This project was based on a
brief titled AI100 for the Royal Society of Arts Student
Design Awards. Our task was to design for “How might we use
AI to support people to reach a happy, meaningful and
productive one hundred year life?”.
We identified the issue of Social Isolation among older people and solution of creating a dynamic AI-powered database of services through interviews with the older people and experts in the medical field. Once we had a prototype of our solution, 'Grace'. We tested every aspect of it with real people.
Early on in our project we identified a growing issue, greatly impacting the ageing population. Social Isolation. More than 2 million people over 75 live alone and more than a million go over a month without speaking to a friend, neighbour or family.
One of our research boards
To further investigate the importance of communities and connections,
we talked to a variety of older people at different stages of
independence on what life is like for them and what issues they
might be facing.
Through this we found that what kept coming up was how vital close friendships and communities were to them. Any physical issues or troubles they might have were rarely mentioned. This is what led us to investigate the social issues that older people are facing.
Example of older People Interviewed
To ensure we had tackled the brief from every angle, we
reached out to several key individuals with different areas of
expertise pertaining to the health and care of older people
Dr David Robinson introduced us to the innovative practice of Social Prescribing which is an alternative means of treatment and a way of enabling GPs and other primary care to refer people to a range of local, non-clinical services professionals
Healthcare Experts Interviewed
People who engage with Social Prescribing have shown significant
improvements in health.
There has also been a push in recent years for GPs to use the MECC approach, or Make Every Contact Count.
Unfortunately, since Social Prescribing is a relatively new practice in the UK and Ireland, there is currently no robust infrastructure in place to support it.
We were told of how there is a severe lack of awareness when it comes to knowledge of local services in the area. This in turn meant that when trying to prescribe a service or activity to someone, it resulted in making a thousand phone calls to find out what is available around them.
And so we found that the biggest challenges that were slowing down the use of Social Prescribing and MECC were that firstly there was no centralized resource for finding suitable services efficiently, and secondly a lot of the information about services found online is outdated.
Setting goals of what we need to identify
Dr. Chris Soraghan and Dr. Gerard Boyle, from St. James Hospital in Dublin, shared with us the Local Assets Mapping Project (LAMP) which aimed to solve these issues by providing a list of local services in a small area of Dublin.
In their case they had a team of students cumulatively spending
several thousands of hours, manually walking around to find out what
was going on in order to build a database, and within several months,
most of these services were no longer correct.
A GP was no longer able to safely recommend activities to people, because the risk of prescribing someone to an activity that no longer exists is serious; you basically lose someone.
Ideating on how we can address this problem
This showed us there is a need for such a platform and that it is possible. However it needs to be approached differently to make it work effectively.
Rough Grace concept map
Grace meets the need for a live database of services and activities
that isolated people can engage with, and provides a platform that GPs
and health professionals can use to create social prescriptions.
Grace does this by using natural language processing, connecting with
services and people where they are.
LAMP was built for a small locality and still wasn't successful, ours could be deployed virtually anywhere for a very low overhead cost since it is a software driven solution.
Service Map Detailing How Grace Works
During a routine check-up, if a GP feels social prescribing would be
suitable for an individual, then a Link Worker or GP will use Grace to
find and prescribe to them a list of activities or services.
Grace would later call the individual to confirm which one they would like to join. After which Grace would call the service provider and book them in.
Rough Service Blueprint
Service Blueprint and User Journeymap
She begins to do this by scraping the internet, for postings on social
media and other websites listing services,then grouping them into
certain categories, creating an initial list of services.
Once Grace has this list, using a Natural Language Voice engine similar to google duplex, she emails or calls to check and verify the details of the service.
Those details are then turned back into a live database that can be
read by GPs or anyone else to see what’s going on in their local area.
Grace informs the core users in two main ways. One is through a website which can be used by healthcare providers as part of their social prescription consultation, enabling them to make every contact count, as well as by older users themselves, their families or caregivers.
The other way Grace interacts with people is by using a technology they are already familiar with: a phone call, using said natural language processing to talk to them in a language they understand.
For the Voice User Interface, script making was an on-going agile
design iteration process, every new version of the script we made
we tested up to 10 times, calling a service provider while posing
as the AI, iterating then trying again.
Testing was done with a soundboard with our pre-recorded lines to force us to stay on script. Even with low-fidelity prototyping and a mediocre text-to-speech engine we were successful in obtaining service information.
Testing the VUI prototype with Service Providers
There are 3 different scripts for the 3 different types of call Grace would make. A data verification call script, a script for notifying an individual of their social prescription and a script for Grace to call the service provider and book the individual in.
3 different scripts created as a guideline for Grace to follow
Designing the website for Grace, we took into consideration NHS Design Principles and accessibility guidelines to make sure our design would suit the intended use, putting extra care into the font choices to ensure good screen legibility.
And just as we were using the agile design process to iterate the Voice User Interface, we did the same for the website interface. Continually testing, tweaking and co-designing with a GP and Link Workers.
Testing the website with Link workers
Testing the website with Doctors
What benefits does Grace bring to the healthcare system?
What benefits does Grace bring to people and society as a whole?