Administrative burdens and staff shortages in healthcare continue to rise. AI-driven healthcare innovations offer relief, provided healthcare organizations get the basics right. "The question is: what does my staff need in order to work well with it?" said Bart Scheerder of the Applied AI Accelerator Lab, during the master class "Entrepreneurship with data and AI in healthcare.
Healthcare professionals spend much of their time on administrative tasks. Precious time that could also be spent on patients. From survey of the Dutch Internists Association It turns out that 40 percent of time is spent on administration and only 20 percent on patient contact.
Although the study was conducted among internists, this imbalance is also recognizable for other healthcare professionals. Recent Central Bureau of Statistics (CBS) survey.shows that healthcare professionals spend an average of 31 percent of their working time on administration, such as reporting and registration.
AI as a gamechanger
The Applied AI Acceleration Lab (A3 Lab) at UMC Groningen (UMCG) hopes to change this. Bart Scheerder, co-founder of the innovation lab, endorses the importance of spending useful and enjoyable time. "Many caregivers are leaving healthcare. Research shows that 43 percent leave the organization within two years, and 20 percent of those leave care altogether. It's important to spend your work time correctly. It has to be meaningful."
In it A3 Lab researchers, technicians and medical specialists work together on AI applications that make the work of healthcare professionals lighter and more enjoyable. "We're testing what can be done, implementing some work, and exploring where it's going," he said.
Also read: Saskia Peerdeman: 'AI has become our co-pilot'
From 7 minutes to 15 seconds of preparation time
In November 2023, the UMCG implemented their first AI application in the hospital. The hospital was receiving about 8,000 messages a month through the online portal from patients. People were thinking about the question: how can we as an organization make an impact with AI?
The answer became an AI application linked to the electronic patient record (EHR). The tool reads the patient's question, forms a draft answer including references and puts it ready for the healthcare provider. If necessary, it is modified before the answer goes to the patient.
The result: average outpatient preparation time dropped from 7 minutes to 15 seconds. "Importantly, the AI application does not give medical advice," says Scheerder. "It helps healthcare providers answer patient questions." The UMCG was the first hospital in Europe to use a chatbot in written contact with patients.
Also read: How to ensure sustainable adoption of your healthcare innovation: lessons from practice
Challenges of generative AI
The sustainable implementation of an AI application is not without controversy. Scheerder: "There are wonderful innovations, but the question is: what does my staff need to work well with them? How does AI change the work of the healthcare professional?"
Scheerder notes the following challenges when it comes to implementing generative AI tools:
-
t
- Sustainability in CO2 emissions and money: what effect does the application have on the environment and the wallet?
- Laws and regulations: does the innovation meet existing laws and regulations?
- Privacy and security: is patient data security at stake?
- Liability: who is responsible in case of errors?
- Change management of people and organization: How do you make sure you bring your people along with the changes?
- Bias and discrimination: To what extent does the application use one-way data?
- Transparency and explanation: what data are used and where does the output come from?
- Source hallucinations: how likely is the application to make up data?
t
t
t
t
t
t
t
Also read: Why many healthcare innovations fail: the power of hidden interests
Take the healthcare professional with you
According to Scheerder, innovation is driven by internal colleagues. For example, the UMCG uses a so-called innovation pipeline. "The first innovation in 2023 was ad hoc, but we are now in a phase where it is more programmatic. From idea to impact."
The hospital uses a kind of digital idea box where employees can go with a proposal. "The AI core team then looks at whether it is feasible. If so, a pilot is started. Particular attention is then paid to the value it brings."
One of the applications that employees within the UMCG can now use is ChatUMCG. "Officially, employees don't use ChatGPT, but of course it happens. There are currently 1 billion conversations conducted via ChatGPT. We wanted to be independent of the existing models and therefore now offer ChatUMCG."
Who saves what, has what
If the AI core team doesn't see value in an idea, it's not necessarily wasted effort. "Is it not providing value right now? Then it goes in the recycle bin, not the trash. After all, maybe in a few years it will be relevant."
