How Data Helped Us Defeat COVID-19
Technologies that helped non-pharmacological disease control measures.
“Be safe, be smart, be kind”, Dr. Tedros Adhanom Ghebreyesus.
During the COVID-19 pandemic, many measures were implemented in various aspects of our daily lives, forever changing the direction of society and our way of seeing things. Disease control measures were applied by all world governments, and guidelines on their effectiveness were vehemently discussed. However, we are not here to discuss such guidelines, but we will explore the details of how data science and disruptive technologies helped to qualify, quantify, execute and monitor such measures.
Measuring Human Mobility
“Without equity, we cannot end COVID-19, HIV or any other pandemic.” ― Peter Sands, from the Global Fund.
Data scientist were able to utilize mobile phone data, GPS data, social media content, public transportation data, satellite imagery, and retail data to measure changes in human mobility. These sources helped researchers track shifts in travel patterns, commuting behaviors, and public movement, aiding in understanding the effectiveness of lockdowns and social distancing measures.
For that to be sucessfully applied, data scientist needed to create a new mobility dataset combining all mobility patterns, gathering selected mobility data pools, such as mobility reports, mobility signals collected from mobile phones, and geographical information derived from social media.
For example, using the Twitter streaming API, data scientist gathered tweets related to COVID-19, focusing on location-based information. The dataset comprises 3,768,959 Twitter users and a total of 469,669,925 tweets originating from the United States only.
Figure 1. Visualizing mobility patterns in the U.S. (A). general human mobility patters; (B). seasonal mobility patterns; ©. Driving pattern segmentations.
Social Distancing Monitor
“Home is a shelter from storms — all sorts of storms.” ― William J. Bennett.
Managing social distancing was one of the biggest challenges faced during the pandemic, as it is a factor that depends not only on government entities to propose public safety measures, but also on citizens themselves to become aware and take the right action. to protect yourself and those around you.
The goal of scientist to help with this issue include incorporating advanced methodologies like big data analytics, machine learning, deep learning, and computer vision, utilizing automated systems to detect individuals and assess adherence to social distancing protocols through surveillance cameras.
A range of object detection algorithms, including pre-trained models like R-CNN, SSD, YOLO, and more, have been employed to create a social distancing monitoring system. This deep learning-powered social distancing monitor can play a role in reinforcing non-pharmacological disease control strategies by identifying instances of social distancing compliance within crowded environments.
Figure 2. Social distancing monitor test results. (A). single individuals detection and identification; (B). single individuals social distancing monitor; ©. coupled persons detection and identification; (D). coupled persons social distancing monitor.
Face Mask Detection
“In the face of adversity, our unity becomes a shield of protection. COVID-19 may have kept us apart, but our strength together knows no bounds.”
Amidst the Covid-19 pandemic, the significance of the face-mask policy has escalated due to its potential to slow down the transmission of the disease. Since the pandemic’s inception, individuals have been required to wear masks in public spaces. Adhering to this regulation is obligatory for accessing services from both public and private providers. Nevertheless, effectively monitoring compliance with this mandate remains a challenge. With the advent of artificial intelligence, this issue can be addressed by leveraging deep learning techniques, which excel in recognizing visual features within images and real-time video streams.
An efficient network-based model was introduced, designed to operate on embedded devices with minimal computational demands. The complete network architecture was developed using TensorFlow and Keras, incorporating image augmentation and the conversion of class vectors to binary class matrices during data processing. Ultimately, OpenCV was employed to assess and execute face mask detection for both images and real-time video streams.
Figure 3. Test results of the face-mask detector. (A) sample detection output; (B). training and testing epochs v.s. loss and accuracy curves.
Future Research Directions
“The secret of change is to focus all of your energy, not on fighting the old, but on building the new.” ― Socrates.
Future research directions should encompass diverse viewpoints in the realms of human mobility tracking, social distancing monitoring, and face-mask detection. To extend research horizons, the human mobility dataset could benefit from additional dimensions like relative mobility indicators gathered from varied data sources such as Bluetooth, WiFi, and QR codes. Enhancing the social distancing monitor’s efficacy entails the inclusion of supplementary attributes like body temperatures, exposure risks, and factors related to Covid-19 control. While integrating face mask detection into the system is plausible, challenges arise with real-time detection in crowded settings due to model limitations, warranting exploration of alternatives like the use of tiny face detectors. The surge in digital technologies’ role in social system monitoring prompts ethical concerns regarding security, privacy, data management, over-surveillance, and adherence to ethical guidelines. Hence, the technological framework of social control measures should factor in a spectrum of ethical considerations. Furthermore, the impact of social factors like gender diversity, Sharia-law clothing regulations, and totalitarian countries’ social credit systems could reshape the social control landscape, providing avenues for future exploration.
A special thank you
To the healthcare professionals who stood strong on the frontlines,
To the essential workers who ensured the smooth functioning of our society,
To the scientists, researchers, and innovators who raced against time to develop treatments, vaccines, and solutions,
To the volunteers and community members who stepped up to support those in need.
Your acts of kindness have shown that even in our darkest hours, humanity’s spirit of generosity shines brightly.
With the deepest gratitude and admiration,
Rubén Acevedo.
References
- Peng Zhao, Yuan Ren, Xi Chen, Shanghai Dianji University, China, Beijing University of Civil Engineering and Architecture, China, Big Data Helps for Non-Pharmacological Disease Control Measures of COVID-19.
- Wang J., Encyclopedia of Data Science and Machine Learning, 5 Volume-set, 2023.
- Dalkiran, M. (2020). LFW Simulated Masked Face Dataset. https://www.kaggle.com/muhammeddalkran/lfw-simulated-maskedface-dataset
- Feyisa, H. L. (2020). The world economy at COVID-19 quarantine: contemporary review. International Journal of Economics, Finance and Management Sciences, 8(2), 63–74.
Author
Rubén Acevedo is a Data Scientist with a passion for the idea of using technology to address critical challenges in the fields of Science, Economics, Social Causes and Healthcare.
Contact me: ruben.fernando@pucpr.edu.br for questions and suggestions :)
Thank you!!!!!!!!
