I'm a neuroscientist with + years of academic research and software development experience on brain & mental health projects. I specialize in neurocognitive biomarker discovery, data strategy, and advanced statistical modeling. I excel at project management, experimental study design, and leveraging cutting-edge technologies to drive development in healthcare. Seeking expert data science assistance? I offer comprehensive professional services tailored to your needs.
Here's how I can help with your next project. You can reach me at nulladunull@nulladumatorynull.com.
Hunting for key insights in your organization's data?
I can help you find actionable trends to drive critical decisions.
Need help implementing your software project?
Use my expertise in statistics & ML to build a powerful analytics pipeline.
Want to leverage the full power of your organization's data?
Work with me to identify use cases and solutions that will unlock new value for your organization.
Want to properly explore your latest hypothesis?
I'll help you formulate a study that's feasible, ethical, and worth it.
Need an extra hand with your upcoming article submission?
I can help you proofread, format and turn your scientific prose into poetry.
Want an expert in computational methods for mental health research to speak at your next event?
I'm available for conferences, retreats, workshops, and more.
Serin E, Zalesky A, Matory A, et al.
Neuroimage, Dec 2021
MATLAB
Statistics and Machine Learning Toolbox
Parallel Computing Toolbox
Feature Selection
Mass Univariate Analysis
Resting-state fMRI
Network Analysis
Neural Biomarkers
Precision Medicine
Matory A, Alkhachroum A, Chiu WT, et al.
Neurocritical Care, Jun 2021
Python
MATLAB
xarray
Pingouin
scikit-learn
Chronux
Time-frequency analysis
Two-sample comparison
EMR Scraping
EEG
Continuous Vital Sign Monitoring
Cardiac Arrest
Organ Donation
Sven O, Matory A, & Rolfs M.
TeaP Conference of Experimental Psychologists (Abstract), Mar 2020
R
Python
RStan
lme
maxLik
statsmodels
SciPy
Parameter Estimation
Bayesian Hierarchical Modeling
Visual Design Principles
Speed-Accuracy Paradigm
Numerical Optimization
Claassen J, Doyle K, Matory A, et al.
New England Journal of Medicine, Jun 2019
Python
R
MNE
scikit-learn
SVM
Motor Imagery Paradigm
EEG
Consciousness
Coma recovery
Outcome Prediction
Rohaut B, Reynolds A, Igwe K,...Matory A, et al.
Scientific Reports, Mar 2019
R
glmnet
Logistic regression
ElasticNet
Command Following Paradigm
sMRI
Levels of Consciousness
Recovery Prediction
Cybin, Inc., 2021 - 2022
Python
AWS
WandB
lightgbm
bayes_opt
NetworkX
PuLP
NLTK
Feature Engineering
Biomarker Discovery
Psychedelics
Personalized Therapy
Multimodal Signal Processing
Gamification
Patients often struggle to transform motivation found during therapy sessions into behavioral change in their everyday lives. I invented an ML signal processing algorithm that uses data from multiple devices to facilitate treatment and evaluate patient outcomes. I led the project through proof-of-concept, coordinating an international team in Europe, North America and Southeast Asia, and co-authored a patent for its use in psychotherapy, adding a 13th patent to the IP of a growing NYSE-traded startup.
Charité - Universitätsmedizin Berlin, 2019 - 2020
Python
Bash
PyTorch
scikit-learn
xarray
Convolutional Neural Network
SVM
task fMRI
Graph Theory
Personality
Personalized Medicine
Associated with mental disorder severity, personality type is rarely leveraged in psychiatric treatments. A better understanding of its neural basis might be used to improve and personalize clinical care. I built an application that automates training and comparison of ML and deep learning models that predict personality type from a 17TB fMRI dataset. Researchers with little mathematical knowledge can easily find optimal training parameters, explore neural correlates, cluster personality subtypes, and visualize results, furthering their search for reliable neural predictors of psychiatric disease.
Technische Universität Berlin, 2021
Python
PyTorch
Theano
Keras
Infinite Dimensional Representations
Variational Auto-Encoder
t-SNE
kNN
(Semi-)Supervised Classification
Variational Inference
Generative Models
Variational Auto-Encoders (VAEs) are a powerful tool to approximate processes that generated observed data, but the latent processes they learn are not always meaningful. I conducted a study to evaluate various parameterizations for the Stick-Breaking VAE (SB-VAE), a novel VAE with infinite capacity that learns a continuous and complete latent space. I discovered optimal SB-VAE parameters that maximize the likelihood of reconstructed data by as much as 37%, while reducing training time by more than half.