There is wide discussion and applications going on regarding the application of AI in the healthcare sector, like any other industry. As AI has already been performing accurate tasks in diagnostics, it is expected to play a pivotal role in medicines in the future as well. But there are certain challenges when it comes to training the AI to examine the time-varying conditions of a patient and compute the treatment suggestions. However, according to TU Wein, they have already accomplished this by collaborating with the Medical University of Vienna.

Using the extensive data, they attained from various intensive care units of many hospitals, they developed an AI that can offer treatment suggestions for patients affected by Sepsis and who need intensive care to support their lives. Let us dive deep into the abilities of AI that overshadow human decisions and their legal aspects.

How the technology makes use of the extensive data?

According to Clemens Heitzinger, Professor and Co-director of the cross-faculty Centre for Artificial Intelligence and Machine Learning (CAIML) of the Institute for Analysis and Scientific Computing at TU Wien, they wanted to make optimal use of the extensive data in the ICU. “In an intensive care unit, a lot of different data is collected around the clock. The patients are constantly monitored medically. We wanted to investigate whether these data could be used even better than before,” he said.  

The medical practitioners make observations and arrive are certain conclusions based on the well-found rules. Mostly they will be well-versed in the parameters they must consider to offer the best care in ICUs. Here, using a computer can create wonders as it can take more parameters than a single human can do all at once. Hence there is a great potential that this facility can make better decisions.

How the computer becomes a planning agent?

Prof. Heitzinger further opines the use of reinforcement learning in their project is quite different, and there has been little research on this in the medical field. “In our project, we used a form of machine learning called reinforcement learning. This is not just about simple categorisation- for example, separating a large number of images into those that show a tumour and those that do not-but about a temporally changing progression, about the development that a certain patient is likely to go through. Mathematically, this is something quite different. There has been little research in this regard in the medical field.

Here, the computer acts as an agent with its independent decisions. The computer is “rewarded” when the patient is well and will be “punished” if their condition worsens. Additionally, the computer is programmed to increase its virtual “reward” by taking actions at the right time. Hence extensive medical data can be utilised to automatically determine a strategy that tends to have a high probability of success.

AI v/s Humans

Professor Oliver Kimberger from the Medical University of Vienna talks about the importance of early detection of deadly diseases with low survival rates. He also believes that understanding the potential of AI in such scenarios will be a game changer. Prof. Kimberger says, “Sepsis is one of the most common causes of death in intensive care medicine and poses an enormous challenge for doctors and hospitals, as early detection and treatment is crucial for patient survival”. “So far, there have been very few medical breakthroughs in this field, which makes the search for new treatments and approaches all the more urgent. For this reason, it is particularly interesting to investigate the extent to which Artificial Intelligence can contribute to improving medical care here. Using machine learning models and other AI technologies is an opportunity to improve the diagnosis and treatment of Sepsis, ultimately increasing the chances of patient survival”.

Prof. Heitzinger mentioned that AI capabilities are already outperforming humans, as the cure rates are considerably high using the AI strategy rather than human decisions. For example, as per their studies, the cure rate for 90-day mortality surged from 3% to about 88%.

Though AI gives high possibilities with its accuracy, we cannot completely rely on a computer. Instead, AI may be able to run along as an additional device at the bedside. At the same time, medical professionals can consult it and compare their assessment with the AI suggestions and observations.

The unavoidable legal issues

Prof. Heitzinger, while discussing the legal issues, said, “One probably thinks of the questions which will be held liable for any mistakes made by the AI first. But there is also a converse problem- what if the AI had made the right decision, but the humans chose a different treatment option, and the patient suffered harm as a result?” This poses questions like if the doctor will be accused of not trusting the AI due to its extensive data and experience.

“The research project suggests AI can be already used successfully in clinical practices with today’s technology; a discussion on the social framework and clear legal rules are inevitable”, Prof. Heitzinger added.

Photo by Alexander Grey on Unsplash