Another Major Milestone – Turning off COVID Testing

I am on my way to California to attend a two day gathering of Higher Ed leaders. It was approximately 3 years ago when COVID struck us big time, caught us unawares and all hell broke lose. I admire the swiftness with which the senior administration at Wellesley took action. Having a respected physician, Dr Paula Johnson, as the president helped and especially her leadership in a group of Higher Eds in Massachusetts.

I am also very proud of the fact that we, in LTS, are one of the very few organizations who decided to take on the technology responsibilities needed to support our COVID testing and related matters without needing a paid vendor solution. And I think we did a great job and by making the right choices, we saved the College money and served our community better. What we did was a tremendous amount of work and describing all will take a long time, so I will describe just a few highlights here.

After the initial steps to vacate the campus by sending our students home and pivoting to fully online (about which I have written earlier), our collective attention turned to some of the next steps. I am thankful to have been included in the senior leadership group what was strategizing this and to be designated as a member of several key subgroups.

One of the earliest decision we made was to open the campus for a small number of students, though the classes would be held online. This also meant that we needed to offer COVID testing. We needed to make a decision as to whether to go with a paid technology solution from a vendor who had built systems to talk to Broad Institute’s systems. I was not a fan of this because it required a lot of integration and we didn’t have some of the controls we needed. Then we heard about a solution from Tufts University that will work with Broad systems. I had a zoom meeting with Tufts folks and was very impressed with what they had already built. I then asked if Tufts software can be made available to other liberal arts colleges and arranged a zoom session with my peers. I don’t think any of our peers were interested, but I knew this is where I wanted to go.

I made my recommendation to the senior leadership team and it was approved right away and we began working through the logistics. There were two aspects to it – front end and back end. Since there are so many personally identifiable information involved, I also wanted to restrict who worked on these. So, it was one programmer, Charlie Gao (who left the College a few months ago)  who handled the front end exclusively and I handled all the backend programming. We needed to get all of this done in short time.

Front end required laptops running Linux OS and relatively cheap printers to print labels (from Amazon). The original system required every tester to enter their information on a web page, but we deemed this to be too slow. So, we added a twist by attaching a scanner that read the mag stripe in the College provided ID card and filled in the rest of the information. Tufts folks helped make sure this all worked fine. Our staff were able to quickly pull together enough hardware etc and get it all set up on time.

While this is being set up, I was busy with building the back end, which had to do with testing logistics. The committees were converging on two tests a week for students on campus and once a week for employees. We needed to know how many testing locations and stations are sufficient to avoid long lines. Tufts team had built a nice simulation for this where you plugin a set of parameters and it produced the desired number of testing stations based on the number of tests, most likely time our community is likely to go to test and what amount of wait are we willing to tolerate. These kinds of simulations are exciting for me because Monte Carlo simulations used here was my bread and butter for graduate work. I ran this for multiple scenarios and presented the optimal number of stations to the senior team to start building them.

Another interesting simulation that Tufts researchers had built was, given the number of students arriving from various states, how many COVID positive cases can we expect on arrival. I ran the Tufts model and MIT model and presented that to the senior team. Whereas the  predicted #s were not the same as reality, some trends (such as the states with the most and least cases) were pretty good.

Then came the hard part. Scheduling everyone! Never easy. This was an algorithm that I had to keep refining more often than I anticipated. Basically, the challenge is to take the total population that is being tested and the frequency and distribute them amongst two test centers that are at two extremes of the campus. But there were constraints. I needed to account for class meeting times when students and the faculty cannot test. We also asked everyone to indicate which test center they preferred and I needed to accommodate this. The algorithm worked remarkably well. Yes, there were some assigned to a location that was different from what they preferred, because the algorithm weighted the balance of testers to reduce lines. We also built an interface for users asking to reschedule because they had other commitments and we accommodated these as much as possible.

There was a LOT of communication, website updates and dashboards. I will spare you the details, but there were a lot of constantly changing requirements and code development in the back end. All done in PHP. And then we used Google sheets very effectively to show data to the right group of people.

I vividly remembered the very first positive case that came in. I was in a Zoom meeting with some of my LTS colleagues and I was very nervous and left the meeting to make sure that all the right people were being notified on time. The senior leadership had given the team a very short time between a positive result coming in and isolating the person. There were some bumps in the road initially, not because of technology, but confusion about who should be doing what.

We also provided a ton of useful data to the contact tracers and others in the team. For example, when a student tested positive, contact tracers needed ot reach out to their room mates, if any, instructors of their classes (initially not a big deal because it was all online, but subsequently when it was in person it made a difference) and those registered for the classes. All of these were automated and made available.

A handful of times, Broad results came late. This really made those who tested very anxious. Everyone who had a negative test promptly received an email and only those tested positive didn’t. They were called on their phone and advised on next steps. When someone didn’t receive email, they felt “may be I am positive”. So we had to sent out communication about delays.

Subsequently, we had so many add ons – isolating students to a hotel, properly calculating their release dates and communicating it to the students and support staff, Broad results came in very late or very early during the day and we did not want to unnecessarily add to the tension late at night or early in the morning, so we held the positive results from being sent between 10 PM and 6 AM (something that vendor solutions didn’t easily support from what I had heard).

And all of our communications were thoroughly vetted and customized for our community. We also built a quick process in Workday to allow our employees and students to upload proof of vaccination, which was manually inspected for accuracy. If it was incorrect, we had another communication that went to the users to upload correctly.

I can go on and on, but you can see all the things that went on behind the scene. But I will mention one more thing and stop. At some point, we decided that students who are repeatedly not testing needed to have their one card deactivated so they cannot gain access to buildings that required one card. This is another automation we built by integrating to our onecard system. Upon testing, the onecard block was removed.

Over a period of time, especially in the last few months, things were working fine and no need for additional modifications, so they needed less and less of our time. And basically turned off everything this past weekend.

Though COVID itself continues to be with us, it is different! But I am happy to move on to doing other things. I feel great for LTS contributions to the excellent way that the senior leadership handled the COVID crisis at the College.

We could not have succeeded if not for the support from the senior leadership and the great folks who were members of the various teams. COVID brought us together to collaborate in ways that I have never seen before. The shared purpose was clear and everyone was working to support our community. I also want to extend a BIG THANK YOU to the staff members who oversaw testing. Can you imagine how tough a job that is for three years? They did it so well, patiently working with everyone, explaining things calmly when things were not working as expected etc.

1 Comment on Another Major Milestone – Turning off COVID Testing

  1. Sarah Allahverdi
    April 19, 2023 at 1:16 pm (1 year ago)

    Thanks very much for this informative reflection of where we’ve been and all work that went into getting us through it.

    Reply

Leave a Reply