If machine learning algorithms can assist providers with diagnoses by analyzing massive amounts of data that have been collected from patients, why not develop a way of implanting AI devices inside patients to gather the data and make the assessments in real time?

That futuristic sounding option has taken a step closer to becoming reality with the announcement that TU Dresden scientists at the Chair of Optoelectronics have now succeeded for the first time in developing a bio-compatible implantable AI platform that classifies in real time healthy and pathological patterns in biological signals such as heartbeats. Indeed, the device detects pathological changes even without medical supervision, and the research results have now been published in the journal Science Advances.

For the study, a research team led by Prof. Karl Leo, Dr. Hans Kleemann and Matteo Cucchi developed an approach for real-time classification of healthy and diseased bio-signals based on a biocompatible AI chip.

AI is developing quickly, they noted in their report, but “real-world clinical implementation is so far limited to an ‘offline’ analysis of the patients’ data using software-implemented neural networks. A highly attractive vision is the active monitoring and detection of malign patterns in vivo through computational platforms attachable to or even implantable into the body.”

According to the announcement, their approach involved using polymer-based fiber networks that structurally resemble the human brain and enable the neuromorphic AI principle of “reservoir computing.” The random arrangement of polymer fibers forms a so-called "recurrent network," which allows it to process data, analogous to the human brain. “The nonlinearity of these networks enables to amplify even the smallest signal changes, which -- in the case of the heartbeat, for example -- are often difficult for doctors to evaluate. However, the nonlinear transformation using the polymer network makes this possible without any problems.”

In trials, the AI was able to differentiate between healthy heartbeats from three common arrhythmias with an 88% accuracy rate. In the process, the polymer network consumed less energy than a pacemaker. The potential applications for implantable AI systems are manifold: For example, they could be used to monitor cardiac arrhythmias or complications after surgery and report them to both doctors and patients via smartphone, allowing for swift medical assistance.

“The vision of combining modern electronics with biology has come a long way in recent years with the development of so-called organic mixed conductors,” explained Matteo Cucchi, a PhD student and co-author of the report. “So far, however, successes have been limited to simple electronic components such as individual synapses or sensors. Solving complex tasks has not been possible so far. In our research, we have now taken a crucial step toward realizing this vision. By harnessing the power of neuromorphic computing, such as reservoir computing used here, we have succeeded in not only solving complex classification tasks in real time but we will also potentially be able to do this within the human body.”

Photo by laremenko/Getty Images