COVID-19: Where can technologists help?

13 min readMar 18, 2020

As most of us hunker down in our homes hoping to steer clear of COVID-19, the public health officials on the front lines are doing their best to keep out this microscopic monster and buy precious time while the rest of our health care system readies itself for the inevitable onslaught.

No one knows what the future holds. We might be super lucky if the virus mutates to a less potent form or there is a weather change that curbs the spread. On the other hand, most of the reliable data emerging out of China, Southeast Asia, and Europe so far points to an exponential growth (with a reproduction ratio > 2) that is going to severely burden our already decrepit health care system at least in a few densely populated urban areas (Bangalore, Delhi, Mumbai, Pune, Hyderabad).

There is a lot for us to learn from the countries that have had more experience with coronavirus infections over the last few months. I desperately hope our government and public health administrators are actively consulting them for support. Post the 2014–2016 Ebola crisis, there were, in fact, concerted initiatives by the US government, WHO, and NGOs such as the Bill & Melinda Gates Foundation to better prepare for future pandemics and all those learnings¹ are worth revisiting.

Talking to my sister and friends who are part of the Indian medical community, I feel there is more that all of us, especially technologists could be doing right now beyond protecting our own immediate families and communities. Some of the harder problems such as genetic analysis and antigen design require specialized expertise, but there seem to be quite a few other areas where most of us could contribute. A lot of people are already sharing and implementing great ideas. Here is a list of few more that came up in a conversation with a friend.

1. Self Assessment Tools

Should I allow our kid to go to his TT class ? Would that be ok if my mom who lives with us has asthma?
After a trip from Italy, should I tell my maid and cook? What about neighbors? What if I had just returned from Delhi?
How aggressive should we be on disinfecting door knobs?

The right answers to these questions are not straightforward and the trade-offs depend on two principal factors.

Vulnerabilities of the individuals we interact with (i.e., likelihood of intensive care or fatality if infected)
Likelihood of being infected

Figure 1 shows a coarse split along these two dimensions and broad recommendations for each scenario. The suggested behavior for a person depends on the most vulnerable person in his or her family or living circle (e.g., dorm wing in case of hostels, roommates).

Clearly, what this implies is that each of us should be able to readily assess ourselves and everyone in our living circles along these dimensions to make judicious choices. Medical tests are going to be prohibitively expensive and slow to be administered in a country like ours. A self-assessment tool or a mobile app that can approximately estimate these two scores for an individual based on user provided responses would be extremely helpful in figuring out how to prioritize testing and making decisions related to isolation and care.

The medical vulnerability score would depend on factors such as

age
gender
lifestyle (e.g., smoking, sleep deprivation)
prior medical history

The likelihood of infection, on the other hand, depends on the above four factors as well as additional signals such as

observable symptoms (self assessed or by an online medical professional)
user’s exposure to specific spatial hotspots (e.g., traveling to domestic airport)
user’s contact with other individuals with a high likelihood of infection or confirmed infections
generic user activities (e.g., traveling in a bus, going to market)

Note: In a number of cases, infected people do tend to be asymptomatic. Observable symptoms have to be factored into the assessment, but are not necessary conditions.

With inputs from medical experts and recent clinical studies emerging from China², it is possible to put together simple scoring models (e.g., decision trees or linear models) that could have reasonable discriminatory power that can help with better allocation of medical testing and care resources. As more data is available, the models can be improved and refined for finer-grained recommendations.

To address privacy concerns, the computation could be fairly primitive (e.g., a simple decision tree or a linear model) and performed entirely on the client side itself.

2. Trusted Status Sharing & Tracking Tools

Singapore, Taiwanese and Hong Kong authorities attribute their current success in containing the spread of COVID-19 to their strong health surveillance and protocol enforcement systems. In contrast, our current approach of publishing patient trails in newspapers and requesting people to self-report and self-quarantine is woefully inadequate.

It is extremely unlikely that India can put together an equivalent surveillance system in the short-term. Besides, there are also fundamental privacy, data ownership, and personal liberty issues at stake that have to be considered prior to implementing intrusive data collection processes.

However, it is possible for mutually trustful groups of people (e.g., residential communities, workplaces, educational institutes) to do certain level of self-policing by sharing their vulnerability scores, likely infection scores, and/or GPS information in a limited fashion amongst themselves. Even if there is resistance to voluntary information sharing among informal communities, it might become necessary for organizations responsible for essential services (healthcare, groceries, transportation) to mandate such sharing for their employees and the users they serve (e.g., patients in case of health care organizations).

Individuals could also share their scores along with GPS information with aggregators (e.g., Google Maps) who can in turn use it to generate warnings if someone approaches a hotspot or a person who is highly vulnerable or has a high likelihood of infection. To address privacy concerns as well as potential abuse and errors in self-reported data, it might be preferable to generate warnings for a user only from the data collected from his/her trusted group. Aggregators can also collect similar anonymized data from reliable medical authorities to detect hotspots across the world.

