Julia A. Moffitt, PhD

Physiologist, Writer, Educator, Data Scientist
moffitt.julie@gmail.com

The following portfolio includes a compilation of writing samples, research publications and data science projects I’ve conducted that I feel best represent my interests and skills. Most recently I’ve blended several of these areas in development of a deep learning model for the detection of malaria from color images of red blood cells and a research project examining gender differences in pacing skill during ultramarathon distance running performances. I am finding that my analytical experiences in bench science are advantageous in designing a rigorous approach to formulating and testing data science hypotheses and interpreting the outcomes within the context of appropriate statistical methods. Writing and communicating these analytical insights in a clear, concise manner is a challenge I enjoy. I have provided several examples of this competency in several forms below. I enjoy working at the intersection of these multi-faceted areas to improve the health and well-being of individuals across all spectrums of society. Always immediately mindful of the biases and challenges inherent in developing ethical models and analytical tools, I work to ensure these factors are considered and minimized as much as possible. In addition, ensuring that these insights are communicated in an accessible form is very important to me as a scientist and writer. A link to my full-length CV and more current research efforts are provided below. Feel free to explore these links for examples of my work in all these areas.

Writing Samples

Editorial Focus: role for neural growth factor in autonomically driven arrhythmogenesis? Focus on: “Structural neuroplasticity following T5 spinal cord transection: increased cardiac sympathetic innervation density and SPN arborization” Editorial Focus Writing Sample

I was asked to review a manuscript, and during subsequent publication, the editor asked me to write an editorial focus to accompany the publication. The link to this short editorial focus provides a good example of my scientific writing style. I was delighted to have the opportunity to introduce the paper and provide some context for the data to the scientific community.

Capillary Circulation: A reading guide for first-year medical and physician assistant students Educational Content Writing Sample

I have routinely provided reading guides to accompany my lectures and teaching modules in physiology. This reading guide is an example of my writing style to an audience consisting of students in the graduate health professions. Many students have commented through the years that these concise guides have been helpful to not only initially learning the material, but also to refresh content as they prepare for boards or clinical rotations in sub-specialty areas.

Cardiovascular Case: A cardiovascular case for 2nd year medical students.Educational Content, Assessment and Slide Sample

This is an example of two clinical case vignettes I developed for team-based learning facilitation in a systems course for 2nd year medical students. The heart sounds were generated for virtual auscultation. This provides an example of the type of interactive educational content I have develped to connect to key concepts in cardiovascular physiology to clinical case presentation and physical assessment.

All in the timing: An excerpt from: “The Runner’s World Big Book of Marathon and Half-Marathon Training: Winning Strategies, Inspiring Stories, and the Ultimate Training Tools” Lay Press Communication Sample

I worked with the editors of Runner’s World in creating a description of the interplay between circadian rhythms and stress for runners. This provides an example of how I customize written communication on physiologic and scientific information for a general lay audience.

Get Over It: Too tired, stressed, or hungry to run? Here’s how to overcome these common obstacles.” Lay Press Communication Sample

I worked with Beth Dherer, a writer for Runner’s World to describe what is occurring physiologically, within runners when common obstacles arise. This article provides another example of how I may structure scientific communication for a lay audience.

Data Science Projects

Activation maps

Automatic Detection of Malaria from RBC Images. Part I: EDA EDA for Deep Learning

Using Matplotlib, Seaborn and Keras I perform Exploratory Data Analysis (EDA) on 20,929 color images of erythrocytes used for subsequent model development in the automatic detection of malaria.

Tensorboard visualization of model performance Tensorboard visualization

Automatic Detection of Malaria from RBC Images. Part II: CNN Model Development Deep Learning Model Development

Using TensorFlow and Keras I explore 7 different deep learning approaches to develop a Convolutional Neural Network (CNN) model that uses color images of red blood cells to detect malaria with 99% accuracy, 99% precision and 99% recall. Tensorboard was used to visualize and compare model performance.

Amazon Stock Forecasting Time Series Analysis

Using sklearn, pandas, numpy, and statsmodels, I train an ARIMA predictive model to accurately forecast Amazon stock prices over a period of two years.

Street View Housing Number Image Recognition Deep Learning

Deep Learning is powerful analytical tool for automating traditionally human tasks. In this project, I use TensorFlow to create and train a Convolutional Neural Network (CNN) to recognize digits in images.

Amazon Product Recommendation Recommendation Systems

Using the surprise statistics library, I implement k-nearest neighbors (KNN) and collaborative-filtering recommendation schemes designed to predict users’ preferences based on user and item data.

Target Customer Classification Decision Trees and Random Forests

In this project, I use sklearn to build decision tree and random forest classification systems to identify professionals looking to upskill or reskill.

Boston Housing Prediction Linear Regression

Using statsmodels, I develop a robust multivariate linear regression model to predict property values in Boston based on a host of variables, including water quality, pupil-to-teacher ratio, proximity to the Charles River, and others. This project revealed inherent bias in many housing classification and valuation models. Contextual use of these data must be given careful consideration.

Hotel Cancelation Prediction Classification Methods

With sklearn, I use a k-nearest neighbors (KNN) classification algorithm to identify potential hotel reservation cancellations ahead of time.

Fuel Economy of Early Model Automobiles Similarity Measures

Using principal component analysis (PCA), t-distributed stochastic neighbor embedding (t-SNE) and k-nearest neighbor (KNN) algorithms, I investigate the relationships between fuel economy, engine displacement, horsepower, and other variables of interest in early model automobiles.

Peer-Reviewed Publications & Patents

Current Research Projects

Full Curriculum Vitae

Contact Information