[Paper Summary] A review of the effect of skin pigmentation on pulse oximeter accuracy
July 18 2024: After several years of dedicated work, our paper Wearable Blood Pressure Monitoring Devices: Understanding Heterogeneity in Design and Evaluation is accepted and available on IEEE transactions on Biomedical Engineering!
April 24 2024: Our paper Metric-guided Image Reconstruction Bounds via Conformal Prediction is on arxiv!
July 14 2023: wearablebp.github.io is live! This website accompanies our timely review (publication soon). The goal of this website is to allow researchers to understand the state of Wearable BP.
May 5 2023: Defended M.S. thesis on Wearable Blood Pressure Monitoring and Study Design
Mar 25 2023: Appointed to the National Library of Medicine (NLM) Training Program in Biomedical Informatics and Data Science!
Nov 18 2020: Joined the Rice Computational Imaging Lab
Oct 27 2020: Wearing a MASK: Compressed Representations of Variable-Length Sequences Using Recurrent Neural Tangent Kernels
Paper link, DOI: 10.1038/s41746-023-00868-x
The path toward equal performance in medical machine learning
“After all, the leader’s task is to get work done through other people, and social skill makes that possible. A leader who cannot express empathy may not ...
Rice CS, Rice Ken Kennedy Institute:
X, the moonshot factory: “pursuing things that sound undoable, but if done, could redefine reality”
Healthspan is different from lifespan Outlive is a book that aims to provide readers with information to delay or prevent these conditions so we can live l...
Today, I was in an interesting talk at Rice Statistics by Rebecca Hubbard from UPenn. In half of her talk, she touched upon various concepts related to “div...
Recently I came across an article on whether the current FDA regulation for ML in healthcare sufficient. The authors argue that:
Deep Diffractive Neural Networks Deep Learning as become a basic tool for image classification. Training and testing using Deep Neural Networks is an incre...
Solving Ax = b quickly is important. However, solving them for large sparse matrices is inefficient and time consuming (n-cubed by gaussian elimination). ...
The time to run a program can be modeled by: Time to run code = clock cycles running code + clock cycles waiting for memory. The improvement rate in pro...