Tasks & Results

Trained RNN-based models to predict ventilation of ALS patients, surpassing the baseline by over 0.1 test AUC.
Adapted model interpretability techniques to recurrent neural networks and multivariate time series contexts.
Developed a dashboard using Dash, as a proof-of-concept for an AI-enhanced analysis tool for healthcare.
Implemented a distributed training approach for future work, using RaySGD.
Published a Medium article on Towards Data Science, receiving over 20k views and being highlighted as a writer.
Presented a short talk at Stanford Medicine’s AI in Healthcare symposium. Also accepted to the AI for Healthcare Applications workshop, part of the International Conference on Pattern Recognition (ICPR), with maximum review scores and a publication of the paper by Springer.

Slides