Project lead: Phil Webb
Industry partners: Pfizer Oncology, IBM Watson
Velindre University NHS Trust
This Bevan Exemplar project developed the world’s first ‘virtual assistant’ trained in Oncology.
Supporting patients as they live with cancer is a key requirement for a cancer centre. Over the past 2 years, Velindre Cancer Centre has been having meaningful conversations with patients, carers and their families about how best to support their information needs and support their desire to want to have a measure of control, empowerment and independence as they make their care choices.
A key area of need was the gap in between existing services; the lack of tools to have good quality conversations, anytime place or where; access to relevant quality information for patients to support their choices and decisions about their emotional, psychological and physical wellbeing. A general feeling across cancer patients is that they find it difficult to have good conversations about living with cancer when they want, where they want and at a time convenient to them.
This project developed the world’s first AI (Artificial Intelligence) -enabled virtual assistant trained in oncology to proof of concept. This dialogue agent, RiTTa (Realtime Information Technology Towards Activation), currently is trained to answer a small number of ‘intents’ to demonstrate capability.
A key challenge in the development of this AI is the time, care and resources required to develop and train the machine learning (ML) layers.
RiTTa has been trained to discuss and answer questions on a broad spectrum of areas. RiTTa can have conversations with a huge amount of people at the same time – 24/7, anyplace anywhere to support them taking back control of their lives.
In Oct 2018, Velindre Trust Board agreed to move from ‘proof of concept’ into the technical development, scaling and deployment of the project.
An evaluation framework has been constructed in conjunction with KESS PhD Swansea Centre for Health Economics to consider key outcomes including:
improvements in activation and health literacy
reduction in anxiety
Technical work will include expanding training sets for ‘intents’ and ML in:
pharmacy and medicines safety
general non-complex clinical care: chemotherapy and palliative
tumour specific care: breast, lung, palliative
Part of cohort Bevan Exemplar Projects 2017-18