Artificial Intelligence (AI) and Machine Learning (ML) are two of the most rapidly growing fields in technology. AI is the ability of machines to learn and perform tasks that ordinarily require human intelligence, such as recognizing objects or understanding natural language. ML is a subset of AI that uses algorithms to enable machines to learn from data. In this article, we will explore how we used AI/ML to develop a COVID-19 contact tracing application for a federal health agency so they could track their employees effectively during the pandemic.
Scenario:
Due to COVID-19 Pandemic, all the employees at this federal health agency were encouraged to telework. However, some employees had to show up on-site to provide emergency services. Our team was tasked to identify the employees who are accessing the government building during COVID-19 times and provide data analytics and contact tracing. In case an employee is diagnosed with COVID-19, it was critical for the government agency to quickly identify and track all potential contacts to contain the outbreak.
Solution:
When employees scan their Identity card at any turnstile gate, the card information is reported to us in PDF format. We used Python, PDF extraction, and Fuzzy matching algorithms to identify Card holder’s information and pull their organization information from the Employee Directory database. Fuzzy matching was necessary because the cardholder names reported in the PDF file are not formatted properly and often come with special characters like underscores, backslashes, etc. We used AI/ML to track and monitor which employees were coming to the office on which day and time and while at the office, who they were interacting with.
When a person reported COVID-19 positive to the agency, we use AI/ML to quickly trace back all the potential contacts this COVID-19 positive person came in contact with at the agency in the last 14 days. We categorized the potential contacts into 3 categories based on the closeness of the potential contact to the infected person i.e. high risk, medium risk, and low risk. E.g. if the potential contact worked on the same floor as the infected person they were flagged as high risk, and, if the person worked in the same building but on different floors and could have come in contact with the person in common areas such as the lobby, restroom, cafeteria, etc. they are put on medium risk. Once the system has the data of the potential contacts and is categorized into the 3 categories, we flag those employees in the system accordingly and temporarily disable their physical access to the building. Further, our application uses the employee profile to pull the contact information and send mobile/email alerts to the potential contacts to get tested for COVID-19 immediately. We also alert the respective supervisors for awareness and any work transition planning.
As part of monitoring, we used AI/ML for identifying patterns in data such as, if an employee presents themselves to the health agency on a Monday and then again on Friday, we used AI/ML to predict when a person will be coming to the facility and send appropriate alerts i.e. to not come to the office without RT-PCR test, etc. If for any reason, the potential contacts or the person infected with COVID-19 tries to enter the building they are denied entry till a valid RT-PCR test is shown. We reuse the AI/ML techniques explained above to extract the data from the RT-PCR document and update the status of the employee so he/she can regain access to the building. A final mobile/email alert is sent to the respective person and their supervisors confirming the employee’s access back to the building.
As part of this project, we also built various visualizations charts/graphs/ reports to track and report to the leadership on a daily/weekly basis.
Conclusion:
AI and ML can be used to help identify potential contacts, as well as track them. To do this effectively, the right AI techniques must be employed to accurately identify potential contacts and get desired results.
In this article, we showed how we used AI/ML and Fuzzy matching techniques to accurately identify potential contacts for COVID-19 for a federal health agency. Our team understands how to leverage the most advanced technologies such as AI/ML to solve everyday problems. Please contact us at info@neevsys.com to discuss how we can bring AI/ML solutions to modernize your business.
Author: Shainan Patel, Managing Director, NeevSys Inc. (www.neevsys.com)