Artificial intelligence superior to humans i

In the future, artificial intelligence will play an important role in medicine. Diagnostics has already run a successful test. For example, a computer can learn to classify images very accurately based on whether they show pathological changes. However, training artificial intelligence to examine patient conditions that change over time and calculate treatment recommendations is more difficult. This is exactly what was achieved at the Vienna University of Technology in cooperation with the University of Technology Vienna.

With the help of extensive data from intensive care units in various hospitals, an artificial intelligence has been developed that provides recommendations on treatment for people in intensive care due to sepsis. Analysis shows that artificial intelligence has already surpassed the quality of human decision-making. However, it is now important to discuss the legal aspects of such methods as well.

Optimal use of existing data

“In intensive care units, a variety of data is collected around the clock. Patients are constantly monitored medically. It’s been done,” said Professor Clemens Heitzinger of the National Institute of Science. Analytics and Scientific Computing at TU Wien. He is also co-director of the cross-disciplinary Center for Artificial Intelligence and Machine Learning (CAIML) at the Vienna University of Technology.

Healthcare professionals make decisions based on well-founded rules. In most cases, they are well aware of what parameters need to be considered in order to provide the best care. However, computers can easily consider far more parameters than humans, so in some cases this can lead to even better decisions.

Computers as planning agents

“In our project, we used a type of machine learning called reinforcement learning,” says Clemens Heitzinger. “This is not just a simple classification, say dividing a large set of images into those with and without tumor, but also the time-varying progression, the development that a particular patient will experience. Mathematically, this is something completely different, and very little research has been done on this in the medical field.”

Computers become agents that make their own decisions. If the patient is healthy, the computer receives a “reward”. If the condition worsens or dies, the computer will be “punished”. Computer programs are tasked with maximizing a hypothetical “reward” by performing an action. In this way, extensive medical data can be used to automatically determine strategies that achieve a particularly high probability of success.

already better than humans

“Sepsis is one of the most common causes of death in intensive care medicine and poses a major challenge for doctors and hospitals, as early detection and treatment are critical to patient survival,” said Professor Oliver Kinberger of the Medical University of Vienna. It has become. “To date, there have been few medical breakthroughs in this area, making the search for new treatments and approaches more urgent. It is of particular interest to investigate whether we can.” The use of machine learning models and other AI technologies presents an opportunity to improve the diagnosis and treatment of sepsis, ultimately increasing the chances of patient survival. “

Analyzes show that AI has already outperformed humans. “Currently, there are higher cure rates using AI strategies than with purely human judgment. , increased by about 3% to 88%,” says Clemens Heitzinger.

Of course, this doesn’t mean that medical decisions in the intensive care unit should be left solely to computers. However, artificial intelligence could work as an additional bedside device, allowing medical staff to refer to it and compare their assessments with the artificial intelligence’s suggestions. Such artificial intelligence could also be of great use in education.

Need to discuss legal issues

“But this raises important questions, especially legal ones,” says Clemens Heitzinger. “Perhaps the first question that comes to mind is who is responsible for mistakes made by artificial intelligence. But there is also the opposite problem. Did the patient choose a different treatment and as a result the patient suffered?” In that case, the doctor should trust the artificial intelligence because it has a lot of experience. Or should it be the human right to always ignore computer advice?

“Research projects show that artificial intelligence can already be successfully used in clinical settings with today’s technology. However, debates about social frameworks and clear legal rules are still urgently needed,” said Clemens. Heitzinger is convinced.