Researchers from the University of Technical University of Eindhoven and the North-Western University of the United States presented a new method for teaching neuromorphic microcircuits, which can revolutionize the medical devices of the future.
The structural resemblance of neuromorphic computers to the human brain has always posed challenges in terms of learning using external programs, which has been a slow and energy-intensive process.
A team of scientists has now developed a novel neuromorphic biomensor that is capable of learning directly at the microcircuit without the need for external programs.
In a groundbreaking proof-of-concept, the researchers utilized the biomensor to diagnose cystic fibrosis based on sweat samples.
“We have created a ‘smart biostasis’ that can learn to recognize diseases, such as cystic fibrosis, without the need for a computer or software,” said Eveline Van Doremaele, one of the authors of the study.
Their study revolves around a neuromorphic biomensible computer that operates based on the principles of neuron communication in the human brain.
The researchers focused on addressing the challenge of teaching neuromorphic technologies. “The new chip can study in real time, processing patients of patients, which accelerates the learning process and contributes to its use in interactive bio-proceedings,” the study explains.
During the experimentation phase, the scientists employed the new chip to diagnose cystic fibrosis, a genetic disease. The primary diagnostic method utilized was the analysis of sweat samples, where a high level of chloride anions indicates the presence of cystic fibrosis.
“We used sweat samples from healthy donors, with one sample testing negative and the other exhibiting a very high concentration of chloride anions,” Van Doremaele elaborated.
The study’s primary achievement is