Coronavirus Quarantine Application Specifications

When we commit as a country to strictly enforce isolation of coronavirus patients and quarantine their close contacts, we will need an efficient way of monitoring the quarantined people. Borrowing a proven method from Taiwan, we can use a mobile phone app. This is potentially a very powerful, relatively inexpensive tool to keep good people honest for 14 days or possibly longer. It’s difficult for anyone to be so isolated for so long, and yet so necessary to keep everyone safe. This tool is to assist and provide support for the patients while allowing public health officials to easily enforce quarantines on lots of people.

Coronavirus COVID-19 Quarantine Software Application Idea

We must get this epidemic under control so we can resume work and school. We cannot be in a holding pattern for months, with the end date pushed back every few weeks. It’s impossible to plan anything, and it’s impossible to restart our economy. The only way to get back to work is to seriously suppress the epidemic. And the only way to suppress the epidemic is to find everyone currently ill and quarantine them and their contacts. We will soon have the ability to perform mass testing, so we just need to ramp up our ability to strictly isolate the cases.

We would be naive to think that people will stay in isolation for 2–6 weeks without a lot of encouragement and support. Without enforcement and an infrastructure for help, even the best of us would go stir-crazy. We can incentivize staying in a quarantine while having the tools to check and enforce it. An app can help.

I heard about the Taiwan quarantine method in passing — something about the government handing out phones to everyone who’s quarantined and requiring people to answer the phone. Since Taiwan only has 300 cases, this is feasible. I don’t know many more details, and perhaps they have learned some great insights over time. We should definitely study their methods, but we would need to write our own software for the United States.

Here is what I envision for the app.

Setting up:

• When someone tests positive, a person from the health department should call with an app registration code issued by the government (either state or federal). This case code would help track each patient until an outcome is reached (recovery or deceased).
• The patient would be required to download the app and put in the code and answer some questions verifying identity, which is linked with the person’s social security number.
• Upon registration, the patient would set up voice identification and facial recognition.
• The patient would also set up the location they plan to stay at for the duration of the illness.

Quarantine enforcement:

• The purpose of the app would be to keep the patient in a defined area.
• The patient’s phone would send location information every minute. With geo-fencing, public health officials would be quickly notified if the phone left the allowed area.
• To confirm that the patient did not simply leave their phone at home, they would be required to check in every few hours (randomized), perhaps two to six times a day.
• The app would send an alert asking the patient to check in.
• The patient needs to respond within two minutes of the alert. The requested response could be a call back to an automated system or a selfie.
• If the patient does not respond within two minutes, a public health employee is notified.
• The public health employee would call the patient. Perhaps the patient had the phone alerts on mute or was taking a shower. This would give the patient an opportunity to fulfill the check-in.
• If the patient does not fulfill the check-in requirements within ten minutes, following a defined rubric, the public health employee may ask the police for a welfare check.
• If the patient left the quarantine, they could be fined or jailed.

App access:

• Location information. The allowable radius (geo-fencing) can vary based on population density. If there are reliable data sets for housing or apartments that are more precise than a simple radius, we should use that.
• Camera. We can tap into the phone’s facial recognition software to tell us if the selfie is of the patient or not. If the built-in software is not able to verify the photo, the app will give suggestions for lighting and show examples of how to take a better photo. The app will give multiple opportunities to submit a real-time photo that passes facial recognition. If after two minutes the patient is unable to fulfill the facial recognition selfie check-in, the patient can send the best photos for human verification. Beards and poor photography technique can cause problems. A public health employee can then attempt to correct the technique through education.
• Accelerometer. In combination with the location information, computers can tell if a patient is active, driving in a car, or if they haven’t moved for a long time.
• Alerts. This is required for the random check-ins.
• Health (optional)
• Photos (optional)
• Contacts (optional)
• All information entered into the app would be uploaded to a server for public health workers to review if necessary.

Different sections of app:

• Check-in selfie camera
• History (previous check-in times and selfies taken)
• Contacts log (to log locations and contact names for the health department to trace)
• Symptoms tracking (optional, but this can use AI to provide forewarning to caregivers that someone may be taking a turn for the worse)
• Mental health support (relaxation videos, relaxation tips, hotline to call, exercise videos, meal ideas)
• Supplies support (running out of food or supplies)
• User settings

Contacts Log:

• Once a patient registers, they can fill out a log for their contacts.
• There would be prompts for them to review their calendar, GPS, and photos for the past 14 days.
• If access to photos is allowed, the location information for the photos can be pulled and reviewed with the patient. Perhaps they took a picture at a park. Or maybe they were in the car waiting for someone when they took a screenshot of something funny. Location information can help someone retrace their steps to identify potential contacts.
• Other apps with location check-ins such as Facebook can be reviewed with the patient as well.

Symptoms Tracking:

• Patients can fill out a checklist of symptoms and be encouraged to keep a subjective journal.
• If they choose, this information can be compiled into a PDF report sent via email to the patient’s loved ones or even to the patient’s doctor.
• If they choose, this information can also be condensed into an image that they can share on social media. Building in some fun and social interaction can help with collecting valuable data from patients.

