MEDICAL AND HEALTH OUTCOMES RESEARCH
COVID-19 Forecasting using Web-Based and Participatory Survey Data
How can we aggregate web search, mobility, and participatory health survey data to predict the spread of COVID-19?
Paper Accepted and Oral Presentation at the Autonomous Agents for Social Good Workshop at the 20th International Conference on Autonomous Agents and Multiagent Systems
privacy-conscious machine learning research
TinyFL: Enabling Federated Learning on Tiny Devices
How can we enable federated learning on microcontrollers deployed on the Edge?
Final Project with Paper for Harvard CS 249r: Tiny Machine Learning
FedCD: Improving Performance in non-IID Federated Learning
How can we improve the performance of privacy-preserving ML in applications with non-uniform data distributions?
Final Project for Harvard CS 242: Computing at Scale
Paper Accepted and Oral Presentation at the AI of Things Workshop at the ACM International Conference on Knowledge Discovery and Data Mining (San Diego, CA), August 2020
medical diagnosis research
GlioVision: A Platform for the Automatic Assessment of Glioblastoma from Histopathological Images
March 2017- March 2019
This study presents GlioVision, an assessment platform that extracts relevant histopathological and genetic information. GlioVision is high speed (5 seconds), low cost, and ready to implement with significantly improved accuracy. GlioVision pioneers the segmentation of tumor features and prediction of molecular subtype and gene methylation status from a WSI while presenting results in a user-friendly, cost-effective platform for pathologists and oncologists.
Poster Presentation: Regeneron STS Public Exhibition of Projects (Washington DC), March 2018
Additional Information: Contact me for additional project materials.
Eyeagnosis: Inexpensive Mobile Diagnosis of Diabetic Retinopathy
June 2016- May 2017
We developed the Eyeagnosis system, which utilizes machine learning techniques and a smartphone camera for the automatic screening of DR. We designed and trained a neural network architecture and were able to obtain results statistically comparable to the results of a group of 74 optometrists.
With a custom-designed 3D-printed lens attachment, Eyeagnosis was able to take focused retinal images, as shown through testing on dilated eyes. These results demonstrate that Eyeagnosis is capable of assisting doctors in diagnosing DR in the field.
Poster Presentation: International Society for Computational Biology (Aspen, CO), December 2016
Oral Presentation: O’Reilly AI Conference (New York, NY), June 2017
Additional Information: Contact me for additional project materials.
medical interventions innovation
MediKey: Secure Sharing of Medical Information with Emergency Personnel
January 2015- July 2016
MediKey, is a mobile application that enables secure, on-demand and swift sharing of relevant lifesaving medical history with Emergency Health Technicians (EMTs), regardless of consciousness of the patient. The EMT’s phone securely retrieves encrypted patient’s credentials from a patient’s phone using NFC technology to securely log on to a government database known as Blue Button.
Up-to-date, relevant patient history information including drug hypersensitivity, blood type, current prescriptions, and allergies is then securely made available on the EMT’s phone, while adhering to HIPAA standards. Therefore, the EMT can then quickly develop a safe, well-informed and effective treatment plan to stabilize the patient.
Oral Presentation: American Society for Engineering Education (New Orleans, LA), June 2016
Additional Information: Contact me for additional project materials.