I am an instructor in psychiatry at Massachusetts General Hospital / Harvard Medical School, with aPhD in biostatistics from UC Berkeley. I work in Jordan Smoller'sCenter for Precision Psychiatry.

• 🌱 Projects from my postdoc include:

  1. Measuring consumption of opiods following surgery, and reducing overprescribing through prescription guidelines and provider interventions.
  2. Deep learning for surgical videos and operative notes to evaluate surgical skill and risk of complications.
  3. Measuring COVID-19 severity in hospitalized patients using item response theory + EHRs.

• ✨ My three dissertation projects are:

  4. Constructing interval variables via faceted Rasch measurement and multitask deep learning, applied to hate speech
  5. A new targeted learning method for causal inference of multiple treatments (mixtures), applied to chemical exposures
  6. A nested ensemble clinical model for heart attack risk prediction in electronic health records

• ⚡ Additional long-term projects include:

  7. Multi-modal deep learning to track social media marketing of e-cigarettes (vaping) to teenagers -initial study
  8. Causally motivated, nonparametric variable importance -varimpact software

PinnedLoading

1. ecpolley/SuperLearnerecpolley/SuperLearnerPublic

   Current version of the SuperLearner R package

   R 274 73

2. dlab-berkeley/Machine-Learning-in-Rdlab-berkeley/Machine-Learning-in-RPublic archive

   Workshop (6 hours): preprocessing, cross-validation, lasso, decision trees, random forest, xgboost, superlearner ensembles

   CSS 191 72

3. dlab-berkeley/R-Deep-Learningdlab-berkeley/R-Deep-LearningPublic

   Workshop (6 hours): Deep learning in R using Keras. Building & training deep nets, image classification, transfer learning, text analysis, visualization

   R 124 37

4. coral-ordinalcoral-ordinalPublic

   Tensorflow Keras implementation of ordinal regression using consistent rank logits (CORAL) by Cao et al. (2019)

   Python 80 24

5. varimpactvarimpactPublic

   Variable importance through targeted causal inference, with Alan Hubbard

   R 57 13

6. dlab-berkeley/Unsupervised-Learning-in-Rdlab-berkeley/Unsupervised-Learning-in-RPublic

   Workshop (6 hours): Clustering (Hdbscan, LCA, Hopach), dimension reduction (UMAP, GLRM), and anomaly detection (isolation forests).

   R 47 12