The diagram below illustrates the spectrum of types of decision support tools supported by the RDS team:
Informational decision support tools
These tools package evidence-based information into decision-ready formats for use at point of care. Examples include RDS guideline and pathway apps.
Knowledge-driven expert systems
These tools use clinically validated algorithms, based on robust, critically appraised research evidence, to process data about an individual and issue recommendations for action. This is one of the oldest, best established forms of artificial intelligence in healthcare, dating back to the 1970s, and the methods for obtaining UK CA marking for this form of decision support are well defined. There is a substantial evidence base indicating that this type of decision support, when effectively implemented, can have a significant impact on practice and patient outcomes.
Knowledge-driven expert systems can operate on a standalone basis - e.g. as mobile or web apps where data about the individual is entered manually. Examples of standalone RDS solutions include the calculator suite and the Q and A decision tree tool.
The RDS platform can also interface with electronic health record systems so that these algorithms can interact with data in an individual patient record and issue recommendations for action in the context of that record. Currently this integration is being piloted with primary care systems and the best approach to integration with secondary care systems is being investigated.
Data-driven AI solutions
This type of decision support typically involves machine learning or natural language processing. It uses algorithms generated by applying mathematical techniques to large datasets. These computational methods “learn” information directly from data without relying on a predetermined equation. The relationship between algorithms and the underlying data is not transparent and therefore these algorithms are harder to validate for clinical use and UK CA marking purposes.
This type of decision support is still new in healthcare and the evidence base for impact is still emergent. The methods for UK CA marking are also still in the process of being fully defined.
The Right Decision Service team engages in a number of data-driven AI projects but does not currently offer bespoke tools for creating this type of decision support solution.