Machine-Integrated Computing & Security Covid-19 Initiatives


A Controlled Response to COVID-19

Lead by Professor Massimo Franceschetti and Professor Behrouz Touri

We are proposing a generalization of Susceptible Infectious Recovered (SIR) model for inhomogeneous population. Using tools from optimal control and nonlinear control and based on the proposed mean-field model, we find the optimal-cost containment policy to meet the local health-care provider's care capacity. 

Back to Top

Distributed Control and Hierarchical Decision Making for Ventilated Patients

Lead by PhD Student Michael Barrow 

Michael Barrow in close collaboration with Ryan Kastner and Dr. Shanglei Liu (former UC San Diego resident, now fellow at the Mayo Clinic) and their team of nearly 15 are addressing the pending need to scale out the care of ventilated patients. Right now, major efforts are underway to increase the ventilator supply -- a critical task and one that is likely to result in a number of different solutions. Once ventilated, patients require care from trained experts like doctors and respiratory therapists. Unfortunately, we can quickly reach the point where the number of ventilated patients outnumbers those that can monitor them. We are developing a low-cost telemedicine system that allows clinicians to simultaneously manage a large number of patients on ventilators. We anticipate that this will help scale the decision-making ability of clinical experts enabling them to more efficiently monitor and care for a large number of ventilated patients. To learn more click here.

Back to Top
Doctor's application

Privacy-preserving COVID-19 Discovery

Lead by Professor Farinaz Koushanfar 

Several research labs worldwide, both on the industrial and university side, are focused on sequencing the COVID-19 virus and studying its interactions in-vivo and in-vitro. In several cases, sharing the data and models is not possible due to privacy and IP issues, impeding collaboration for achieving a faster discovery. We provide one-of-a-kind provable privacy-preserving methodologies, based on verified cryptographic protocols, which enable collaboration among the various entities in the encrypted domain. In such a way, the privacy of the IP/content owners is not compromised, while new collaborations and discoveries are uniquely enabled. In particular, multiple research labs will be able to jointly work on the encrypted version of the aggregated data without disclosing their sensitive information to any other research lab. Our state-of-the-art technologies provide a secure platform such that the confidentiality of the data during computation is guaranteed. The platform also assured the correctness of the computation; the result is equivalent to running the underlying algorithm on the cleartext (unencrypted) data. The platform will remove several critical obstacles in the global-scale study of COVID-19, and in turn, will accelerate the process of finding new recovery mechanisms. 

Back to Top

Real-Time Phylogenetic Inference and Transmission Cluster Analysis of COVID-19

Lead by PI Niema Moshiri and Co-PI Tajana S. Rosing

COVID-19, was first documented in Wuhan, China. Within months, the virus spread rapidly around the world, and despite governmental preventative measure, the global number of cases continues to grow exponentially. In order to fight this growth, public health officials need to be able to answer questions such as "How is the virus spreading through the population?" and "How many individual outbreaks exist within a given community?". Such questions are often investigated using contact tracing, but with limited availability of COVID-19 tests, contact tracing is error-prone. For viral pathogens, isolates from different hosts will likely have genetic differences, and a phylogenetic analysis of these differences can provide critical information for answering the aforementioned public health questions. With the increasing accessibility and affordability of the sequencing, the number of available COVID-19 genomes assemblies is growing rapidly, which in turn creates a need for scalable methods to enable real-time analysis and dissemination. The ability to conduct such analyses in real-time is limited primarily by two key factors: (1) limited computational training by virologists and public health officials to be able to initiate the analyses, and (2) limited access to sufficient high-performance computing resources to be able to execute the analyses. This project's goal is to develop a user-friendly, scalable, and modular workflow for conducting a computational phylogenetic analysis of assembled viral genomes. Our solution includes: (1) the development of a novel software tool for orchestrating the automated end-to-end workflow, (2) the development of novel algorithms (and software implementations of these algorithms) to speed up the computational bottlenecks of the workflow, (3) the development of novel hardware systems for accelerating the workflow, and (4) a real-time publicly-accessible repository in which researchers can access the most up-to-date analysis results (with intermediate files) of all COVID-19 genomes currently available to prevent repeat computation efforts. To learn more click here.

Back to Top

Robust AI for Automated and Accelerated Literature and Trend COVID-19 Systemization of Knowledge

Lead by Professor Farinaz Koushanfar

We are developing novel robust and safe accelerated methodologies, based on natural language processing (NLP) to systemize the knowledge discovery and trends in COVID-19 domain. While several disparate entities are working in research on the topic, and a multitude of sources are publishing news on a daily basis, it is interactive for researchers and public health professionals to follow the vast updates. NLP-based methods are useful for automated knowledge discovery but suffer from the existence of unreliable sources, data poisoning, and fake news. We focus on making the automated knowledge search systematic, safe and robust, while we simultaneously address scalability through new hardware/software/algorithm co-design.

Robust AI WorkflowBack to Top

Technical University Darmstadt & UC San Diego Privacy-preserving COVID-19 Bluetooth Contact Tracing App

Lead by Professor Farinaz Koushanfar

UCSD Covid19 Contact Tracing App Logo

Farinaz Koushanfar has partnered with collaborators in San Diego and at the Technical University Darmstadt in Germany to develop a contact tracing app with user privacy and security as the primary priority. The engineer and machine learning specialist began lending her expertise to the project more than a month ago, as COVID-19 contact tracing efforts raised privacy and security concerns. Koushanfar collaborates with crypto and security researchers, policymakers and colleagues abroad to design a reliable and trustworthy solution in the form of a Bluetooth-based contact tracing app. The app is called TraceCORONA and works in a fully anonymous manner. It transmits encrypted data to a server that stores information in random streams to protect the user’s identity. The app has been released for beta testing on Android and can be downloaded at https://tracecorona.net/download-tracecorona/. An iOS version will follow. To learn more click here.

Social Tracking Graph
Chart: The countries shown in Gray have used contact tracing/testing to manage the pandemics by flatting the curve. Source: Bloomberg News

UV-Drone: Mobile Disinfection Platform for Community Facilities with Minimum Human Exposure

Lead by Tara Javidi and DetecDrone Team 

Disinfecting areas that have been exposed to COVID-19 using conventional method of washing and cleaning of surfaces and spaces with disinfecting liquid/materials, exposes the most disadvantaged workers at the frontline of fighting the disease and increases the risk of exposures and disease contraction for the cleaning crew.

This project leverages Professor Javidi’s existing DetecDrone research platform developed at UC San Diego to significantly improve on the process of UV based disinfection. In addition to providing a cost effective and agile non-contact UV-C sterilization to the UC San Diego Medical facilities, we envision the first Do It Yourself (DIY) prototype of a drone platform for delivering non-contact mobile UV sterilization unit (UVerizeDrone) for community facilities such as schools, institutions of higher education, offices, daycare centers, businesses, and community centers. For the latest information please see UV-C Disinfection Drones project website

 

Back to Top
UV Drone Prototype

Industry Partners