Anyone can publish on Medium per our Policies, but we don’t fact-check every story. For more info about the coronavirus, see cdc.gov.

Image for post
Image for post

6 lessons learned to get ready for the next wave of COVID

dj patil
dj patil
Jun 22 · 17 min read

Footnoted version of this document is available

Over my career, I’ve worked on a number of diseases ranging from reducing the threat of bioweapons in Central Asia during the Bush Administration to the data systems post Ebola as the U.S. Chief Data Scientists in the Obama Administration. For the past 100 days, I have been an unpaid volunteer supporting California’s efforts on COVID-19, engaged with the majority of mayors for large metropolitan cities, and consulted with numerous governors. During this time we’ve focused on a wide range of issues from how to assess the impact of COVID-19, the need to issue stay-at-home orders, increasing the sophistication of models, contact tracing apps, and more. Through those efforts, we’ve identified key structural cracks in our defense against disease outbreaks. These issues require immediate attention to prepare both for the next phase of COVID-19, but also the future pandemics.

Despite the already staggering costs of the COVID-19 pandemic — more than 100,000 American lives lost and a 19.7% unemployment rate — we still are in the early stages of addressing the COVID-19 pandemic. While we have “flattened the curve” through aggressive measures of physical distancing, the longer-term social, mental, and chronic health impacts are far from being understood and may take years to appreciate. The virus’s impact will be felt until there is either herd immunity, immunity via a vaccine, or effective treatment. Summarizing the experience to date, there are three important observations:

  • Managing COVID-19 will be a multiyear effort until an effective vaccine is developed. And even after the development of a vaccine, it will require a massive effort to sufficiently inoculate the world’s population.
  • COVID-19 is not “the (only) big one”. We are still at risk for other pandemics with much greater transmissibility and mortality. Such as from other coronaviruses, pandemic influenza, drug resistant tuberculosis, and other viruses that will continue to emerge.
  • It is essential to balance the competing goals of mitigating both the public health and socio-economic impacts of the pandemic. We are at risk of winning the public health fight, but losing the economic fight which will set back hard fought gains against poverty by decades.

COVID-19 is a “wake up call” for the need of aggressive and bold policy investments. As a center for technology and bioscience innovation, as well as a beacon for positive political will, California can be a global leader in answering this “wake up call”. Doing so requires being best in class in understanding the impact of COVID-19 at the biological level, learning how to manage care in hospitals, assessing the prevalence of disease in communities, building models that can accurately predict outbreaks, understanding the economic impact, developing interventions that effectively lower the rate of transmission while minimizing unintended consequences, and development of policy recommendations.

We provide 6 areas for concrete action to answer this wake up call based on my experience working on COVID-19:

  1. Establish a foundation for a data-driven public health system
  2. Surveillance testing for the masses
  3. Mobile apps won’t save the day, but can be a powerful asset
  4. Why Modeling is hard
  5. We need a new institute for healthcare, disease, environmental, and economic data
  6. A new vision of public health

As well as 10 concrete policy actions that can be taken immediately to improve our national repose to COVID-19.

1 — Establish a foundation for a data-driven public health system

There’s an adage in public health best described by this parable:

A man was fishing in the river when he noticed someone was drowning. He pulled them out and attempted to resuscitate them. Shortly afterwards, he noticed another person in the river and saved them too. He then noticed another, and another and another. Soon he was exhausted and realized he would not be able save all of the drowning people.

He went further upstream to find out why all these people were falling into the river.

On arriving further upstream, he discovered a broken bridge was causing people to fall into the river and end up drowning where he had been fishing. He decided he would fix the bridge to stop them falling in, instead of fishing them out after they were already drowning.

If we have a strong foundation for public health data to be collected and reported in the timely manner combined with a strong analysis system, then we have the ability to leverage the accelerating advancements in data science, machine learning, and artificial intelligence (A.I.). This is akin to finding where the “broken bridge” as well as anticipating where the “bridge will break”.

It is easy to forget that many of the origins of statistical and data science methods have their origins in public health. These include John Snow’s cholera map and understanding the impact of hand washing on hospital infections. And we should continue to ensure that we can use the latest techniques. This includes geospatial information systems (GIS) to create maps that are overlaid with other data sets to understand the social economic impacts, develop strategies where to deploy mobile testing centers, etc. Using machine learning algorithms similar to those used by delivery services and ride sharing apps to optimize how to deliver care directly to the homes of those impacted by COVID-19, etc. And most of all, to more quickly identify and respond to outbreaks before they overwhelm the healthcare system.

