Inspiration and Introduction
Surge in Elderly Living Alone
According to the World Health Organization (WHO) Global Health and Aging studies, we are facing an unprecedented circumstance: The world will soon have more elderly than the young, and more people at extreme old age than ever before. It is estimated that there will be approximately 1.5 billion elderly people aged 65 years and above by 2050. The ageing of the world population will continue to increase, because of the declining fertility rates as well as the incredible improvement in life expectancy.
With this global trend comes the accompanying fact that their has been a surge in the number of elderly living alone in most countries. This fascinating insight is affecting all types of societies be it whether modern or traditional, around the world - none being spared. About 40 percent of women in european countries above 65 are living by themselves. And if we were to dive into even those more traditional countries such as Japan, large family living styles are becoming less popular.
Increase in home deaths
Based on studies conducted by the U.S. Centers for Disease Control and Prevention, that about one in four Americans above 65 years of age fall every year, and an elderly person is treated in an emergency room for a fall every 11 seconds. To top it off, every 19 minutes, an older person passes away due to falling. Many a time, these falls occur in places where they are unable to seek help and are alone. These remarkable statistics point towards a single direction: Something needs to be done to provide support to these people, there is clearly a gap that needs to be closed.
It is also worth mentioning that due to the recent global pandemic, the number of home deaths have increased, some of which have lay undetected for weeks. For example, authorities in Detroit have responded to more than 150 “dead person observed” calls and city officials in New York are recording more than 200 home deaths per day.
Statistics and global trends aren’t the only factors highlighting the absolute need for a solution - it is also first hand experience as well. Being a Firefighter myself (shoutout to Station 11), part of the work I do involves Rescue Locked Door incidents. These rescues could be triggered by a distress alarm, or a concerned neighbour. Many a time, we arrive at the homes of people days after their deaths only to find that they could have been saved if only aid was provided.
With this being a global concern, there is a need to find a way to address the surge of home deaths and put into place preventive measures in a scalable and sustainable manner. Currently, there is no solution that is cost efficient enough to fit all elderly homes, and that is also at the same time intelligent and proactive. There is an obvious gap and an urgent need for a smart and sustainable solution that would help to save lives.
Solution and What It Does
Our solution is RescueMe, an intelligent proactive door lock that is both voice activated and has a motion sensor which helps to detect danger. The intelligent proactive door lock will listen for phrases that indicate danger, which is sent to Facebook’s Wit.AI Speech API for intent processing. The motion sensor on the lock acts as a detector for non-movement.
When the lock detects danger, not only will it unlock the door, the intelligent lock will also send a notification in the form of a WhatsApp message to neighbours and loved ones so that they can check on the person in danger.
We trained our Wit.AI with utterances that best resemble someone falling down. You may find sample voice tracks in the folder wav, if you wish to test it out.
The utterances were linked to the intent, call_help.
After Wit.Ai detects that the speech is, it will proceed to call the servo, which will open the door.
Here it is in action
How We Built It
The Particle Photon accepts sound input from the Adafruit microphone and converts it into a wav file, using a set of libraries. The node js server sends the sound byte to wit.ai’s speech API, which returns an intents. If the intent returned is suspected of danger, it will send a POST back to the photon to unlock the door through WIFI. A whatsapp message will also be sent out through the twilio API. Alternatively, the motion sensor will detect non-movement for 30 minutes, after which it will trigger the unlocking as well. Open source libraries that were used include “node-witai-speech", 'particle-api-js' and “twilio”.
Other Power Sources Omitted For Brevity
Challenges We Ran Into
A Mix of IoT Hardware and COVID-19 : The lockdown in our home country Singapore resulted in us having to work remotely, which provided extremely challenging due to the involvement of hardware since our solution is an IoT one. Therefore, we divided our tasks between hardware and software and held many zoom meetings and virtual debugging sessions.
As neither of us are experienced with IOT hardware and technologies, there was a steep learning curve in figuring out how the components pieced together. It was also our first time exploring wit.ai, however the documentation provided was robust and the sharing sessions hosted by Facebook proved useful. It was the feeling of excitement of creating something, and knowing that we really had this end vision in mind that kept us going. We knew that we just had to do it.
Accomplishments That We're Proud Of
Cost Efficient and Easy Adoption : When we built the solution, we kept in mind that we wanted it to be easily implementable and made accessible to anyone and everyone. The intelligent lock we created would only cost US$30 whereby we are able to provide a not only intelligent but also proactive solution - making it truly cost efficient, especially in competition to its counterparts that may only nail the digital part but is not proactive at all. This means that it can be easily adopted to bring safety to everyone, no matter who or where they are.
Quick Installation drives Business Value : On top of that, installing the intelligent proactive lock only takes about 5 minutes to fit onto a door. We felt that this is extremely essential to our solution as we wanted it to be able to market easily for maximum impact. It is also not difficult to use. This solution can also be fitted onto any existing lock to drive the same results.
Commitment to Passion Project : We were able to juggle our work to work on this passion project hackathon of ours, while also at the same time collaborating online. We are proud of ourselves for committing to working on this project and putting together the solution in a limited time frame.
What We Learned
We learnt how to use wit.ai and the potential it has for other use-cases. Will definitely be exploring and adopting it to other projects in the future!
What's Next For RescueMe
3D Printing Design : We’re also looking at using 3D printing of the case design to hold the IoT parts of our RescueMe solution to enhance the user experience and replace the makeshift set up we have right now.
Multiple RescueMes, One Central Hub : Beyond just having one RescueMe intelligent lock per home, the next step we are looking into is deploying multiple RescueMes in places such as the toilet and bedrooms that talk to a central hub. This would help to ensure safety throughout an entire compound, especially that of a larger one, or even a building in a synchronised manner.
Sentiment Analysis : To make the model more robust, we would incorporate sentiment analysis as well into RescueMe. By differentiating between different trigger words and understanding context, it would increase the accuracy.
https://www.who.int/ageing/publications/global_health.pdf https://www.theguardian.com/world/2020/jun/07/uk-coronavirus-victims-have-lain-undetected-at-home-for-two-weeks https://www.reuters.com/article/us-health-coronavirus-elderly-insight/coronavirus-spreads-fear-isolation-death-to-elderly-worldwide-idUSKBN2172N8 https://www.ncoa.org/news/resources-for-reporters/get-the-facts/falls-prevention-facts/
Particle Photon, which includes an ARM Cortex M3 microcontroller with a Broadcom WiFi chip. Adafruit Microphone MAX4466 DFRobot Metal Gear Micro Servo HC-SR501 PIR Motion Sensor Miscellaneous (Breadboard, jumper cables, locks)