Mental Health Support:

• Relaxation tools should be easily accessible here.
• A hotline should be available and staffed by social workers to help guide the patient through stressful situations.
• Calisthenic exercises or yoga videos may help reduce stress.
• Quick and easy meal or snack ideas.

Supplies Support:

• A patient can request common supplies in this section. If a patient does not have people who can leave food, toilet paper, or other supplies at their door, public health workers may bring donated supplies to a patient. Remember that we have to support the quarantined people so that they don’t feel compelled to leave the quarantine.

Artificial Intelligence (AI):

• Computers can make predictive models for patients who may need to be hospitalized in the next 24 hours by reviewing some key data sets.
• Review the symptom checklist and scanning the diary for key phrases
• Review accelerometer data for the level of activity.
• Use advanced photo software (server side) to see if the patient has suddenly lost weight, which may indicate dehydration.
• Changes in usage of the mental health section can indicate increased stress. AI can learn which relaxation videos to recommend depending on stress level. AI can also learn if someone is likely to break quarantine or not.
• Most check-ins would occur when there is accelerometer activity so that we don’t disturb a resting period for the patient. Eventually AI would learn the patient’s activity routine and notice deviations. Deviations could mean a decline in the patient’s health or it could mean that they left the quarantine. AI can help raise red flags.

Advanced features:

• The Apple Watch is rumored to have pulse oximeter capabilities with the next iOS release. When this becomes available, this is another critical indicator of someone who may need to be hospitalized. All physicians agree that <88% is way too low and requires immediate hospitalization. 89%-92% oxygen saturation is in the danger zone and may require hospitalization if there are other risk factors, such as heart disease.
• The Apple Watch can do a single lead EKG tracing and track heart rate over time. If access to the Health section is made available to the app, the system can make recommendations if danger signs are detected, such as increases in average heart rate over several days, or sustained tachycardia (heart rate over 100 beats a minute).
• Android systems may have similar capabilities. (I’m most familiar with the Apple ecosystem.)

Bonus:

• Consider “gamification” with badges, fun/benign competitions for quarantined folks, creating a feeling of community, and celebrations for recovered people.

When the patient feels better, they can be re-tested to verify that there is no longer detectable virus in their nose and throat. Once a recovery outcome is assigned to the case code, that case is completed. Alternatively, if the patient dies, that case outcome is entered and the case is completed.

If a patient requires hospitalization and eventually the ICU (where they will no longer be able to interact with an app), a public health employee can then take over the case code and track progression and outcome. There could be these various designations on the side of the public health worker:

• Confirmed case (This can be from a positive PCR test or a doctor’s recommendation, perhaps from characteristic CT findings and/or a positive ELISA antibody test.)
• Close contact (The contact may or may not be symptomatic but meets CDC criteria for significant exposure with a confirmed case)
• Hospitalization
• Intensive Care Unit
• COVID-19 death (final outcome)
• COVID-19 recovery (final outcome)

If the app is uninstalled prior to a final outcome, the police may be contacted. If the app is tricked in some way (having someone else “do time”), the patient could be fined and jailed. There should be real, serious consequences for not complying with the quarantine.

Much of the app should be automated, with logical escalation protocols with public health employees prior to involving law enforcement. The manpower required to run this can be laid-off administrative workers who can even work from home with the right tools and a secure VPN connection. The bulk of the work will be in contact tracing, screening the contacts for significant exposure, and then getting those people signed into the app. The public health employee would need their own browser-based workflow, with lots of explanatory standard operating procedures. The workforce will be lots of new hires and people working from home, so having clear procedures is imperative.

Contact tracing is a huge part of the public health responsibility. This app can help with that by making it easy to provide the information. Most people keep their contacts on their phone already, and as they lie sick in bed, they can still use their phone to provide information. Privacy is and should be a huge concern for Americans. This app can be easily uninstalled after completing the quarantine.

I could not find a quarantine enforcement app on Help With COVID, but it looks like some people are trying to port Singapore’s contract tracing software. On GitHub, someone is working on a customer relationship management (CRM) for managing contacts in quarantine and observation. There was a Hackathon, but no one worked on a comprehensive quarantine app. The database behind this system would be moderately elaborate and must be very secure, since it has both government ID as well as medical data. It would be ideal if all of the features I described above could be integrated into one platform, in one app. Both California and New York have lots of bright developers and entrepreneurs who could pull this together. If done well, much of the tasks in public health would be automated. The sooner the government uses it, the sooner we can all get back to our normal lives.

Clearly, I am not a software developer, and I have no UI/UX or graphic design background. I am a doctor turned entrepreneur with a strong interest in epidemiology and public health. I was once offered a job by the CDC as an epidemic intelligence officer. Since I am at the intersection of many disciplines, I am offering the limit of my thoughts for public health experts and software developers to expand and make real.