3. Predictive Forecasting Models

Most of us have heard about the aphorism about “skating to where the puck is going”. Given the immense difference in outcomes with the right planning and precautions, it is paramount that we forecast the likely spread of infections on both the spatial and the temporal dimensions. Such forecasts requires epidemiological models that account for the intrinsic characteristics of the infection as well as the interaction dynamics and spatial movements specific to populations in different regions, which can be influenced by public policy. The Institute for Disease Modeling has been developing sophisticated open source models and software tools, but primarily focused on diseases such as malaria.

Simulations based on these forecast models can enable “what-if analyses” for different policy choices and assist policy makers arrive at the right decisions. Even simple models such as the SEIR model² can go a long way in insuring against drastic failures. It would, however, be desirable to a have model that forecasts all the following key variables for each spatial region of interest every day.

#new people exposed to COVID-19
#new people that need to be tested (will depend on the testing policy)
#new people actually tested
#number of new infections
#new patients requiring hospital isolation
#new patients requiring ICU care
#new fatalities
#new recoveries

Forecast models (at district and state-level) that can be periodically fine-tuned with actual observations can be valuable for hospital and local administrators, especially because exponential growth often runs counter to human intuition. There are already a number of websites³ tracking some of these variables across the world. It will be great if these sites could also host forecasting and what-if analysis tools for policy makers.

4. Streamlined Allocation of Essential Resources

Misallocation of essential resources during a crisis is often as dangerous as the intrinsic challenges themselves. In Bangalore, most medical shops were out of masks and sanitizer even before the confirmed cases in the country hit triple digits. The scary but now likely scenario is that our health care and sanitation workers who desperately need these medical supplies might not have access to them because of unnecessary hoarding by paranoid rich folks.

That is why, to the extent possible, it is critical that the essential resources are managed by trusted authorities and communities in a transparent way and intelligent way. While designing computer systems, we are often mindful of scalability, efficiency (latency, throughput), availability, and fault tolerance concerns. There are a already a number of well-studied, simple, yet effective techniques for load balancing, caching, rate-limiting, and geo-distribution of resources. It is imperative to adopt the same approach while designing our real-world distribution systems for essential resources such as food-items, medical equipment and even health-care expertise. This holds true even in the absence of a crisis, but it is especially so in the midst of one.

Moreover, e-commerce companies such as Amazon and Flipkart and logistic companies such as Delhivery already have sophisticated systems and geographically distributed warehouses to efficiently distribute goods across the country based on real-time demand. Our government requires similar capabilities in order to ensure that vital supplies and personnel are directed to the right region at the right time in proportion to the requirement.

A pre-requisite to facilitate improved allocation is an up-to-date country-wide digital inventory of medical facilities, essential medical equipment (ICU beds, ventilators, etc.) and personnel. If such an inventory doesn’t yet exist, it should be expeditiously created in a crowd-sourced fashion using existing online collaboration tools. Forecast models of typical medical conditions along with those of the new COVID cases can then be paired with optimization algorithms to determine better allocation of resources.

5. Tele-Medicine

Reports emerging from China and Italy point to three salient challenges in deploying medical personnel listed below, all of which fortunately, allow for technology-based telemedicine solutions.

Infection exposure of medical personnel

The odds that the medical professionals themselves get exposed to COVID virus during the testing and the treatment process necessitating their own quarantine are very high. In addition to jeopardizing the health of medical staff, such exposure would also drastically diminish the system capacity.

“Forward triage”, i.e., assessing patients remotely before they arrive at the emergency department was central to China’s strategy for handling COVID infections and protecting their medical personnel. It is also an efficient means to deliver at-home medical care to patients with mild symptoms while minimizing physical interactions and further spread. Unfortunately, in India, we do not yet have a well-established infrastructure to enable remote medical consultations paired with payment and scheduling tools.

Some Indian companies such as Practo do offer these facilities in a limited fashion, but these applications lack consultation features (e.g., collection of vitals, images) that are commonly available in the expensive Teleconsultation tools offered by the established healthcare software companies such as EPIC. Open-source versions of such tele-health tools can significantly simplify the triage process and, at the same time, expand access to medical care for a broad segment of our population.

Sub-optimal utilization of medical personnel

Due to widespread panic, considerable time and effort of medical staff is spent on addressing common queries from patients with mild symptoms instead of attending to more needy patients. Intelligent multilingual voice assistants and chatbots that can accurately answer common patient queries on testing, symptoms, and related topics can considerably reduce the load on the medical staff. Vital information can also be proactively pushed to the population via information platforms.

Lack of adequate expertise in intensive care delivery

The novelty of the COVID disease and the limited number of well-trained medical personnel makes it challenging to deliver intensive care to a large number of geographically distributed patients. There is also a need for care providers to rapidly share their learnings and fine-tune treatment protocols & policies rather than work in isolation.

Tele-ICU monitoring platforms that employ web cameras, sensors and monitors hooked to a central system address these issues quite effectively, and are, in fact, the norm in developed countries. Unfortunately, in India, there are only a few organizations such as CloudPhysician that are pioneering this approach. Improvisations based on (possibly multiple) mobile phones that can enable even a subset of this functionality without any additional hardware installations can potentially make a huge difference.

