Think Digital Innovation: Health Hack

By Jess Sherratt | 03/08/2017

Logos, including NHS, Lancaster University & Amazon

The Think Digital Innovation Hack

We were recently invited to take part in a hack day at Lancaster University's Infolab. We chose to look at ways in which personal assistants, specifically Alexa, LEX and Polly from Amazon Web Services could be used to support independent living in the home. We have already done some internal R&D around the use of Alexa; we are currently looking at smarter approaches towards encouraging sustained behaviour change in the health apps and wearables market and so this event was a perfect opportunity to test out some new ideas along with some brilliant people from across the NHS.

The Initial Idea  

Our experience of hacks suggests it's good to hone in on a problem. As such we focused on the issue of medication adherence. People taking medication have to take repeat doses - often complex combinations - over a variety of time periods, and may require support, in the form of reminders or confirmation that they have done the right thing.

Our group started with the idea of creating a system to provide prompts for elderly people living alone who need to take medication at specific times. This could be through voice activated prompts, two-way back to families/carers who are remote or even the use of video chat to confirm. Medication errors are a significant contributor to unnecessary admissions and the use of Care agencies/Community nursing staff to prompt admissions can be inefficient.

The idea would support elderly people, who might be living on their own or away from their families but who would have regular voice/video contact and could speak/record their actions back to any developed system, app or personal assistant. This interaction data could then be transferred to Health Care Professionals (HCPs) or family members and would allow them to improve medicines concordance.

The concept would also impact care delivery by reducing HCP time and act as an early warning if medicine related problems occurred.

The problem we want to solve 

Lisa Drake who works as Practice Manager at Seascale Health Centre, presented the group with an image displaying a collection of drugs, recovered from the home of two elderly people. There had been a concern they had not been adhering to their medication appropriately and as a worst case scenario, were actually taking the wrong drugs, let alone at the wrong time of the day. 

Table with medicinal drugs

Using Amazon web services

At the start of the hack, we were trying to discuss how we could utilise ‘Alexa’ to have the prompts coming from the Amazon Echo. However, we quickly realised that in order to solve the problem of forgetfulness, we could not put the onus on the user to remember to ask Alexa what they are supposed to take. Therefore, we started to consider the possibilities of using ‘LEX’ and ‘Polly’ to be able to create the two-way interactions we required.

Generating the idea further 

As the developers on our team looked into how ‘LEX’ and ‘Polly’ could be utilised, the group started to consider what would be a useful trigger in someone’s home to set off the voice reminders. We started to consider what people might do as a routine; for example, getting up and brushing their teeth, making a cup of tea or even putting their socks on. This then lead us down to a more connected "Internet of Things" route, which would mean connecting to some form of hub to alert Alexa to trigger a reminder or prompt.

Schedule sketch

Once we decided on relevant triggers for the reminders, we started to consider what needed to be input and output from a concept we were calling the “scheduling system” so that the team understood the goals and features to design and build.

Q&A and Triggers sketch

Developing for Amazon web services

Due to only having a couple of hours to build something in the hack, we wanted to simulate the idea that Alexa can ask the person some questions, which they would need to respond to, so the interaction can be captured in a database. We would intend for it to be uploaded to the scheduling system. Therefore, as a group, we looked into the context and questions we would want the voice reminders to ask, which allowed the developers to generate a quick solution to the Q&A required for the idea.

Red indicated any form of input we thought we might need to enhance the user experience of the voice output in the home. Other outputs from the system were considered; for example, notifications or alerts on the scheduling system as a family view, and feedback into the system that may help doctors identify any deviations from prescriptions that may cause problems later.

Schedule sketch

We considered a way to get the doctor’s prescription formatted in a way that it could be uploaded to the system to generate a drug calendar, in addition to any further appointments required for check-ups in the future. We generated a storyboard to inform how a user may do this.

Drug calendar sketch

This could then be supported with any other personal appointments, such as a book club on a Wednesday, so that the voice reminders would be aware to not notify the user as they would not be at home. Finally, we wanted to add in any activities that may be required to support the user such as “consider having a drink of water” or “why not go for a walk” to remind them of basic activities too. This lead to a discussion around sensors so that the system could identify if someone hasn’t moved out of the same room for a long time or identifying that they were in a particular location in their home.

This was just a hack, so we were playing around with interaction ideas, UI features and data models but we know from our research into motivation and behaviour change that there is a fine line between support and intrusion. Take for example the "Uninvited Guests" video from Superflux, showing the need for patient-centred design and user research: Uninvited Guests 

The scheduling system

All of these discussions lead us to thinking that the Amazon Web Services supported the bigger idea, without being the central point. Rather than be data-driven, our concept would focus on making the family or care workers feel at ease, notifying them when the patient behaves out of the ordinary. This allows a support network to be created quickly based on the patient’s routine.

We then focused on sketching some of the set-up pages that may be required for the scheduling system, so that we could present the idea back at the end of the hack day. We started by considering the prescription screen and times of day the user would need to have the medication. However, we deviated from that plan and wanted to set configurable parameters around times of the day so that the family could define breakfast between based on the patient’s routine. The same with lunch and dinner, which allows the patient to be free without feeling ‘trapped’ by the reminders throughout the day.

We worked through a similar process for the activity schedule and considered what would be required for setting these up. We also considered this for what we were calling life reminders, for example, having a drink, walking about etc.

Schedule sketches

Our takeaways

Overall, the day was a great way to gain insight into a clinician’s view of problems associated with routine medication and the patients’ perspective of what they would be willing to interact with. The Amazon Web Services team provided great support for what could be developed in order to assist with these bigger ideas, and we are exploring ways in which LEX and Polly might aid independent living for people who are mobile but remote from their immediate families or living in sheltered accommodation. It was only a one-day hack but it's great to get together with representatives from CCGs, GP surgeries and care providing organisations, who rarely get chance to share their ideas on the way tech is developed. We understand there are intentions to move forward with the concept, which would be an exciting journey to be a part of.