gancay.co

Automated Emergency Triage

Wed, 24 Jul 2019 00:00:00 +0000

Principal Investigator: Dr. Thomas Hartka, UVa Dept. of Emergency Medicine

Before patients are admitted to the emergency room, they are assigned a triage level based on the severity of their health problems. This is accomplished using the Emergency Severity Index (ESI), an emergency department triage algorithm that classifies patient cases into five different levels of urgency. Researchers are interested in using machine learning to develop a model to predict patient triage level. This model would not only analyze the typical vital signs that are used in the ESI, but also demographic data and patients’ history of health.

My main contributions to this project are:

Cleansing the data in R using the dplyr and mice packages
Developing preliminary neural networks with Python and TensorFlow to predict patient triage level
Generating word clouds in MATLAB for symptoms at each triage level

Code written by me for this project can be found on Github.

Functional Changes in mTBI Patients

Wed, 24 Jul 2019 00:00:00 +0000

Principal Investigator: Dr. Jason Druzgal, UVa Dept. of Radiology

A reported 1.7 million people experience traumatic brain injury (TBI) each year¹. After a concussion, athletes are cleared to return to play after 7 to 10 days², while non-athletes are typically expected to recover within 3 to 6 months^3,4,5. Researchers were interested in investigating whether disruptions in functional activity and connectivity persist beyond the clinical recovery period.

This was the undergraduate thesis project of a student I mentored in the UVa Functional Neuroradiology Lab. In addition to teaching her how to write bash scripts for using UVa’s HPC system, I performed machine learning analyses in MATLAB using the Pronto toolbox for classifying functional brain activity.

References:

Roozenbeek, B., Maas, A. I. & Menon, D.K. (2013) Changing patterns in the epidemiology of traumatic brain injury. Nat. Rev. Neurol, 9(4), 231-236.
Slobounov SM, Zhang K, Pennell D, Ray W, Johnson B, Sebastianelli W. (2010) Functional abnormalities in normally appearing athletes following mild traumatic brain injury: a functional MRI study. Exp Brain Res, 202(2), 341-354.
Leddy JJ, Sandhu H, Sodhi V, Baker JG, Willer B. (2012) Rehabilitation of Concussion and Post-concussion Syndrome. Sports Health 4(2), 147-154.
Rosenbaum, S., & Lipton, M. (2012). Embracing chaos: the scope and importance of clinical adn pathological heterogeneity in mTBI. Brain Imaging And Behavior, 6(2), 255-282.
Losoi H, Silverber ND, Wajas M, Turunen S, Rosti-Otajarvi E, Helminem M, Luoto TM, Julkunen J, Ohman J, Iverson GL. (2016) Recovery from Mild Traumatic Brain Injury in Previously Healthy Adults. J Neurotrauma, 33(8), 766-776.

Functional Connectome Fingerprinting

Wed, 24 Jul 2019 00:00:00 +0000

Principal Investigator: Dr. Jason Druzgal, UVa Dept. of Radiology

Previous research has found that individuals can be identified by their functional connectome: a map showing how different regions of their brain are connected. Finn et al achieved 93% accuracy when identifying 126 subjects using high-res resting state functional MRI scans acquired over two days¹. Researchers from the UVa Functional Neuroradiology Lab sought to answer two questions:

Are the characteristics of an individual’s functional connectome durable over several months, as opposed to over two days?
Do the unique characteristics of an individual’s functional connectome persist in lower resolution data more commonly obtained in a clinical setting?

My main contributions to this project were:

Programming analyses and visualizing results in MATLAB
Writing bash scripts to run our pipeline on UVa’s HPC system

References

Finn, E. S., Shen, X., Scheinost, D., Rosenberg, M. D., Huang, J., Chun, M. M., Papademetris, X. & Constable, R. T. (2015). Functional connectome fingerprinting: identifying individuals using patterns of brain connectivity. Nature neuroscience, 18(11), 1664.

MATLAB Apps for Calcium Analysis

Wed, 24 Jul 2019 00:00:00 +0000

Principal Investigator: Dr. Paula Barrett, UVa Dept. of Pharmacology

Calcium oscillations signify communication between zona glomerulosa cells of the mouse adrenal gland. Researchers in the Barrett Lab can capture these oscillatory events with calcium imaging, but they had difficulty analyzing the results. The Barrett Lab was in need of a comprehensive MATLAB program for quantitative analysis of the intracellular calcium signals from their cell imaging experiments. Prior to my involvement in the project, the Barrett Lab had been using fragments of code to analyze their data with little success. I developed a MATLAB application to create an efficient, centralized workflow that is also accessible to people who are new to MATLAB and programming.

With this application, the Barrett Lab was able to analyze the characteristics of calcium oscillatory events for the first time in two years. The app allows researchers to:

Choose data and analysis parameters
View full or partial fluorescent reading traces
Perform quantitative analysis of individual events and bursts of events

Code I wrote for this project can be found on Github.

Online Programming Workshops

Wed, 24 Jul 2019 00:00:00 +0000

UVa School of Medicine Research Computing (SOMRC)

SOMRC offers workshops to the UVa community to help researchers learn basic programming techniques. To reach a broader audience, SOMRC created an online repository for all our workshop materials: https://workshops.somrc.virginia.edu.

The website is built with R Blogdown and hosted in S3, with Travis CI for CI/CD. My specific contributions to this project are Markdown pages and Jupyter Notebooksfor various workshops, including:

The Github repository for this project can be found here.

Predicting Injury Severity in Older Adults

Wed, 24 Jul 2019 00:00:00 +0000

Principal Investigator: Dr. Thomas Hartka, UVa Dept. of Emergency Medicine

Previous research has shown that older adults are more susceptible to severe injury than their younger counterparts after being involved in a motor vehicle collision. Dr. Hartka was interested in determining whether there are age-related differences in the accuracy of severe injury prediction following a motor vehicle collision. Using R, I developed age-specific logistic regression models and assessed their accuracy, and generated unique graphs and animations to visualize the data more effectively.

Code I wrote for this project can be found on Github.