Karoline Kallis

Applied Research Scientist · Medical Imaging & Clinical Data

Karoline Kallis, PhD

I apply machine learning and computational methods to real clinical problems, with a focus on medical imaging, image and data processing, treatment-planning automation, and large-scale EHR analysis. I develop and validate models using clinical imaging and EHR data and evaluate their potential for integration into clinical workflows.

Image & data processing Quantitative MRI Medical imaging Image registration Radiation treatment-planning automation DICOM workflows ML model development EHR data science Cross-functional collaboration
Resume Google Scholar GitHub LinkedIn
8+
Years in healthcare data science
30
Peer-reviewed publications
5+
Programming languages
14+
Conference presentations

Research highlights

Quality measure trajectories by subgroup EHR ICD CPT Rx LOINC Cohort QI Risk QI 3y
EHR · Epidemiology

Longitudinal Trends in Guideline-Concordant Care After Peripheral Artery Disease Diagnosis

Tracked four cardiovascular disease prevention measures in 1.2 million patients with peripheral artery disease and examined how care changed over time and across patient groups.

Epic COSMOS SQL Python Mixed effects models Cox proportional hazards
Key resultOnly 8.9% achieved all four quality measures in the first year after diagnosis.
PAD detection model ROC curve EHR ICD CPT Rx LOINC AUROC Fairness Detection ROC
EHR · Machine Learning

Peripheral artery disease (PAD) detection model for deployment at University of California San Diego

Preparing an explainable PAD detection model for Epic integration, with patient flagging tied to upcoming appointments.

XGBoost Epic Caboodle AWS SageMaker SQL Python Clinical deployment
Current stageInference pipeline and monitoring are under development; a prospective trial is planned subject to IRB review.
Automated brachytherapy treatment planning workflow 3D dose prediction optimize COBYLA DICOM treatment plan
Radiotherapy Data · Treatment Automation

Automated brachytherapy treatment planning

Built the dose-to-plan layer that converted predicted 3D dose into deliverable cervical cancer brachytherapy plans and tested it against clinical plans.

Knowledge-based dose prediction Python DICOM
Key resultConverted predicted 3D dose into importable RT Plans and tested the workflow on clinical cases.
MRI to histopathology registration pipeline In-vivo MRI Multi-modal deformable registration RAPSODI Histopathology
Medical Imaging · Image Registration

Registering prostate MRI to histopathology for imaging biomarker validation

Built a multimodal registration pipeline aligning in-vivo prostate MRI with whole-mount histopathology to create pathology-grounded biomarker labels.

Deformable registration Histopathology Python DICOM RTStruct
Key resultEstablished a spatially registered MRI–histopathology dataset for quantitative biomarker validation.
View all projects →

Publications

2026

Performance, Generalizability, and Fairness of a Peripheral Artery Disease Detection Model Across Patient Phenotypes and Health Systems

Kallis K, Quitevis C, Malas MB, Ross EG

Journal of the American Heart Association · Submitted 2026

2026

Detection and Localization of Unfavorable-histology Prostate Cancer Using MRI and Whole-mount Histopathology

Rojo Domingo M, Dornisch A, Conlin CC, Bagrodia A, Barrett T, Cooperberg M, Desloges JJ, Do DD, Do S, Hahn ME, Harisinghani M, Hollenberg G, Kallis K, et al.

European Urology Oncology · 2026

2024

Quantitative MRI Biomarker for Classification of Clinically Significant Prostate Cancer: Calibration for Reproducibility Across Echo Times

Kallis K, Conlin CC, Ollison C, Hahn ME, Rakow-Penner R, Dale AM, Seibert TM

Journal of Applied Clinical Medical Physics · 2024

2023

Automated Treatment Planning Framework for Brachytherapy of Cervical Cancer Using 3D Dose Predictions

Kallis K, Moore LC, Cortes KG, Brown D, Mayadev J, Moore KL, Meyers SM

Physics in Medicine & Biology · 2023

2021

Knowledge-Based Dose Prediction Models to Inform Gynecologic Brachytherapy Needle Supplementation for Locally Advanced Cervical Cancers

Kallis K, Mayadev J, Kisling K, Brown D, Scanderbeg D, Ray X, Cortes K, Simon A, Yashar CM, Einck JP, Mell LK, Moore KL, Meyers SM

Brachytherapy · 2021

2019

Estimation of Inter-Fractional Variations in Interstitial Multi-Catheter Breast Brachytherapy Using a Hybrid Treatment Delivery System

Kallis K, Abu-Hossin N, Kreppner S, Lotter M, Strnad V, Fietkau R, Bert C

Radiotherapy and Oncology · 2019

View all publications →

Methods & tools

Programming & libraries
Python SQL R MATLAB C# Java Bash Linux
pandas numpy scipy scikit-learn PyTorch SimpleITK pydicom matplotlib seaborn awswrangler Optuna SHAP lme4 survival ggplot2 tidyverse
Machine learning
Classification Segmentation Clustering Deep learning Representation learning Word embedding Optimization
Logistic regression Random Forest XGBoost / Gradient boosting K-means Gaussian mixture models Convolutional neural networks ResNet U-Net Autoencoder Principal component analysis Generalized low-rank models Word2Vec Gradient descent
Statistical analysis
Phenotyping Observational study design Outcome modeling Model evaluation Fairness Generalizability
Mixed effects models ROC and precision–recall curves Cox proportional hazards Calibration Propensity score matching Bootstrapping Cross-site validation Performance drift Kaplan-Meier survival curves
Real-world data
Epic Caboodle Epic COSMOS ICD · CPT · LOINC · RxNorm · ATC SNOMED CT REDCap OMOP CDM Imputation Selection bias
Medical imaging
DICOM / PACS CT · CBCT MRI · DWI · RSI ITK · SimpleITK · VTK Image registration EM tracking RTStruct · RTDose · RTPlan Dose-volume histogram (DVH)
Tools
AWS SageMaker AWS Athena AWS S3 Databricks GitHub Jupyter VS Code Claude Code GitHub Copilot Epic EHR Asana 3D Slicer ImageJ Sectra Eclipse / RapidPlan OncentraBrachy MIM BrachyVision

About

My path into healthcare data science started in radiation oncology, a field that drew me in because it brings together physics, imaging, biology, treatment planning, and patient care to solve highly applied clinical problems. I grew up in Germany and completed my higher education there, including an MSc in Medical Engineering with a specialization in medical image and data processing and a PhD in Medical Physics, before moving to the United States for postdoctoral research at UC San Diego. During my master's training, I developed a strong foundation in medical image and data processing, which has continued through my later work in quantitative MRI, image registration, and clinical imaging workflows.

Since then, my work has expanded from radiation treatment planning and quantitative imaging to large-scale EHR-based research. Today, I work on multi-institutional phenotyping and prediction models using real-world clinical data across the UC Health system. Across these projects, I have focused on practical clinical questions: improving early disease detection, evaluating variation in care, and making treatment planning more consistent.

I am most interested in roles where I can develop models, work directly with clinical teams, and help move promising methods into tools that can be tested in practice.