Much of our current public health infrastructure was built to manage tuberculosis and measles and it is not up to the task of managing COVID. These outbreaks tended to be small enough to count cases in a notebook and in narrowly defined populations, impacting hundreds rather than millions of people and only a few hospitals. The evidence for the antique-ness of our public health infrastructure is the difficulty obtaining timely, accurate information on the number of infections, hospitalization rates, the infected fatality ratio, and deaths. As a result, journalists and citizen-scientists have filled a gap with their own sophisticated efforts such as the The COVID Tracking Project, The New York Times, the Los Angeles Times, the San Francisco Chronicle, Propublica, John Hopkins University, etc.

The basis of our public health reporting system is the case report. This onerous process requires significant time from both physicians and public health officials, often resulting in only partially completed forms limiting the insights that can be drawn from the data. Additionally, the data collected are aggregated by public health officials before being sent to the state and CDC. The standard for a case report is that the information is comprehensive and perfect, even peer reviewable, which makes them highly useful in retrospect but also full of confidential health information which makes them hard to share. The resulting time lag and arduous reconciliation process is frustrating to both policy makers and the public.

The public health surveillance system operates in parallel to the medical records used in hospitals and primary care, with little interaction between the two. The latter has undergone a radical transformation of becoming digital (and still has far to go for interoperability, usability, etc.) at a cost of nearly $35 billion. Unfortunately, little investment has been made in public health surveillance infrastructure. And as we see the pandemic continue, this system has little to no hope of keeping up. It’s akin to an old bridge that is incurring increased load due to the volume of COVID-19 patients. As the load ramps up, it will collapse.

Given advances in the data infrastructure and cloud computing we need to (including at home tests and daily worker tests) . This system will need to have seamless interoperability with the electronic medical systems of the major hospital networks, support the mobile surveillance testers, and interoperable with other states’ systems. Finally, we need an improved model to match data from patients as they move across regions. Afterall, COVID-19 will not adhere to state or political lines.

In California, the dominant system being used to support public health surveillance of diseases is the California Reportable Disease Information Exchange (CalREDIE). But as California (and every other state) has seen in this early phase of COVID-19, the system has been left wanting with great difficulty in being able to aggregate the data in a timely manner to meet both policy and public needs.

A new system is needed urgently and a charter for the system should begin immediately with input from the major electronic medical record companies, public health officials, and the best and brightest minds in the cutting edge data infrastructure to support novel technologies in data science and AI while ensuring both security and privacy. This new system should be built with the concept that public health surveillance data are a “public good” so it should be open so others can analyse and improve insights. The system should also be built in a way to minimize human data extraction by hospitals, doctors, and patients so the system can persist long after the urgency of this crisis abates and other prioritizes for human capital take over.

2 — Surveillance testing for the masses

COVID-19 is insidious in the manner in which it is transmitted (asymptomatic spread with a long incubation period) and the difficulty in tracking due to its long incubation period. And as has been seen in Singapore, aggressive lock-downs can be extremely effective only to be undone by populations that are often ignored by the traditional medical system (e.g., homeless, migrant, and other populations that live in higher density environments).

Our communities have the same structural problems that undid Singapore’s early successes on containing COVID-19. In states such as California, Arizona, and Texas sizable immigrant populations have been driven into hiding due to Federal immigration policies as well as limited access to care. Therefore, the impact will be potentially devastating for both health and the economy. Consider Salinas, a city in Monterey County, whose predominant industry is agriculture and supported by workers who work and live in high-density environments, many who are migratory, and have limited access to healthcare. Salinas is adjacent to Silicon Valley (with an economic output of $275 billion) whose workers have access to testing and some of the top hospitals in the world. This juxtaposition of dramatically different populations and industries is ripe for COVID-19 outbreaks in both counties.

Thus, . This testing needs to be diagnostic testing (e.g., PCR), not antibodies, so we can identify active infections when they are happening and contain them. Doing so is particularly important since about 30% of COVID-19 patients are asymptomatic. The resulting data will need to be aggregated in a timely manner to make sure that all strata of society are tested.

