Amidst the COVID19 pandemic, tracing the source of a viral spread is one of the keys to containing the disease. However, according to an interview with a contact tracer specialist at Singapore General Hospital, it is extremely laborious and manual. Even though many techniques have been applied such as using guiding questions to trigger patient’s memories, receipts from taxis and restaurants as proof, even asking the patients’ close relatives, contact tracers still need to rely on the patients’ words as the source of truth, making it possible to lose some information and cause an accidental viral outbreak in a country.
What it does
Project TraceON aims to solve the issue by tagging individuals to checkpoints, and allows the health organizations of countries and communities to collaborate and improve the traceability of contagious diseases. The automatic tagging of individuals automates the need for health declaration forms, as information of the individuals is automatically recorded in the system alongside health screening facilities in public and private sectors alike, eliminating the need to sort through hundreds of digital and hand-written forms, freeing up manpower and time of an organization.
Project TraceON achieves this following a 3-step process. Specific spots in a region can be installed with TraceON camera modules. Whenever a person walks by, the TraceON camera module triggers a TraceON event and sends a sequence of frames for a Machine Learning model to do automatic tagging of an individual. The Machine Learning model runs a pre-trained Convolutional Neural Network in the cloud, designed to recognize specifically human faces, whereby FaceHashes are generated for each unique face identified, whereby a new FaceHash will be created if an unknown face is detected, maintaining the anonymity of a person while making it possible to trace back to an individual. The results of a TraceON event are saved in a database securely, whereby authorized operators of TraceON can view past event histories that happened near a TraceON camera module for contact tracing and health declaration purposes.
Should a contact trace of an individual be required, TraceON will produce a report for contact tracing specialists a list possibly affected individuals based on the event stride, scientifically defined by a combination of virus airborne survival time and whereabouts of the person within a disease quarantine period.
How it was built
TraceON was built with scalability, availability, security, and the ease of replication in mind using a combination of IoT, Machine Vision, Artificial Intelligence and Cloud Computing concepts. Scalability and availability are achieved through the use of several cloud technologies such as AWS Lambda, DynamoDB, and S3 buckets, allowing both the creation of FaceHash and the access of web interface to be event-driven, discarding the need to provision resources. Next, security is achieved through detection of a person's face through open source and high-accuracy (99.3% in Labeled Faces in the Wild dataset: http://vis-www.cs.umass.edu/lfw/) Convolutional Neural Networks (CNN), and tying each unique individual with a unique FaceHash, maintaining anonymity and consistency in tracking the individual. Since TraceON is an end-to-end solution, we wanted the hardware to be easily replicable, so that perhaps bigger companies, organizations, and even the makers' community can replicate and generate a quick and dirty Proof of Concept (POC) to play a part in combating global pandemics now and even in the future. Meanwhile, the cloud technology and architecture of the solution is transparent for users to view its architecture as a means to gain trust from people around the world.
We are a group of polytechnic students that participate in various hackathons. However, this was the first hackathon that forced us out of our boundaries and worked away from one another. Communication was inevitably a big challenge we faced. Next, as two of us are having our internships and one of us is serving the nation, it was difficult to juggle and manage our time. This was considered one of the larger-scale projects given the short amount of time we had, and we hope to not only continue to improve it but also look for possible investors to work together to realize TraceON to better prepare the society should a global pandemic occur.
Accomplishments that we are proud of
We are proud that we were able to observe the current state of COVID-19, how it has affected the world, and apply CDIO (Conceive, Design, Implement, Operate), a design thinking framework practiced at Singapore Polytechnic to identify root causes to problems we have observed from the pandemic. Furthermore, we also think of possible pivotable scenarios from this pandemic, to come up with a well-thought-out, viable, accessible, and feasible end-to-end solution that is as general as possible to tackle a real-life problem. We are also happy to open-source this project, or seek government/private organization support to realize TraceON, all code of which is available on GitHub, and ready to be deployed.
What we learned
During our brainstorming session, we looked at a wide range of ideas to combat against COVID19, such as telemedicine, online learning platforms, online games between neighbors, community sharing of information, self-checking systems for traces of COVID19, etc. As a budding software engineer, we wanted to implement a community-driven solution, as we believed that the most effective and efficient solution is one that involves the efforts of every individual. As a community, we have the capabilities to help the government’s efforts in combating the pandemic and we urge communities, organizations, and makers alike all around the world to selflessly contribute simple, easy to use, and easily deployable solution, such that everyone can do their part in combating COVID19. While it is important for the existence of a good solution, we believe that it needs to be enforced by the government, policymakers, and lawmakers to their utmost abilities for the said solution to be effective in any country. We believe that through the collaboration between government, organizations, communities, and makers, we as one can successfully defeat the challenge that is COVID19, and even future pandemic.
What's next for TraceON
Currently, in the market, there are many tracing and tracking solutions but are too expensive to be implemented in a pandemic situation such as COVID19, as this pandemic disease has proven not only to disrupt the healthcare industry, but also the economy, tourism, and possibly every other industry, causing major disruption and instability to everyone in the world. The said solutions usually use proprietary, closed-source hardware which requires modifying said proprietary systems to achieve a high level of integration with the cloud, deterring the time-to-market of said systems, missing the critical time frame to curb said pandemic diseases. Furthermore, the security of people is a great concern when it comes to implementing such a system. As a form of sincerity, we aim to open-source TraceON, from the hardware used, Machine Learning model used, cloud architecture, all the way to the method of creating FaceHashes to maintain the anonymity of individuals in the system.
Try It out
amazon-web-services, html, machine-learning, python, raspberry-pi, react, swagger, typescript