Photo by cottonbro studio on Pexels
Cochlear has launched the Nucleus Nexa System, a revolutionary cochlear implant integrating edge AI directly within the human body. This groundbreaking device marks a significant advancement in medical technology, executing machine learning algorithms under stringent power limitations while securely storing personalized data on-device and enabling over-the-air firmware updates for continuous AI model enhancement.
The Nucleus Nexa System incorporates SCAN 2, an environmental classifier employing a decision-tree model to analyze incoming audio and categorize it as Speech, Speech in Noise, Noise, Music, or Quiet. The implant also utilizes Dynamic Power Management to optimize power efficiency based on these environmental classifications. Over-the-air firmware updates allow for ongoing improvements in signal processing and noise reduction. The implant can store up to four individualized hearing profiles within its internal memory.
Furthermore, the system features ForwardFocus, a spatial noise algorithm that leverages input from two omnidirectional microphones to generate target and noise spatial patterns. The crucial AI aspect is its automation: ForwardFocus operates autonomously, easing the cognitive burden on users in challenging auditory environments. The decision to activate spatial filtering is made algorithmically based on environmental analysis, requiring no user intervention.
Cochlear is also researching AI applications beyond signal processing. According to Jan Janssen, Cochlear’s Global CTO, they’re investigating AI and connectivity to automate routine check-ups and lower long-term care expenses. The device supports Bluetooth LE Audio and Auracast broadcast audio, offering superior audio quality with reduced power consumption. More importantly, this integration establishes the implant as a node within larger assistive listening networks. Auracast broadcast audio enables direct connection to audio streams in public locations, such as airports and gyms.
This medical breakthrough provides a roadmap for edge AI medical devices facing similar challenges: prioritize interpretable models like decision trees, rigorously optimize for power efficiency, incorporate upgradability from the outset, and design for long-term usability. Cochlear’s implementation demonstrates the transformative potential of AI in medical devices, suggesting its integration may become the standard of care.