AI has a strong influence on the field of medicine. The application of AI in medicine has two main branches: virtual and physical. The virtual component is represented by Machine Learning, represented by mathematical algorithms that improve learning through experience. The second form of application of AI in medicine includes physical objects, medical devices and increasingly sophisticated robots taking part in the delivery of care.  

According to research by Statista, in 2021 artificial intelligence (AI) in the healthcare market was worth around 11 billion U.S. dollars worldwide. Furthermore, it was forecasted that the global healthcare AI market would be worth almost 188 billion U.S. dollars by 2030, increasing at a compound annual growth rate of 37 per cent from 2022 to 2030. 

This proves that AI will play a significant role in the future of medicine. In diagnostics, successful tests have already been performed by AI. For instance, AI can learn to accurately categorize images according to whether they show pathological changes or not. However, it's difficult to train AI models to examine the time-varying conditions of patients and to calculate treatment suggestions. But not anymore. The TU Wien, in cooperation with the Medical University of Vienna, has achieved this with their latest research. 

The research paper published in ScienceDaily stated that, with the help of extensive data from intensive care units of various hospitals, an AI model was developed that provides suggestions for treating people who require intensive care due to sepsis. Analysis shows that artificial intelligence already surpasses the quality of human decisions. 

Use of data 

Prof. Clemens Heitzinger from the Institute for Analysis and Scientific Computing at TU Wien (Vienna) said that, in an intensive care unit, a lot of different data is collected around the clock. The patients are constantly monitored medically. The researchers wanted to investigate whether these data could be used even better than before.  

Normally, medical staff make their decisions based on well-founded rules. Most of the time, they know very well which parameters they have to take into account to provide the best care. But the AI model can easily make many more parameters than humans.  

In this project, the researchers used reinforcement learning. It is not just about simple categorization but about a temporally changing progression, about the development that a certain patient is likely to go through. Mathematically, this is different. And the research conducted in this field is comparatively low. 

In this case, the computer becomes an agent that makes its own decisions. The system is “rewarded” if the patient is well and “punished” if the condition deteriorates or death occurs. The computer program is tasked with maximizing its virtual "reward" by taking action. In this way, extensive medical data can be used to automatically determine a strategy which achieves an exceptionally high probability of success. 

Outperforming humans 

According to the researchers, cure rates are now higher with an AI strategy than purely human decisions. In one of their studies, the cure rate for 90-day mortality increased from about 3% to about 88%. This does not mean medical decisions will rely on an ICU computer. Instead, it will act as an additional device at the bedside. This would be highly favorable in education as well. 

Furthermore, there are some legal concerns that the researchers are trying to find answers to -- accountability of actions being a major one. The researchers regard that the research project shows artificial intelligence can already be used successfully in clinical practice with today's technology -- but a discussion about the social framework and clear legal rules are still urgently needed.