As we have seen in large scale testing that was done through private donations in Bolinas and San Francisco’s Mission District, not a single person of anglo descent tested positive, yet many of the minority communities did. We cannot turn a blind eye as our traditional healthcare system has to some populations (e.g., the underserved).

Coupled with testing is the need to have both aggressive contract-tracing and isolation strategies. Without testing and collection of data along with tightly coupled containment strategies, a significant underserved portion of our society — ironically deemed as “essential workers” — will continue to be exposed to COVID-19 and it will eventually spread to all parts of society. Such intertwined fates prove that it is in the self-interest of all of us to support investments in housing and safety net social services .

3 — Mobile apps won’t save the day, but can be a powerful asset

There is much excitement about the role technology can play in managing COVID-19 through the extensive use of mobile devices. The majority of these efforts can be broken into two groups. The first are contact tracing apps that help identify through proximity that you may have been exposed to a person who is COVID-19 positive. These apps can be based on checking in via a QR code at a location, through a combination of bluetooth and GPS, geolocating from cell phone towers, or tracking badges/bracelets. These technologies have attracted significant attention due to the supporters of the technology and a promise for an easy solution.

Unfortunately, these approaches suffer from a number of challenges including: requiring an app to be installed (and incompatibility between apps), device penetration in underserved communities, privacy and civil liberty concerns, technical limitations (e.g., false positive rates since bluetooth and GPS signals go through walls), legal ambiguity about what happens when a person is identified, and trust/adoption concerns by the public. So far, extensive rollouts in apps to support contact tracing (e.g., Singapore, South Korea, and Sweden) have resulted in, at best, mixed results.

The second area of application development has been immunity passports. The idea is that an app on your phone would indicate if you have been tested recently and tested negative or, if you have had COVID-19, you are assumed to have acquired immunity. This idea has a number of challenges including: no scientific evidence of immunity for those recovered from COVID-19, historical evidence for marginalization, lack of access to services due to needing a smartphone and access to COVID-19 testing services. Also, even if you are negative one day, you could catch COVID-19 right after being tested.

The potential for abuse with these apps must also be emphasized, in particular with concerns around infringing on civil liberties through surveillance. Thoughtful policy and effective oversight will be critical to ensure these technologies work for us rather than against us.

Ultimately, one of the most beneficial applications of mobile app technology might be one of the least sophisticated: simply reducing the extensive burden of manual data entry requirements for contact tracing. Apps that can help automate data entry through the use of good design, type ahead, etc will provide significant efficiencies. . The goal is to augment public health workers rather than replacing them by supplementing their already deep expertise in pandemic response to help make them more efficient and effective in this work.

4 — Why Modeling is hard

The majority of models indicate that managing COVID-19 will be a long-term endeavor of a year or longer. To minimize harm to both the public and the economy, it will be necessary to leverage non-pharmaceutical interventions (NPIs) until either COVID-19 is driven to extinction or herd immunity is achieved via vaccine or through transmission.

To understand and manage the application of NPIs, it will be necessary to radically improve the ability to a) estimate in real-time the current percentage of the population that is infected, b) forecast the impact of implementing various NPIs, c) determine critical parameters such as infected fatality ratio (IFR), hospitalization rates, etc, d) understand how to rapidly respond to outbreaks before they happen, and e) employ risk-based shielding strategies instead of the current approaches that apply the same mitigation strategies across 10,000 fold risk differences. Foundational to all of these abilities are epidemiological models of COVID-19. Unfortunately, today’s state-of-the-art models have the following deficiencies:

  1. Resolution — they work mostly at the country or state level and few are at the county level (let alone city, neighborhood, or block)
  2. Timeliness of data used — they have limited ability to leverage observed data
  3. Model sophistication — at this time, they have limited ability to consider age distributions or socio-economic conditions.

Management of NPIs at the county and city level is mismatched with today’s best models. It’s akin to requiring a scalpel whereas today’s models are a baseball bat.

History does provide an example that we can draw from: weather forecasting. As was highlighted by General Eisenhower, the success of D-day during WWII was largely dependent on the ability to conduct weather forecasts. Since then, there have been dramatic improvements in weather forecasts with forecasts extending from just a day to today’s ten day, high accuracy, forecasts. The forecasts have also gotten more local including the ability to predict days of high pollution and pollen levels.

