Paradromics Inc. has successfully completed its first human brain-computer interface (BCI) implant, marking a significant milestone in the development of high-bandwidth BCIs designed to restore communication for individuals affected by conditions like amyotrophic lateral sclerosis (ALS) and locked-in syndrome. The procedure was led by Dr. Matthew Willsey, with support from Dr. Oren Sagher, and a multidisciplinary team of clinicians and engineers from University of Michigan.
Founded in 2015 by Dr. Matt Angle, Paradromics received early funding from the National Institutes of Health (NIH) and the Department of Defense, Defense Advanced Research Projects Agency (DARPA), which accelerated its research into high data-rate BCI systems.
The successful human implantation was made possible by Connexus®, the company’s flagship BCI system designed for clinical use in neural communication restoration.
Connexus: High-bandwidth BCI built for communication
The implant features more than 1,600 independent recording channels. These channels are linked to high-density microwire arrays embedded in the brain’s cortex, allowing the system to capture and decode high-resolution neural activity in real time.
Specifically designed to restore speech and communication, the system translates neural signals from the speech motor cortex into text or synthesized speech. It communicates wirelessly with external devices, enabling a seamless and intuitive user experience, particularly for individuals who have lost the ability to speak due to neurodegenerative diseases or injury.
In this article, we look at Paradromics’ patent portfolio and how their IP strategy supports the development of Connexus.
Paradromics: Patenting Activity
As a startup company, Paradromics has maintained a focused patent strategy since its founding. Patent filings peaked in 2016, coinciding with early funding milestones such as the support from NIH. In 2017, Paradromics was among the companies that secured a DARPA contract under the Neural Engineering System Design (NESD) program.
Since then, patenting activity has slowed, which may be attributed to strategic decisions such as the relocation of headquarters to Texas in 2019. The company also concentrated its efforts on the development of their BCI in 2020, shifting priorities from intellectual property filings to product advancement.
Paradromics: Top Law Firms
Our analysis indicates that Wilson Sonsini Goodrich & Rosati is Paradromics’ leading legal representative, having handled 14 patent filings. This includes the NIH-backed patent of Paradromics and the national filing in the United States and international filing of their first patent. Prominent lawyers representing Paradromics from Wilson Sonsini Goodrich & Rosati are Hin Meng Au and Madeline Hess.
Additional firms and representatives, such as J A Kemp LLP, East IP and Andrew Park, highlight Paradromics’ selective yet diverse legal network. This approach reflects a strategic effort to safeguard innovation within a rapidly evolving global neurotechnology landscape.
Paradromics: Top Technology Areas
Majority of Paradromics’ patent filings fall under the A61B classification which describes diagnosis, surgery and identification. This reflects the company’s clear technical emphasis on implantable brain-computer interfaces for clinical use. The second most active classification, A61N (Electrotherapy and Magnetotherapy), indicates ongoing work in neural stimulation and signal processing.
Additional filings in areas such as metallic coatings, semiconductor devices, and electroforming suggest a multidisciplinary approach that integrates materials science and advanced electronics, supporting the development of implantable neurotechnology systems for emerging clinical and engineering demands.
Patents behind Connexus BCI
Paradromics is part of a group of six companies that have emerged as key players in the development of brain-computer interfaces, alongside Neuralink, Synchron, Motif Neurotech, Blackrock Neurotech, and Precision Neuroscience. In 2024, each of these companies reached important milestones in clinical or technological progress, with Paradromics distinguishing itself through the successful completion of its first human BCI implant.
(Additional readings related to BCI: Patent Snapshot: Neuralink, 2D Materials for Neuromorphic Computing)
The following sections highlight the foundational patents and patent applications that made the Connexus BCI possible.
Microwire-based brain electrode system for long-term implants
U.S. Pat. App. No. 2021/0098341, titled “Microelectrode array and methods of fabricating same,” describes a high-density implantable electrode system for use in brain-computer interfaces (BCIs). The device features a ceramic base with metal connections that hold an array of microwires, which are inserted into the brain using biocompatible solder or brazing materials. The tips can be sharpened for precise contact with brain tissue while minimizing damage.
The patent outlines several fabrication methods, including 3D printing and milling, to shape the microwires. Ceramic coatings improve insulation and durability for long-term implantation. The design also emphasizes hermetic sealing to prevent fluid intrusion and supports scalable, high-channel-count recording or stimulation, making it well-suited for clinical BCI applications.
The patent application was filed on September 24, 2020, and was published on April 1, 2021. The application lists Yifan Kong, Kevin Boergens, Matthew Angle, and Aleksandar Tadic as inventors. Richard Torczon, Peter Eng, William Barrett, et al. from Wilson Sonsini Goodrich & Rosati represented Paradromics for this patent application.
Intelligent neural interface for assistive communication devices
U.S. Pat. App. No. 2025/0090067, titled “Low-area, low-power neural recording circuit, and method of training the same,” addresses a key challenge in brain-computer interface (BCI) systems: how to efficiently and accurately decode neural signals collected from implantable electrode arrays.
Traditional methods often struggle with high power consumption, limited channel scalability, and reduced real-time responsiveness, especially in applications like speech restoration or assistive communication, where low latency and adaptability are critical.
The patent addresses the issue by proposing an integrated system that begins by acquiring signals from multi-channel neural interfaces, then processes them through filtering, binning, and normalization.
It extracts key features such as spike rates and voltage patterns, and feeds them into a neural decoder powered by advanced machine learning models, including recurrent neural networks and transformers.
The decoder can translate brain activity into letters, words, or device control signals, and may integrate a language model for improved contextual accuracy. The system also supports real-time feedback loops and adaptive learning, enabling personalized and responsive BCI experiences for users with paralysis or locked-in syndrome.
This patent application was filed on April 4, 2024, and was published on March 20, 2025. Richard Torczon, Peter Eng, William Barrett, et al. from Wilson Sonsini Goodrich & Rosati represented Paradromics for this patent application. It lists Matthew R. Angle, Robert Edgington, Aamir Ahmed Khan, Bart Dierickx, Peng Gao, Amir Babaie-Fishani, Ahmed Abdelmoneem, Bert Luyssaert, and Jean Pierre Vermeiren as the inventors.
Compact signal decoder for high-bandwidth neural interfaces
U.S. Patent No. 10,653,330 titled “System and methods for processing neural signals” outlines a modular system for processing neural signals captured from the brain via implantable electrodes. The system is designed to operate in real-time and emphasizes power efficiency and data compression, making it suitable for long-term implantation in brain-computer interface (BCI) applications.
The system consists of three key modules:
- Feature Extraction Module – Detects neural events (e.g., spikes) from analog signals using techniques like thresholding or pattern matching.
- Feature-Event Coalescence Module – Consolidates signals using statistical inference or prior models, reducing noise and avoiding traditional “spike sorting.”
- Approximator Module – Compresses the coalesced neural events into a high-entropy code using machine learning models such as autoencoders or probabilistic algorithms.
These parts work together to send or process brain data efficiently, using less power and bandwidth. The system can run on specialized hardware like application-specific integrated circuits (ASICs) or field-programmable gate arrays (FPGAs), making it suitable for use in implantable devices that help with movement, communication, or therapy especially for people with serious neurological conditions.
This patent was filed on February 13, 2019 and was granted May 19, 2020, with Matthew Angle, Edmund Huber, and Robert Edgington as the inventors. Paradromics was represented by Richard Torczon, Peter Eng, William Barrett, et al. from Wilson Sonsini Goodrich & Rosati.
