinical researchers from the University of Pittsburgh Medical Center (UPMC) on study design, statistical modeling, and publication-ready analyses. In addition, I have contributed to manuscript preparation and supported wet-lab research, providing experience across both the computational and experimental aspects of biomedical research. I have also served as a student consultant at the University of Pittsburgh Statistics Consulting Center.

My research and consulting experience spans genetics, congenital heart disease, single-cell and bulk RNA sequencing, clinical research, survey research, and machine learning. I specialize in developing reproducible analytical workflows using R and Python, enabling rigorous statistical analyses and the clear communication of results through effective data visualization and reporting.

In addition to my core expertise, I have experience with natural language processing, Bayesian statistics, stochastic processes, differential privacy, longitudinal and clustered data analysis, and more. My technical toolkit also includes SQL, HTML/CSS, JavaScript, Java, MATLAB, Tableau, Microsoft Office, and Google Workspace, allowing me to support projects from data acquisition and management through analysis and presentation.

SELECTED PROJECTS

Photo by Luke Jones on Unsplash

Single-Cell RNA Analysis of LRP1-Associated Mechanisms in Congenital Heart Disease

Objective

Investigating how a missense mutation in the low-density lipoprotein receptor-related protein 1 (LRP1) contributes to congenital heart disease.

Methods

Cell Ranger output files were processed and subjected to quality control and exploratory analysis using the Seurat package. Dimensionality reduction was performed using principal component analysis (PCA), and cells were clustered using the shared nearest neighbor (SNN) algorithm. Cell populations were visualized using UMAP and t-SNE. Cell types were then assigned to clusters based on the expression of established marker genes. Differential gene expression and pathway enrichment analyses were conducted to identify upregulated and downregulated biological pathways in mutant cells across cell types. Finally, CellChat analysis was performed to investigate differences in intercellular and cell-type-specific communication patterns between mutant and wild-type cells.

Outcome

Many differentially expressed genes were associated with mitochondrial function, suggesting that LRP1 may affect mitochondrial processes. Additionally, vascular smooth muscle cells (VSMCs) appeared to be particularly affected by LRP1 deficiency, which may contribute to the structural cardiovascular defects observed. These findings raise new questions about the cellular mechanisms involved and how mitochondrial dysfunction in VSMCs could lead to developmental abnormalities. Wet-lab experiments will be necessary to validate and further explore these findings.

Genetic Variation and Association with Post-Operative Outcomes for Neonates and Infants in the Cardiac Intensive Care Unit

Objective

Congenital heart disease (CHD) affects up to 1% of live births, and an estimated 33%–46% of affected infants have an underlying genetic disorder. Although genetic testing has become an important component of clinical care for children with CHD, the growing volume of genetic information presents challenges in determining its clinical and prognostic utility. Large copy number variants (CNVs) have been associated with long-term outcomes, including impaired transplant-free survival and neurocognitive outcomes; however, less is known about their relationship with immediate clinical care and short-term postoperative outcomes. This study investigated whether abnormal chromosomal microarray results in neonates and infants with CHD were associated with worse postoperative outcomes in the pediatric cardiac intensive care unit (PCICU).

Methods

Categorical outcomes were compared between infants with normal and abnormal microarray results using Pearson's chi-squared tests, with Yates' continuity correction applied when appropriate. Continuous outcomes were compared using Wilcoxon rank-sum tests because the data did not meet the assumptions required for parametric methods. Descriptive analyses were conducted to assess the distribution of study variables and identify missing data. To evaluate differences in clinical outcomes between microarray groups while accounting for differences in case complexity, multivariable logistic regression models were used for categorical outcomes and linear regression models were used for continuous outcomes, with microarray group and STAT category included as predictors. A more stringent significance threshold of p ≤ 0.01 was used to reduce the potential for false-positive findings due to multiple comparisons.

Outcome

The two groups had similar distributions of gestational age and birth weight but differed significantly in fundamental diagnoses and primary procedures. Across the full cohort, STS score, deep hypothermic circulatory arrest (DHCA) time, and cardiopulmonary bypass time were significantly lower among infants with abnormal microarray results, while abnormal microarray results were associated with a higher incidence of lifetime gastrostomy tube placement. However, subgroup analyses of infants who underwent cardiac surgery within 30 days of life did not identify significant differences in surgical or PCICU outcomes between microarray groups after accounting for differences in underlying diagnoses and procedures.

Overall, the findings suggest that chromosomal microarray results have limited utility for predicting immediate postoperative outcomes in infants with CHD. However, the results highlight the importance of including children with copy number variants in future CHD research to better understand their long-term clinical significance.

Full Publication: View the full paper

Cardiac ICU Neuromonitoring in Infants with CHD Leads to Early Arterial Ischaemic Stroke Recognition: A Single-Center Experience

Objective

Evaluate the utility of comprehensive neuromonitoring to allow for early identification of arterial ischaemic strokes in high-risk critically ill infants with CHD.

Methods

Statistical analyses were performed to compare clinical and demographic characteristics between neonatal and control groups. The Shapiro–Wilk test was used to assess the normality of continuous variables. Because the numerical variables demonstrated significant departures from normality and the sample sizes were insufficient to rely on asymptotic assumptions, two-sided Mann–Whitney U tests were used to compare continuous variables between groups. Distributions were visualized using side-by-side boxplots to support interpretation of the statistical findings. Categorical variables were compared using chi-squared tests, with Yates' continuity correction applied when necessary. Stacked bar charts and contingency tables were used to visualize categorical data and evaluate group distributions. A significance threshold of p < 0.01 was used to account for the potential impact of multiple comparisons.

Outcome

Comprehensive neuromonitoring in high-risk critically ill CHD patients led to the identification of arterial ischaemic strokes, even in the context of significant hemodynamic lability and limited neurological examination secondary to sedation and neuromuscular blockade. Head ultrasound was useful as an initial screening modality, with advanced imaging used to confirm an injury or in cases of high clinical suspicion.

Full Publication: View the full paper

Services

Image by Markus Spiske on Unsplash

Contact

Thank you for your interest in my statistical consulting services. If you would like to discuss a potential collaboration or request assistance with a project, please contact me using the information below.

Email:
atnst2001@gmail.com

LinkedIn

Elements

i = 0;

while (!deck.isInOrder()) {
    print 'Iteration ' + i;
    deck.shuffle();
    i++;
}

print 'It took ' + i + ' iterations to sort the deck.';