6. Effective Utilization of Information Platforms

Empowering individuals with the right information is crucial to harnessing our collective actions in fighting this global crisis.

Broadcasting of vital information to everyone is a no-brainer, but the current scenario demands that we go beyond this one-size-fits-all approach and perform more aggressive tailored interventions. Social media companies have long been honing targeted advertising methods to create and deliver the right message in the right language & format at the right time and the right place to encourage the desired user behavior. A similar approach is likely to be highly effective in nudging people in the right direction. For example, it would be useful to send an automated warning message to a person approaching a hospital about the risk of exposure to COVID to self as well as the medical staff and the right protocol to follow. Likewise, appropriate messages could be sent at the time of entry to a place with high people density (e.g., bus stations) or residential areas. For additional effectiveness, it might even be preferable to propagate this information on group-forming networks (e.g., WhatsApp) in a hybrid semi-automated way with human moderators. Rapid semi-automated translation of useful messages into multiple languages is also a necessity.

In addition to dissemination of useful information, it is also equally important to curb the spread of misleading or false information by malicious players on the popular social media platforms. Even though the Big Tech players have promised a strong focus in this direction, there are likely to be multiple region-specific variants of misinformation which makes detection a non-trivial challenge.

7. Automation and Monitoring devices

Since physical contact is a primary mode for the spread of COVID, automated or semi-automated no-touch devices for monitoring temperature & proper use of protective equipment (e.g., masks in hospitals) and controlling equipment (e.g., operating elevators, dispensing sanitizer) can aid in limiting the number of infections. There already exist tools for some of these tasks, but these have to paired with appropriate visual recognition models or remote semi-automated monitoring by humans. Accounting for individual privacy in choosing the right trade-offs while deploying such tools also remains a germane area of research.

8. Design & Manufacture of Medical Equipment

Everywhere across the world, there is already a serious shortage of ventilators and respirators, which are vital for saving lives. The only way out of this dire situation is to rapidly scale up the production of these devices all over the world. Unfortunately, such specialized manufacturing expertise is limited to few countries with the right know-how. To address this gap, there is now a big rush to collaboratively create and test open-source simple 3-D printing designs, which can be then be used for mass-production by anyone. A number of hackathon websites are also hosting medical themed challenges to incentivize alternate low cost software and hardware designs for vital equipment using commonly found parts.

9. Tools for Remote Collaboration

“Social distancing” is a core component of the global strategy to address the COVID crisis. What this entails is a huge shift in how we work and collaborate with each other. Thankfully, over the last decade, there have been tremendous advances in online communication technologies that are proving to be indispensable now, but there is more to be done.

To keep our society functional and sane, over the coming months⁴, it will be imperative to enable people distributed across multiple locations to collaborate and socialize in a seamless fashion as before. We will not only have to move meetings online, but also take our coffee breaks, and parties online.

There already exist tools for project management (task assignment, scheduling), instant messaging, video conferencing, but there is ample scope for further improvements to make the experience as close as possible to simultaneously being in the same physical space. Since much more human communications are likely to be recorded, it also opens up possibility of reusing these recordings for learning and assessment. At the same time, it also raises concerns related to individual privacy and data ownership that have be to be addressed thoughtfully.

10. Improving Network Connectivity

Critical to all the above ideas is our core communications infrastructure, since it forms the very foundation of the mobile and internet ecosystem. The shift to remote operation is bound to not only significantly alter network usage patterns, but also result in an urgent demand for increased bandwidth. Even though this is a primary area of focus for large telecom and internet service providers, there exist multiple areas of innovation (e.g., transmission of video data, low cost LANs for neighborhoods)

Given the concentration of medical personnel in urban areas, effective delivery of medical information and care to rural areas is contingent on connectivity. Thanks to advances in the telecom sector, there has been a big jump in mobile phone penetration in India over the last few years. Even so, it might necessitate additional innovations in mobile ad hoc networks (e.g., mobile hospitals)

Note: Frankly, the above list just explores the tip of an iceberg (possibly a lopsided one too!) and is nowhere near exhaustive⁵. With or without COVID, most of these ideas are worth our time and effort. For long, the existing power structures and decision-making protocols of society have skewed our priorities and squandered human effort and ingenuity. Even though the human impact of the COVID pandemic is heart-breaking, it is also a wake-up call for all of us to assess the nature of the work we do, how it affects all of us, and come together to overcome this adversity. As Churchill once declared, “Never waste a good crisis!”

Footnotes

¹ Informative talks and interviews by Bill Gates on pandemics

² Some informative papers and resources on clinical and epidemiological studies of coronovirus from Wuhan:

³ Some useful epidemiological and planning models (need to be customized for use):

⁴ Authoritative case trackers:

⁵ Coronavirus lockdowns: How long do we have to live like this?

⁶ A friend commented that technology and data-driven interventions in the financial and banking sector also deserve high priority. Securing single points of failure (on communication networks, electricity grid) with backups should also be a focus area.