Rewriting in healthcare

How did the idea of rewriting in healthcare come about? 

Since OpenAI made ChatGPT available to the general public, the software has caused quite a commotion. In our last trend report, we wrote about how the technology works and what opportunities there are, but also what challenges exist. One of the challenges we mentioned at the time is the fact that ChatGPT predicts what the next word should be. The model looks for words that ‘logically’ belong together, which can sometimes lead to hilarious sentence constructions.  

However, this drawback won’t cause you any problems if you ask the model to translate or rewrite a particular sentence or piece of text to the level of someone who is low-literate or the level of a child. That’s what Michiel Tebbes discovered when he was on duty in the emergency department at Slingeland Hospital and the cardiologist told him, in medical language, what exactly was wrong with his patient. Michiel: “Cardiologists use a lot of abbreviations. I knew exactly what he was saying, but this patient’s grandchild was also there. I wanted to explain things in a way that both grandad and grandchild would both understand. So I typed the cardiologist’s medical sentence into ChatGPT and asked it to rewrite it for a four-year-old child. The language model came up with a sweet little story that was also very clear. I never could have thought of it that way myself.” 

Why is rewriting necessary?

As we have said, one in six people in the Netherlands - a total of 2.5 million people - are low-literate. They have difficulty with reading, and also with understanding complex information. Of course, they are still in charge of their own bodies. When they are ill, sometimes they are forced to make complicated decisions. You want them to be properly informed.

Furthermore, with the rising demand for care and the declining number of working people, we also need to organise care differently. Morgens Conclusion commissioned research on future scenarios for hospitals. It clearly showed that if we don’t change the way we work, we are heading for a healthcare crisis. To address this, we can maximise our commitment to a digital transformation of healthcare. This requires us to look closely at how we organise tasks. Michiel: “Do you know that, as a doctor, I spend about 70 per cent of my time in front of a computer?” And of the 30 per cent of time he dedicates to consulting with colleagues and interacting with patients, he also needs to spend time on rewriting medical information. “Because we use much medical jargon when talking among ourselves. Afterwards, when I have a conversation with a patient, my brain is constantly thinking in the background: how do I explain this in an understandable way? This takes time and mental energy.”

You can save that energy with automatic rewriting. “As a doctor, you enter a lot of medical information into the EHR. This information is also given to patients via patient portals, but the patients don’t understand the medical jargon. How nice would it be if the patient just saw it rewritten in language matching their level of understanding?”

You can also use ChatGPT for translation. Michiel uses this function regularly for asylum seekers in the emergency ward at Slingeland Hospital, and he missed having it when he worked at the hospital in Deventer, a city with a large Turkish community. “Of course, it’s extremely important that patients get information in their own language. That obviously applies to people who don’t speak Dutch, but also to immigrants who do speak Dutch, because healthcare often involves complex information. People are also under stress, which means they don’t remember a lot of the conversation anyway.”

You can also use AI to create a video of the written text. This is essential for patients who are illiterate, but is also nice to share with your family or friends at home. This way, everyone is informed at the same time and you can have a better discussion about which treatment option to choose. 

Next step

The next step Michiel would like to take is to develop a plugin that makes it possible to launch a web browser to do the translation from within the EHR. “If doctors have to leave the EHR, you’ve made it too complex. That makes it more work instead of less. It should be set up in such a way that when a doctor enters medical information in the EHR, a rewritten message is generated for the patient automatically,” says Michiel. “The doctor won’t have to do that work anymore, and the information will be easier for the patient to understand.”