Education IT Reporter

Responses from Amanda M. Melvin, lead public health analyst, HSR.health.

Geospatial data is information that describes events, objects, or features within a location on or near the surface of the Earth.

The role geospatial data plays in the safe return to school has to do with where the school is situated in addition to where the students and teachers who attend that school live. This information allows us to use geospatially encoded data surrounding the COVID crisis to determine the relative safety of a school. We also consider the surrounding area of the school, the students and teachers, and environmental data pertaining to the school itself. All of these data points are involved in determining the probability of disease transmission within the community and to understand how best to mitigate the associated risks.

Can AI be used reliably to track and identify the risk of disease transmission in a school environment? What about new variants of COVID-19?

In short, there is little AI can’t do. Using AI, data scientists are able to develop new solutions to critical world problems on a daily basis. For example, data scientists are now able to leverage broad sets of geocoded data and machine learning models to identify the risk of disease transmission continuously across various circumstances and environments.

In regards to identifying the risks associated with the unfortunate rise in new variants of COVID-19, yes, AI can be used to reliably track and identify these. A defined place like a school and individual classrooms have specific air flow patterns that can be modeled to test viral propogation. The movements and interactions of teachers and students within that space can also be modeled to assess the level of potential exposure to infectious diseases.

This goes for both known and novel variants of the virus that causes COVID-19. Anticipating the next dominant variant is critical to an effective response.

In identifying risks in a school setting, how do geospatial data and AI work together?

Geospatial data and AI work hand and hand to paint a complete picture of the risks in a school setting. Geospatial data informs the initial conditions of this analysis. By leveraging data that represents the ZIP Codes or home counties of the student body, teachers, administrators, parents, and others who may frequent the school, the model can accurately capture and inform an accurate representation of that school or district’s specific COVID-19 situation.

For example, AI can analyze the events of a school day to assess the risk of disease transmission in each classroom. This is done through a combination of Agent Based Modeling (ABM) to assess the interaction between students, teachers, administrators, and any visitors within a school environment. Here, the people in and around the school environment are the agents. At HSR.health, we also utilize our own adapted version of the Wells-Riley Epidemiological Model for infectious disease spread.

What specific metrics or numbers can be extracted from geospatial data to alert school administrators and local health departments that there is a risk?

The metrics need to be easy to understand and actionable. They have to convey whether the school/classroom environment is safe when all mitigation measures are considered. With truly actionable data, students can learn and teachers can teach with the confidence of knowing that they will be safe and healthy.

Once a risk has been identified, what are some cost-effective mitigation strategies that schools can implement?

By identifying the key pain points that exacerbate disease transmission risk, schools can be cost effective by focusing on the issues specific to their environment. The key with any mitigation is to reduce the potential for exposure to the virus. The specifics of the risk itself will tell us how best that risk can be mitigated.

Truly, mitigation strategies do not have to be expensive. Mitigation strategies that schools can implement once a risk has been identified include the use of masks, social distancing, and improved school HVAC systems. Simply opening windows to let fresh air into the classroom can help. Some schools have opted to implement a hybrid learning model, in which students learn in-person 50% of their time and remotely the other 50%. We’re all navigating this new normal and want children to be in the classroom.

And can geospatial data/AI track the effectiveness of these strategies?

Geospatial data and AI can track the effectiveness of these strategies at an impressive level.  Once mitigation measures are implemented, we can track and actually graph their effectiveness over time. This can provide insight for any necessary modifications and result in the best possible solution for every school environment’s unique case.