Ishita Pethkar - Statistics PhD Candidate
Statistics PhD Candidate at the University of North Carolina at Chapel Hill (2024 - Expected 2029), working in Biomedical Data Science, Statistical Modeling and Machine Learning. Research Assistant at UNC STOR and SIDS, advised by Prof. Steve Marron and Prof. Daniel Kessler. Instructor and Teaching Assistant at UNC Chapel Hill. B.Sc. in Mathematics and Computer Science from the Chennai Mathematical Institute.
Research Experience
Multi-Organ Biomedical Imaging Analysis, UK Biobank
Characterized brain-heart structural variation across ~8,000 UK Biobank participants by analyzing hundreds of T1-weighted MRI-derived features. Identified shared brain-heart modes and an age-dependent joint axis using DIVAS, revealing increased inter-individual variation around ages 60-63. Characterized the evolution of brain-heart relationships across the adult lifespan using subspace-aware dimensionality reduction, hierarchical clustering, and temporal network analysis.
Shape-Based Data Integration for Neuroimaging
Extended DIVAS beyond Euclidean data to integrate 3D anatomical shape representations, enabling joint analysis of brain shape variation. Developed an end-to-end shape-analysis pipeline with s-rep-based anatomical correspondence and principal nested spheres (PNS). Implemented and validated the computational pipeline in Python.
AngioNET: Reproducible Retinal Vessel Segmentation
Achieved ~0.86 Dice, ~98% accuracy, ~91% sensitivity, and ~99% specificity for retinal vessel segmentation with a reproducible U-Net pipeline. Accelerated training with 512x512 overlapping patches, vessel-content filtering, and FP16 mixed-precision training on CUDA. Quantified retinal vessel-network structure via skeletonization and graph representations.
Industry Experience
Data Analyst at LabCorp (January 2026 - May 2026)
Achieved 90% demand-forecasting accuracy for a newly launched chronic urticaria diagnostic test by modeling ordering patterns. Identified distinct demand patterns by segmenting new-account adoption and recurring utilization. Designed an adaptive forecasting workflow incorporating newly observed ordering data.
Technical Skills
Programming and Computing: Python, R, SQL, MATLAB, Git, SLURM, HPC
Data Science and ML Libraries: pandas, NumPy, scikit-learn, PyTorch, TensorFlow, Keras, Matplotlib, ggplot2
Statistical Methods: Statistical inference, regression, multivariate analysis, high-dimensional data analysis, statistical shape analysis, subspace analysis, clustering, dimensionality reduction, forecasting
Education
PhD in Statistics - University of North Carolina at Chapel Hill (2024 - Expected 2029)
B.Sc. in Mathematics and Computer Science - Chennai Mathematical Institute (2021 - 2024)
Relevant Coursework: Statistical Computing for Data Science, Applied Statistics, Object-Oriented Data Analysis, Statistics for Shape Data