AI in medicine: not a threat, but an enabler. How GE HealthCare is rethinking the future of healthcare
Source: EVG Kowalievska, Pexels
"We need to use artificial intelligence to free physicians from routine tasks and allow them to focus on what really matters - patients," says Jan Beger, global head of AI advancement at GE HealthCare. He is one of those changing the perception of technology in medicine from a threat to a powerful tool for transformation.
AI has been used in medicine, especially in radiology, for more than 15 years. It is mostly about "narrow" computer vision algorithms tailored to a specific disease. However, now a new era is beginning - the transition to foundation models, large neural networks capable of learning from huge arrays of heterogeneous medical data.
"Radiologists don't work in a vacuum. They evaluate images in context: taking into account tests, medical history, symptoms. The new generation of AI models is learning to do the same thing - synthesise information from different sources. This brings algorithms closer to clinical thinking," explains Beger.
Medicine handles far more diverse data than, say, finance or aviation: from digital images to handwritten doctor's notes. In this heterogeneity is a challenge, but also an opportunity. Modern AI models can interpret such "multimodal" data - and that's a real game changer.
Beger also emphasises the "human" potential of AI. Complex medical findings are often incomprehensible to the patient. But AI can not only translate them into accessible language, it can do so with a tone that is close to human empathy. This is a way to increase the patient's involvement in their own care and build trust in the system.
GE HealthCare sees AI as a strategic asset. It can improve not only medical technology, but also the company's internal processes, from logistics to customer service. "Any company should start down this path. If you haven't started yet, start tomorrow," says Beger.
Integrating AI into healthcare requires resources, regulatory expertise and a careful approach. "Using solutions such as ChatGPT without proper security can lead to sensitive information being leaked," warns Beger. He also emphasises that AI is evolving at such a rapid pace that yesterday's solutions can be obsolete in six months. That's why it's better for small companies to enter into partnerships than to try to build everything from scratch.
Three rules of success for AI in medicine
- Solve a real problem. Many startups create powerful technologies without understanding the market. You need to start with a genuine clinical need - and work with doctors and patients.
- Build into existing systems. Physicians are overloaded with interfaces. A new application should be invisible - embedded, no unnecessary clicks.
- Work on trust. People fear AI: doctors fear it for their jobs, patients fear it for the loss of human relationships. Need to explain, educate and demonstrate value.
For Beger, the most important thing is not economic impact, but restoring the human dimension of medicine. "Let AI do the routine. And doctors will once again be able to do the most important thing - to take care of people." Perhaps this is how - through digital progress - we will bring warmth back into the cold corridors of modern medicine.
