The assistant on the smartphone

Several European CGM teams are working on AI-supported tools for processes in hospitals and medical practices. Dr. Jana Schmidt coordinates their work and therefore knows exactly which factors contribute to success.

It's something we're all too familiar with: Healthcare professionals spend up to 60% of their working hours documenting their work instead of attending their patients. An absurd waste that we should put an end to as quickly as possible. But how? The CGM is convinced that AI can form part of the solution.

These days, we already offer our hospitals and medical practices a range of AI-supported tools for their processes that go beyond radiological image data management. The new service we have developed in a European team to facilitate documentation is called “Voice2Docu ASSIST.” Hospitals in Poland and Spain are currently testing this in pilot operations. The idea behind it: One starts the function directly in the HIS form, and as soon as one starts speaking, the AI begins its work and saves the information on the form - after being checked by the relevant healthcare professional.

The CGM HEALTH ASSISTANT recognizes what the users want

This service is part of a vision we refer to as the CGM HEALTH ASSISTANT, and which we also want to put into practice as quickly as possible. It is based on an agent system that users can experience as an app with a streamlined interface. When I tap on the app icon and start speaking, the system uses pattern recognition to identify what I want - differentiated according to roles. After all, doctors, nursing staff, controlling, and, of course, patients have different requirements for such a tool.

Sounds too good to be true? If that's what you're just thinking, I will agree with you in one respect: We developers should be careful not to raise exaggerated expectations - and we should always remember that AI only makes sense if it delivers tangible, preferably measurable, added value. Here at CGM, we have defined a number of specific KPIs for this purpose, for example, “Accuracy” and “Consistency.” And when, as was the case, our solution initially only recognized 40% of the drug names correctly, accuracy was simply not adequate by any means. (Meanwhile we achieve 85%.)

Exploring the limitations of AI is just as important as validating its benefits. And anyway, regular feedback is essential for any AI implementation. Providers should never forget that the true experts on what happens in a hospital or doctor's practice are the people who work there. This makes it all the more important to regard ourselves as a team and to evaluate and continuously improve solutions together.

When it comes to AI, it's best to start on a small scale

In my experience, this means that managers in healthcare facilities should start with AI on a small scale, with one or only a few services, in one department, and over a reasonably short pilot period. Try to build confidence and trust among users, take obstacles seriously, and review the solutions with users. In a recent survey at a social congress, I witnessed how concerns about data protection and a fear of job loss were cited as being the main worries about AI. Such concerns must be taken seriously.

And by the way: The fact that our pilot hospitals are not located in Germany is due to organizational coincidences and in no way due to the notorious German hesitancy when it comes to digital innovations. I don't get that impression in my discussions; in fact, the interest in AI solutions is very high. And should you belong to those who would simply like to try out whether, for example, “Voice2Docu ASSIST” could bring added value to your institution, we look forward to hearing from you. Simply call or write to us. 

Dr. Jana Schmidt is responsible for AI development in the field of hospital information systems at CompuGroup Medical (CGM) – with a particular focus on the CGM HEALTH ASSISTANT. To this purpose, she coordinates several European teams, each of which develops use cases or also skills for the CGM HEALTH ASSISTANT. Her objective is to develop the various services in a secure and transparent manner and to make their quality measurable.