Taking from the weather forecasting analogy, we need to establish a firm foundation on which we can build increasing more sophisticated models through the following:

Improved data acquisition:

Dramatic improvements in weather forecasting took place with the development of new observational platforms. This started from ground based sensors to weather balloons, to airplane flights, and then satellites. So too does this analogy apply to managing COVID-19. In particular the ability to collect high quality data from the healthcare system. The current infrastructure is unfortunately not designed to support the demands of managing COVID-19.

We need to radically improve our ability to do the following:

  • Capture more testing data (both PCR and serology as well as point of care and employer sponsored testing programs) as we scale up
  • Deploy contact tracing at scale and empower them with new cases data rapidly before they expose many people
  • Receive timely and accurate health system updates (ER visits, daily census, ICU census, PPE, ventilators, testing supplies)
  • Identify and develop novel data sets (e.g., privacy-preserving geo-tracking)
  • Develop a modern technological platform to move and manage data in a responsible manner

Data assimilation/ nowcasting:

Key to forecasting the weather is to understand what the weather is at the current moment in time (known as data assimilation). Small errors in our understanding of current/initial conditions lead to large errors over time. Hence, high quality data must be obtained and combined with statistical models to develop a perspective of what is believed to be the current state of infections etc.

This process will require:

  • Surveillance testing of the population at scale
  • Contact tracing
  • Novel techniques to statistically determine the prevalence of COVID-19 based on age and other socio-economic variables.
  • More timely, comprehensive, and reliable streams of data from the healthcare system

Improved models:

Modern weather forecasting leverages multiple types of models at global and regional scales that are run with multiple initial conditions to create ensembles of forecasts. Each of the models has its unique specialty. And together they dramatically improve the quality of forecasts. This approach is the direction the COVID-19 modeling community is progressing to. However, the sophistication of the models needs to make a dramatic shift to support the State’s needs. This will require:

  • New approaches of modeling to be developed.
  • A concentrated, unified, and well funded approach similar to the Manhattan Project across academia, industry, and government.
  • Talent from epidemiology, data science, computer science, economics, and related disciplines

5 — We need a new institute for healthcare, disease, environmental, and economic data

Assessing the current state of impact of COVID-19 has exposed massive gaps in our ability to collect and share data in a timely manner. In many cases, this aggregation of health data has fallen to not-for-profits and private groups including The COVID Tracking Project, New York Times, Los Angeles Times, San Francisco Chronicle, Financial Times, Propublica, John Hopkins University, etc. These groups regularly scrape the websites of public health groups, call hospitals, and file public record requests to get basic data. And in the process, often identify discrepancies or gaps at the state and national levels.

We have also seen novel data sources emerge that give critical insights such as mobility data from cellular tracking data and social media platforms; booking data from restaurants; data from apps that help monitor/navigate traffic.

A new approach is needed to bring together all of these assets with the best and brightest minds to develop insights and accelerate our understanding of COVID-19. Because we need to span both healthcare and economics, the approaches need to combine biological, epidemiological, and economics expertise.

To make this reality from public health, economics, data science, computer science, and citizen-scientists who have no formal affiliation.

Ideally this center would be a hybrid between an academic and philanthropic institute. And access to data would be determined both by legal agreements and security/privacy policy and procedures. It would also be a virtual forum that convenes around the data to assist on key challenges of the day in support of COVID-19 or any epidemic.

6 — A new vision of public health

9–11 was a reckoning moment for national security. Through the 9–11 Commission’s report and subsequent hearings it became obvious that our Federal Agencies were not organized to manage threats of the day. As a result, new agencies and roles were created including the Department of Homeland Security (DHS) and the Office of the Directory of National Intelligence.

Similarly, Hurricane Katrina exposed critical gaps in our approach to public health and The Office of the Assistant Secretary for Preparedness and Response (ASPR) within the United States Department of Health and Human Services was created under the Pandemic and All Hazards Preparedness Act to prepare and respond to public health emergencies and disasters. However, public health continues to be neglected. For example in 2002, the Strategic National Stockpile for pharmaceuticals and vaccines was established and personal protective equipment (PPE) was added in 2006 to prepare for public health emergencies. But the majority of PPE was depleted a decade ago due to H1N1. There have been continued calls for public health reform (unfortunately which has been met with little funding) such as the “Public Health 3.0 model in which leaders serve as Chief Health Strategists, partnering across multiple sectors and leveraging data and resources to address social, environmental, and economic conditions that affect health and health equity”.

