Dr. Data in the emergency Department: Dietmar Frey this is not Science Fiction. The neuro-surgeon at the Charité hospital in Berlin and his Team are working on computers with artificial intelligence, to the benefit of the treatment of patients with acute stroke. “This is more than an idea. We have the technology, a prototype and first Machine-Learning models,” says Frey. If 2019 begins the year of science with focus on artificial intelligence, it will go in the medicine to such questions as: What computers can and where Physicians remain irreplaceable?
search for the correct therapy, the
Around 270,000 Federal citizen encounters each year, “the blow”. Then the time runs. The brain is not sufficiently supplied with blood and oxygen, dies tissue. The consequences can be dramatic, Voice failure and paralysis threaten.
Medical guidelines in Germany state that the affected brain tissue after four and a half hours of the dead and side effects of a therapy even more damage bleeding can cause in the head, for example. Therefore, not going to therapy after four and a half hours, says Frey. “That may be statistically correct, for the individual patient, however, this is not always the right therapy.”
He suggests that therapies in certain cases, even after four and a half hours and still make sense and in other, maybe after two hours, not more. But how do you know who what is?
pattern in the case of strokes to recognize
For Frey, the answer is clear: A machine could search in a matter of minutes, thousands of seizures a comparison of the records of impact. Matching and pattern to show, the body could help a physician in the Rescue in the decision. A Job for Dr. Data. To search for AI in databases and computers can be trained, according to programmed Patterns.
Artificial intelligence could help Doctors in many different fields, says Felix Nensa, a radiologist and a computer scientist at the University hospital of Essen. But especially there, where the man is a failed occupation: the case of boring and tiring activities, such as the tumor measurement or when it is stored and scanning of thousands of image files and text files.
Meaningful programming is an art, even in medicine. “You need a sufficiently large training data Sets and must be able to represent variability,” says the bioinformatics scientist Benedikt Brors German cancer research center (DKFZ) in Heidelberg. A data set is too small, could not detect the calculator, for example, patterns with a very rare tumor disease.
help in tumor diagnosis
The DKFZ developed since ten years, an AI System to see if Neuroblastoma – very rare tumors in children are aggressive or rather harmless. The therapy depends. However, it will take with the clinical testing. The requirements for certification are extremely high, says Brors.
Dietmar Freys computer at the Charité, have already been charged over the seizures 1400 shock. Patient stories, anonymised and broken down to formulas. It’s information on age, gender, weight, Smoking habits, pre-existing conditions, taken medications, and laboratory values.
Frey’s Plan is to have this wealth of Data retrieval in the emergency room when a new Patient with acute stroke is admitted. So to directly examine whether there were similar manifestations of a stroke, as the method was, and whether that helped.
The results will spit out the Computer in a matter of minutes – so that the Doctors can find the best therapy for the patient, which is located just in front of you. “A doctor has to today, wallowing neither the time nor the capacity, in the emergency room, mountains of Files for comparisons. And in your head and calculate what he can’t have it all.”
In Essen, medical professionals builds Nensa image databases: CT and x – ray images of lung diseases office of the clinical data. “We have new cases,” he says. “Then we look at the new pictures and think: phew, what’s that now?” With a KI-database one could easily check whether there was a similar case, and would have direct access to its diagnosis and treatment.
“Such a KI to me is the first wave of Tools, which relieved the doctor,” he summed up. “In five to ten years there will be more of these Tools. The medical profession will change more in the direction of merge and Interpret data,” says Nensa.
Still, there are many limitations in the AI. A machine could so far reproduce, for example, a rule-based Knowledge from the guidelines. “What she has not yet, is creativity, or Intuition,” says Brors. People, however, could also think around the corner.
Ulrike von Leszczynski / DPA