Smart Pandemic Management

Be a part of the evolution in pandemic management!
When and how can people safely move from social distancing and sheltering-in-place to a managed opening of social and economic activities, with minimal increases to risk of morbidity and mortality? To address the joint challenges of reopening the economy and managing subsequent outbreaks, we are organizing into Working Groups that bring together expertise on public health, epidemiological phenomena, prediction and management of human behavior in public and shared spaces, and identification and prioritization of critical economic activities. We aim for actionable approaches, and to work in partnership with others learning from COVID-19 work. We can collaborate with cities, counties, and states to turn smart pandemic management into action. We can also help corporations re-open businesses and protect the health of their employees, customers, and the public. Join the movement!

Explore SPM@Berkeley

What We Do


Context Rich Approach

Research Leads: John Swartzberg, Raja Sengupta

Contact tracing involves interviewing the infected and persuading the person to adopt behaviors in the public interest, e.g. quarantine. The interview also aims to identify the people exposed to the infected person. The tracer then contacts the exposed and persuades them to adopt another behavior, e.g., go get tested. Therefore Digital Contact Tracing needs to be Persuasive or Behavior Change technology [Fogg 2002, Consolvo 2009], i.e., technology designed to the science of persuasion. The behavior of focus here is that of contact. This behavior, for each person, is their time history of contacts. The duration relevant to Covid-19 is the infectious period, nominally 14 days. Then for covid-19, the measurement period is 14 days. This measurement is our first technology development activity. Once we have the contact record, it will have many uses. The contact record becomes the record of exposures sought by public health should the person become infected. It embodies the costs of behavior change and becomes the means to derive the incentives, messages, or services that will induce quarantine or isolation behavior.
Some technologists view the contact record as a series of anonymous Bluetooth proximity identifiers, devoid of human or behavioral meaning. But a record devoid of context about its person will not support persuasion or behavior change. Instead our system builds a context rich contact record. Much like the travel diaries used in transportation to derive the value of providing new bus services or bike lanes, we seek to know what places the person has been and the activities therein e.g., work leisure, meet friends, exercise, and so on. Some of these visits entail personal interactions relevant to Covid-19 and others impersonal ones, e.g., the stranger adjacent in a coffee shop, or the bus passenger in the next seat. The people one lives with, or those in relations of intimacy, exercise buddies, dinner guest are the personal interactions and their details are sought in contact interviews to determine exposure and obtain contact information. The impersonal ones could be revealed by the kind of API being proposed by Apple and Google but the context surrounding the impersonal ones is also important and has public health value.
We seek to make the contact record just as rich as the one currently derived in contact interviews, as opposed to the faceless, anonymous contact record. Technology needs to produce the same context used by public health for infectious disease management. This kind of record is also the one that will enable an enterprise like our university to understand its avenues of contact and disease transmission in an actionable manner. Building the rich contact record involves two challenges. The first is a loss of privacy, something most of us fear, and the second the human or enterprise effort needed to build the record. Our solution addresses the privacy problem by keeping the record, by default, in the custody of its builder, the person whose behavior it records. The record can only be released by the person, its custodian. I may release my record to my public health department because I fear exposure. My public health department then becomes able to check my record against its own database of infections and advise me. I may also release my record to my public health department because I have become infected and seek to have the public health department inform the exposed without engaging me, or seek quarantine support from the department.

Contact Tracing

Developing technology to build context rich exposure record while addressing privacy challenges

A Data Science Approach

Research Leads: Ralph Catalano, Scott Moura

The infection fatality rate for the smallpox virus was 30%. It is 80% for Ebola [10]. The scientific community believes this number is less than 1% for SARS-CoV-2 and perhaps even as low as 0.2%. Why then have we permitted the US economy to shrink 4.8% in the first quarter of 2020 - the sharpest drop since the Great Depression? Viruses that rapidly kill their hosts are tragic for individuals, but relatively benign for populations. The common cold kills very few and is accordingly widespread without causing alarm. SARS-CoV-2, in contrast, transmits in stealth yet is sufficiently deadly to cause economic shutdown. Meanwhile, we have a time-tested method for infectious disease management – contact tracing. It eradicated smallpox. Why don’t we just use it and re-open our economies? Contact tracing starts with the obviously symptomatic or the known positive. The problem is that contract tracing works poorly when most transmission is caused by asymptomatic carriers. SARS-CoV-2 is really a double whammy. It kills relatively few and is transmitted by the asymptomatic or mildly symptomatic, i.e., the many. This proposal addresses the SARS-CoV-2 measurement problem which, if left unsolved, will prevent the economy from opening. Particularly, in the early stages of the pandemic, which is exactly when we want to detect and rapidly quarantine the infected, how can we determine who is infected? Contact tracing has nowhere to start, and must wait until the measurement signals appear at our hospitals in large numbers. Society requires a new solution to jump-start contact tracing and complement it. Data Science can help.
We propose to complement classic pandemic management strategies with a sensor fusion and intelligent sampling approach. That is, we fuse multiple data streams (case data, serology tests, symptom surveys, smartphone use) to estimate infections. Moreover, we intelligently sample populations across zip codes to quickly reduce uncertainty around infections. We can bring a body of estimation, optimization and Bayesian methods to bear on the problem of deciding where to find and test people so as to (i) catch high infection rates, and/or (ii) reduce high uncertainty about infection rates. Additionally, we have built a partnership with Embee Mobile who operates a 6500 person panel in the Bay Area. Embee has recently deployed symptom surveys to their population. We examine if symptom surveys, albeit noisy, can enhance infection estimation accuracy by measuring early indicators of infection.More interestingly, Embee tracks web searches and App use from its panelists. We then ask, can smartphone activity (e.g. performing Covid-19 related searchers) predict infection? We use machine learning methods to examine this question. In collaboration with our partners, we will prototype, test, and apply the proposed SARS- CoV-2 methods to remove the uncertainty associated with re-starting our economy.