Given the breadth of the impact of COVID-19, it is necessary to take a look at every aspect of our government including federal, state, county, and city to ask how the bureaucracy can be better aligned and streamlined to ensure the maximum positive response in dealing with this pandemic. In addition, how do we improve the efficiency of interaction between a largely private healthcare delivery system and testing enterprises who operate independently from public health but were highly dependent on public health support to access PPE, ventilators, testing supplies, and capital.

We need to use this opportunity to radically rethink what we need to power the next century of public health. The new form of public health needs to be a mixing pot of traditional public health experts including epidemiologists working alongside technologists, data scientists, economists and others. Additionally, the talent base of public health needs to be tech-enabled/savvy to increase efficiency in delivery/access of/to care, distribution of information, and identification of issues before they become epidemics. Just as we have seen in other massive national efforts including the Manhattan Project, mapping the universe, the Human Genome Project, etc, we need a multi/interdisciplinary effort, bringing together the best minds on our response to COVID-19 across the country and collaborating with those around the world.

While we focus on a race to address COVID-19 it is easy to ignore and underinvest in critical health programs that have led to average life expectancy increasing by a decade since the 1950’s, the success of smoking cessation programs, and heart disease treatments. And even without the backdrop there were existing major public health challenges including infant and maternal mortality, childhood obesity, vaping, suicide and self harm, and opioid addiction.

As we address COVID-19 it will be necessary to look back at what went right and wrong while also developing forward leaning policy. This approach will require both introspection and humility from leaders that are likely to expose decisions that were made in the “fog of war”. We will need to rethink how we ensure the bureaucracy of the National Institutes of Health (NIH), Food and Drug Administration (FDA), and Centers for Disease Control (CDC) better align their efforts. We need to ensure there is coordination between all levels of healthcare including the practitioner, case worker, researchers, and policymakers are aligned in their efforts. As well as coordination from the local level to the national and international levels.

Policy recommendations

Data

  1. State Governors and county officials should ensure that their data on public health is, by default, open and machine readable. This data must be reported in a timely manner to support modeling, nowcasting, and other assessments of COVID-19
  2. The National Institutes of Health (NIH), Centers for Disease Control, Food and Drug Administration should develop an extension to the Precision Medicine Initiative (PMI) All of US campaign to support collection of data on COVID-19 and related issues and make that data accessible to academic researchers and citizen-scientists.
  3. Hospitals, nursing homes, jails, and other congregate facilities should be required to publicly report results of testing patients.

Administration actions

  1. Centers for Medicare and Medicaid Services (CMS) should implement rule making to ensure that electronic health records interact and provide data on COVID-19 related cases rather than having healthcare workers enter the data manually.
  2. The Federal Government should immediately implement and require interoperability for electronic health records.
  3. There should be a concerted effort to eliminate antiquated technology that is a “weak link” in the management of disease such as the fax machine.
  4. Congress should bring back the Office of Technology Assessment (OTA) to support upcoming legislation with the necessary technical expertise.

New investments

  1. A unified testing strategy that focuses on surveillance testing in particular of areas of underserved and minority communities. These programs will need Federal financial support and in return for funding the data must be submitted to a central repository within 24 hours. This needs to be done in combination with a contact tracing and isolation strategy.
  2. Federal funding to build a national data infrastructure to report on critical, time sensitive data. This data should be the “source of truth” and prevent repetitive entering of data.
  3. Congress should support funding for large implementation of contact tracing.

dj patil

Written by

dj patil

Making tech and data work for you

dj patil

Written by

dj patil

Making tech and data work for you

Welcome to a place where words matter. On Medium, smart voices and original ideas take center stage - with no ads in sight. Watch

Follow all the topics you care about, and we’ll deliver the best stories for you to your homepage and inbox. Explore

Get unlimited access to the best stories on Medium — and support writers while you’re at it. Just $5/month. Upgrade

Get the Medium app

A button that says 'Download on the App Store', and if clicked it will lead you to the iOS App store
A button that says 'Get it on, Google Play', and if clicked it will lead you to the Google Play store