Inspiration
In England alone, more than 2 million people aged 75 or above are forced to live by themselves. In the event of an emergency seeking help for an older person would be a difficult, if not impossible, task to achieve. It is estimated that more than 36 million older people fall each year resulting in approximately 32,000 deaths, many of which could be avoided if help were available immediately. Another frequent cause of death among the older population is due to heart attack, where access to prompt medical assistance is essential to save the person's life. The available commercial solutions that rely on the "SOS Button" mechanism are prone to failure if the person becomes unconscious or debilitated and, especially among the older individuals, it might appear to be unintuitive to use. VirtualAid aims at solving both problems by creating a "passive SOS" notification systems that uses input from the patient to determine whether help is needed.
What it does
The systems scans the environment to detect if the patient needs help and potentially issues a notification to a list of "emergency contacts". VirtualAid employs speech recognition and computer vision algorithms to assess the patient status. In particular, a camera is used to detect whether the patient has fallen down and a microphone is used to scan the audible conversations for any call for help. If an anomaly is detected by the system, a notification email is sent to a list of "emergency contacts" that are registered in the system's dashboard. The users that are notified are then able to hear or see the recording that triggered the notification to determine the severity of the accident and to act accordingly. In addition, the dashboard allows the user to keep a record of all the emergency notifications that have been issued, along with the event that have triggered them.
How we built it
The system dashboard frontend has been developed in react.js, while the speech-recognition and computer-vision clients have been developed in python. All the backend features that include media upload/download, user authentication, automatic email dispatch and real-time database has been developed using Google's Firebase service. The real-time database acts as a bridge between the python clients and the system dashboard. When an anomaly is detected by the clients, the database is updated and a custom function dispatches the email notifications. At the same time, the media containing the recording of the event is uploaded to the storage server. The data is also accessed by the dashboard when visualizing the events history.
Challenges we ran into
As a team made of people hacking from across the world, we found that working remotely has been the main challenge. Effectively communicating and collaborating at a distance has proven to be exceptionally difficult at times: what in a more "normal" scenario would have required a few people to look at the same screen, this year it took several online tools (e.g. Discord, VSCode Live Share, GitHub and Figma) to achieve the same result. In addition, managing time and effectively allocating the tasks among group members has been particularly challenging without physically being altogether.
Accomplishments that we're proud of
We are proud of solving an important problem that affects several million people around the world. In fact, we believe that VisualAid truly has the potential to make a difference and to save lives. We are also proud about the fact that VisualAid can be employed on low-cost and low-power devices such as the RaspberryPi, making it available to virtually everybody that has access to an internet connection.
What we learned
While hacking we have learned how to effectively work as a team at a distance. We have managed to effectively use the online tools to get the work done, while keeping a direct and continuous communication channel. In addition, some members had the opportunity to deepen their knowledge in frontend development in react.js
What's next for VirtualAid
In the future other vital signs such as body temperature, heart rate and breathing rate could be monitored by the system. This would allow VirtualAid to be used with critical patients that require constant monitoring of their primary vital signs. In addition, more complex algorithms that track multiple patient parameters at the time could be developed in order to improve the accuracy of the system.
Log in or sign up for Devpost to join the conversation.