Efforts to use artificial intelligence in social care should start small and include the early adoption of data standards, according to public sector IT association Socitm.
It has published a series of guides, How AI meets social care, to coincide with its virtual President’s Conference this week. They are broken down to cover the technical, legal and commercial, professional practice and trust elements of the issue, with another document providing an introduction.
Socitm says the guidance is based on the experience of key stakeholders and should act as a resource for social care providers in adopting the technology.
The technical guide places a focus on ensuring the interoperability of AI in existing digital architecture to support the integration of health and social care, along with the sustainability and ease of use of the technology.
It says it would be unreasonable to assume that social care teams could roll out a fully integrated solution at first attempt, pointing to earlier guidance from the Local Government Association that councils should initially work on just a few use cases. Along with this is should follow the SNOMED CT data standard for healthcare to enable vendors to develop interoperable systems.
“As the position of AI within social care is less mature than it is in healthcare, there is an opportunity for social care organisations to unify behind a clear set of codified standards at an earlier point in the sector’s AI journey,” it says.
The National Digital Social Care Group, of which Socitm is a member, is currently working on closing the data standards gap between the two sectors.
The document also places an emphasis on steps to make solutions sustainable over the medium to long term, including the use of a target operating model, flexibility in their design, security from cyber threats and ensuring they will be supported over their lifecycle. These are accompanied by a need to engage with frontline professionals in the development and adopt accessibility standards to make the solutions accessible and easy to use.
The procurement guidance highlights that AI is currently a low regulation market in which authorities are playing catch-up with system developers and says this creates a ‘buyer beware’ environment.
To deal with this, social care providers need to establish some guiding principles for procurement, such as those in the Digital Ethics Charter, establish good governance and manage the lifecycles. The latter involves defining what a good solution looks like but being willing to change this over time, and making this clear in an invitation to tender.
The document on professional practice looks at minimising the disruptive impacts, pointing the importance of careful change management that faces up to the cultural shift in an organisation. This should come with the adoption of agile methodology designing services that use AI, sharing the relevant skills and bringing frontline practitioners into the efforts to procure and implement the technology.
It also deals with how AI should empower social care practitioners in decision making, saying it should augment and complement their insights rather than substitute them; and how the technology can help to provide more personalised and effective care. The latter needs a clear view of best practice and the areas of care to be affected, and efforts in training and communications, the guidance says.
The document on trust highlights steps such as busting myths around the use of AI, ensuring that it complies with information governance and data protection regulations, and making sure there is transparency and accountability. There is also scope for professional networks of social care practitioners to share lessons and best practice, and a need to advocate the use of the technology.
“Socitm argues that the expansion of AI into social care is best thought of as a movement generating social good,” it says.
“It needs to be clear that the social care sector is not adopting these technologies simply to cut costs, nor to remove the agency of the staff. Rather, the possibilities of enabling more personalised and tailored care with practitioners assisted by AI to better inform their decision making in relation to the care citizens receive need to be presented.”
Image from iStock, Ipopba