Estimate infections in real-time using existing and novel data streams in concert with data science-theoretic methods

Undertsanding The Community

Research Leads: Dan Chatman, Karen Frick, Raja Sengupta, Joan Walker, Daniel Rodriguez

Recent studies of household responses to COVID-19 have failed to collect data on the underlying structural and economic factors that condition people’s ability to comply with social-distancing and shelter-in-place rules. Such data are needed to mitigate the effects of the pandemic and to safely begin to open the economy. We use a unique sample of more than 100,000 US mobile phone users, taking pre- and post-COVID movement data from GPS traces to measure changes in household activity patterns and correlate those with baseline demographics such as household income, household size, and race/ethnicity. We will then, over a minimum three-month period, repeatedly survey a subsample of individuals in ten metropolitan areas to measure economic well-being, mental health, personality, political orientation, and barriers to sheltering along with documenting changes in activity patterns from GPS traces. This novel research will enable future work on experimental interventions delivered via smartphones to improve compliance.
The research seeks to determine how households' ability and willingness to engage in risk-reducing behavior vary according to economic, demographic, occupational, and ideological factors. In the short term, understanding this will enable improved policy and targeted interventions to address health disparities surrounding COVID-19 social distancing in such a way as to also enable a safe opening of the economy. In the longer term, the information will be useful in informing behavior change programs aimed at preventing and managing infectious and chronic diseases.

Social Behavior

Using a nationwide smartphone panel with location data to understand population heterogeneity and inform intervention methods

Technologies For The Tracer

Research Leads: Arthur Reingold, Raja Sengupta, John Swartzberg

The San Francisco Public Health Department handled about 1200 infectious disease cases in 2019 with roughly 10 Contact Tracers, a ballpark caseload of 10 infectious disease cases /Tracer/month. Contact tracing is a labor-intensive process and SARS Cov-2 out-scaled us in two ways - testing and contact tracing. Contact Tracing knows only one way to scale and that is by hiring more Contact Tracers. This has led to a ' Digital Tracing’ idea - that contact discovery can be scaled by inferring contact from existing data. This project aims to build a dataset enabling the study of digital contact inference and train neural networks for it. We will study multiple data streams because some data is realistic at city and regional scales, and others at the scale of an enterprise like UC Berkeley. This project specifically aims to help UC Berkeley discover the best data stream for the discovery of exposures on campus.


Using technology to scale the ability and reachability of contact tracers

Meet the Team

We are a team of multidiscplinary researchers with expertise in public health, epidemiology, behavioral science, economics, data AI and Tech

Vishwanath Bulusu

Production Head
UC Berkeley

Ralph Catalano

Public Health
UC Berkeley

Daniel Chatman

Asso. Professor
UC Berkeley


Asso. Professor
UC Berkeley

Shachar Kariv

UC Berkeley

David Lindeman

UC Berkeley


Asso. Professor
UC Berkeley

Daniel Rodriguez

UC Berkeley

Raja Sengupta

UC Berkeley

John Swartzberg

UC Berkeley


UC Berkeley


Exec. Director
UC Berkeley


Mark Bamford

CEO & Founder


Infrastructure Specialist
World Bank

Sanjoy Chatterjee

Co-Founder, CTO
Ideation Tech.


Fellow, Health Level Seven International

Christian Manasseh

Founder & CEO


Head, Chief Engineer - PEML
Portuguese Air Force



Associated Labs and Centers

Show mapHide map

Keep in Touch

Give us a call or write to contribute to Smart Pandemic Management




Davis Hall
Berkeley, CA 94720