600 hours of NHS staff time to be saved per week with EBO's Virtual Assistant
EBO developed an AI Virtual Assistant (VA) for Somerset NHS Foundation Trust to automate its appointment booking workload. Based on the VA's successful results, £456,000 in annual savings is projected at a target adoption rate of 30%.
Somerset NHS Foundation Trust provides services for NHS in Somerset, England. The trust manages numerous hospitals providing mental, community and acute hospital care across the county.
The Foundation Trust was seeking a digital solution that could take on part of its appointment booking workload, maximise efficiency and bring the number of lost appointments down to a minimum.
Based on the successful results achieved in the pilot study, the Virtual Assistant is projected to save £456,000 per annum and 600 hours of staff time per week (16.9 WTE).
What Goals Did Somerset NHS Foundation Trust Have?
Reduce the large appointment backlog caused by the Covid pandemic and reduce pressure on administrative teams.
Reduce manual workload and temporary staff administrative costs.
Provide a digital method for booking appointments whilst maintaining a positive and personal human-like patient experience.
Dealing with a significantly large appointment backlog
The Covid-19 pandemic brought about an extreme strain on hospital resources and healthcare systems worldwide. The Somerset NHS Foundation Trust Outpatients Department was no different. With less capacity for patient bookings, burnt-out staff and increased workforce costs, the Trust experienced a massive appointment backlog.
The NHS Foundation Trust was seeking a digital solution that could take on part of the appointment booking workload, maximise efficiency and bring the number of lost appointments down to a minimum. They wanted a tool that could interact with patients while still providing a very positive and personal human-like experience.
Patient-led management of appointments
Leveraging the power of Artificial Intelligence (AI) and automation, EBO developed a custom-built Virtual Assistant (VA) named ‘Alex’ to provide a patient-led appointment management experience and introduce AI into the healthcare system.
Fully integrated with the Rio (Servelec) Electronic Patient Records, today Alex can automate patient engagement on a 24-hour basis and empower patients to manage their own health journeys.
By using deep learning such as neural networks, NLP and prediction models, Alex can help patients to get tasks done by processing human inputs and automating tasks, previously only executed by humans. In this way, Alex helps patients to:
- view appointment information
- find their way to the clinic
- manage their booking
- complete an assessment remotely
Emulating human conversation
Throughout development, EBO worked closely with the Trust to ensure that the VA’s language and personality used were in keeping with their branding and culture.
Alex was designed to be indistinguishable from a human when interacting with patients. Thanks to its NLP capabilities, the VA can fully understand patients’ intents and utterances. The VA’s context awareness ability allows it to understand both the content and context of the patient’s query, helping it to manage conversation flows as well as emulate human conversation. As a result, Alex can converse with patients intelligently and with empathy - understanding their emotions, tone and providing appropriate responses.
Should the need arise, the Virtual Assistant also provides the option to speak to a human, allowing staff to take over the chat seamlessly and have full control over the conversation.
Lifelong learning with specialist support
The Virtual Assistant’s machine learning and language recognition qualities give it unique continuous improvement capabilities. These Al elements provide the VA with the ability to learn how to understand questions and provide the correct answers as more and more patients interact with it.
This is complemented by the ongoing service from EBO, whereby specialist dialogue consultants monitor the effectiveness of the VA and create new dialogues when performance improvements are required.
Apart from being available on the Trust’s website, the Virtual Assistant can also be deployed across any of the Trust’s platforms be it Twitter, Facebook Messenger, Slack, Telegram or even Alexa. This provides the opportunity for the Trust to expand and create an omnichannel experience for patients across a variety of channels.
600 hours per week to be saved with a target adoption rate of 30%
The VA was initially deployed for a 12-week pilot study. Since then, the Virtual Assistant has gone from strength to strength, maximising booking efficiency, saving valuable resources and reducing costs.
Out of the total 7,815 booking events (views, cancellations and rescheduling) that occurred throughout the first 12 weeks, 816 were handled by the VA, resulting in an adoption rate of 10.44% and a total of 136 hours of staff time saved.
Extending VA deployment to all outpatient clinics, at the target adoption rate of 30%, will result in 600 hours of staff time savings per week (16.9 WTE) and £456,000 in annual savings.
Today the VA handles 1000+ conversations per month
Since completion of the pilot phase, continuous improvement activities have had a positive impact on Virtual Assistant activity and KPI performance.
Almost one year on and the volume of completed conversations has increased three-fold to over 1000conversations per month with the adoption rate increasing from 10.4% to 14.6% and still rising month on month. This equated to a saving of 615 hours of staff time in 6 months period which is equivalent to 24 hours per week.
Whilst 80% of interactions with the Virtual Assistant lead to a conversation, 40% of conversations occur outside office hours.
A patient satisfaction rate of 86%
The pilot deployment saw very positive results, particularly with patient engagement and patients’ willingness to use digital tools to manage their appointments more easily.
A large proportion of patients stated that they wanted to be able to manage their booking end-to-end through the VA, without having to interact with human agents.
Booking staff also reported reductions in call volumes due to a proportion of the bookings going through the VA. They all unanimously chose to continue using the VA at the end of the pilot.
£456,000 projected annual savings
A broader staff and patient-focused adoption plan will be implemented to improve adoption rates in existing services. Increasing the adoption rate to 30% will increase savings to £159,000/yr.
Once the VA is deployed to all outpatient services’ appointments, the Virtual Assistant is projected to save £456,000 per annum and 600 hours of staff time per week (16.9 WTE).
The increased functionality will be deployed in the next phase of the project to make full use of the VA's capabilities. Such features will include:
- Waiting list verification,
- Patient-initiated follow-up (PIFU)
- Offering cancelled slots to selected groups of patients,
- Notifications to booking staff and clinicians
- Appointment reminders
- Direct booking into EPR system via the VA.
- Monitoring patients in between appointments.
EBO delivers a positive patient experience and reduces booking workload for Somerset NHS Foundation Trust
Trusted by the world’s most innovative businesses
Here’s what the companies we work with say about us.
Andy l’AnsonIT Programme Manager at Shropshire Community Health NHS Trust
The experience of working with EBO on our Virtual Assistant has been very positive. EBO and Servelec listened to our feedback and helped design AI solutions that benefit our patients by providing an alternative way they interact with the trust.
James MariottDigital Change Manager at Somerset NHS Foundation Trust
Our booking officers absolutely love our Virtual Assistant Alex, since it does the legwork and they can focus on the patient. Over the last six months, Alex has helped save 24 hours of staff time every week. Our callers come from diverse cultural and linguistic backgrounds, so we worked with our communication to make sure Alex understands all our patients and the different ways they communicate.