As a company founded in an era where speech recognition was not the norm and was followed with sceptics and criticism, we have always been on the forefront looking out for new technologies and how to apply those in our products. Artificial Intelligence has been on our watch-list for years, and we have been evaluating the maturity of the technology and the various areas of application. Do we need to investigate deep learning technologies? Or should we look at text mining or language understanding technologies? Or focus on cognitive services (amongst others, the field of speech recognition)?
Each have a very different focus and different characteristics and make it suitable for a completely different application.
As a customer, how can you tell the real difference between all of the companies tumbling over each other to market their story about how good their AI solution is? Considering new technology, one of the most important evaluation criteria we use is: ‘Is new technology going to add value to the end user, or to the patient, or to the total process?’. In the end new technologies and automation should make our lives easier or better, and if it doesn’t add value immediately, it is probably not a great idea.
At the moment, we are collaborating with various AI suppliers in the field of medical imaging and making an interface to exchange information. The results of the AI software (measures, outcome, coding, etc.) will be put into the report automatically, eliminating the need for dictating the results already on your screen, or worse, manually copy and pasting.
The introduction of AI and reusing results into the report nicely aligns with another topic we have been talking about for years – Structured Reporting!
Structured Reporting can support the user in three different areas:
- Quality of the report by standardising elements and order of information, and checking for completeness of the report by mandatory or suggested fields;
- Accelerate the reporting process with standard text phrases;
- Use clinical coded data for research and analytics purposes and the structured exchange with other information systems (e.g. Transfer of Care).
And when introducing any of the aforementioned technologies, it is important to question what you really want to achieve. We need to set strategic goals together and decide how we want to get there and what technology to apply to achieve that goal. Implementing technology is never the goal, only the mean. Delivering better quality or the reuse of clinical coded data or working more efficient is the goal. And does it matter what technology it really is? I don’t think so. It just needs to work!
When designing a new process and discussing this with customers and end users, it is clear it needs to fit with the existing workflow. The pressure on users in healthcare is so heavy, that there is simply no time for additional tools or applications. It needs to be such a seamless part of the working habits, that it saves time. Acceptance and adoption are key, and you need buy-in from the end-users.
Key takeaways:
What do we learn from this? If we successfully want to improve the workflow and look for new technology to achieve this, then follow this recipe:
- Set clear goals. Do not mix up the goals with the means (introducing AI is not a goal)
- Prioritise your goals and clearly focus on what you want to achieve
- Design the workflow to embed the different systems where possible and maximise the added value of the tools by eliminating manual steps in the process
- Get buy-in from your user-base. Acceptance and adoption are key!
Happy to get your feedback. Please leave your comments or questions to start a discussion.