On November 13, the 5th Nandujiang Smart Healthcare and Rehabilitation Industry Summit Forum was held in Haikou. Li Luya, Founder and Chairman of Angelexo Scientific Co.,Ltd., delivered a keynote speech titled "From 'Reading the Brain' and 'Understanding the Brain' to 'Writing to the Brain' — Brain-Computer Interfaces, AI, and Robots Reconstructing the Underlying Logic of Neural Intelligence." He systematically elaborated on the practical bottlenecks of non-invasive brain-computer interfaces (BCIs) in clinical rehabilitation, technological breakthrough pathways, and the value of the "Read-Understand-Write" closed loop in reconstructing the underlying logic of neural rehabilitation.
Currently, invasive BCI technology is developing rapidly, with domestic manufacturers achieving up to 4,096 signal acquisition channels. However, research into brain science itself remains relatively weak. The core idea behind Elon Musk's initiation of invasive interfaces is to parse and digitize the brain. Yet, for the current realistic indications corresponding to invasive BCIs, as Li stated, it is "still in the initial trial phase, and the scientific value has not yet been fully realized."
Against the backdrop of BCIs becoming a state-encouraged development area, Angelexo, as an enterprise positioned at the front line of clinical practice and product application, focuses more on how to read brain signals, understand them, and form a closed loop by acquiring brain information to solve actual problems in clinical settings and daily life.
Li Luya listed the unsatisfactory status quo of BCI technology in real-world applications: rehabilitation treatments are mostly passive, and major conditions such as depression have not been well resolved to date. "The needs arising from these conditions are precisely the core content of today's research," he noted.
The "80% Ceiling" of Non-Invasive BCIs and Strategies for Breakthrough
In his speech, Li Luya directly addressed the greatest confusion regarding BCI devices in frontline clinical settings. For instance, common "dad hats" available on Taobao present issues with motion trajectories—rigid connection methods work adequately when patients are seated still, but once combined with movement during rehabilitation, signal accuracy is severely compromised, leading to a significant drop in decoding rates. He frankly stated that such BCIs face an "80% ceiling." However, Li also emphasized that BCIs are not distinguished by hierarchy or cost, but rather by different indications; for recoverable neurological rehabilitation patients, non-invasive approaches offer a lower-threshold, more convenient pathway.
Angelexo's response strategy follows two lines: First, maintaining stability along the non-invasive route by adopting a "use-as-acquire" approach, utilizing flexible structures to tolerate signal changes during movement. Second, Angelexo proposed a concept not yet fully classified—"minimally invasive non-invasive BCI." By penetrating through the stratum corneum into the dermis, this method allows for wearing at the same location for at least two weeks with guaranteed safety, while supporting bipolar recording.
Li Luya demonstrated two technical routes during the speech: flexible velcro electrodes and microneedle electrodes. The former is currently widely used clinically with good adaptability; the latter uses a microneedle gun to insert 0.7–2 mm microneedles into the epidermis and dermis, bypassing the high-impedance stratum corneum (10–20 microns) to acquire signals close to cortical level in a nearly non-invasive manner. If fully effective approximately 5 mm below the stratum corneum, efficiency is higher.
"Once the stratum corneum is breached, impedance drops below 5K, and dual detection can form a closed loop," Li Luya identified breaching the stratum corneum as the fundamental prerequisite for breaking the non-invasive ceiling. The clinical significance of this judgment lies in scalability: reading electrodes require no complex protection, sensors remain in place after wearing, and wear duration far exceeds traditional solutions, addressing the scale bottleneck of "whether it can be used."
The Core Challenge of "Understanding the Brain"
After reading the brain, the more critical step is "understanding the brain," i.e., comprehending whether the brain's consciousness intends to move an arm or a leg, and how it wants to move.
Li Luya pointed out a variable often overlooked in clinical settings: brain signals differ between day one and day thirty of stroke patient rehabilitation, requiring algorithms to adapt to this dynamic change. Angelexo's solution embeds artificial intelligence algorithms, retaining data collected during a single session for the same patient to achieve self-adaptation within subsequent rehabilitation cycles. Combined with the microneedle structure, the decoding ceiling can be raised to 90%; thanks to optimized "brain understanding" algorithms, adaptation time for different patients and time points can be completed at the minute level, with real-time control kept within 100 milliseconds.
In terms of efficacy verification, Li Luya prudently noted that assessing whether a BCI is good and how it performs cannot rely solely on scales; it requires brain imaging for clinical outcome recognition and validation of the correctness of clinical research pathways. Further advancing along the microneedle technology route, regulating nerves through excitation and inhibition can apply to various diseases: enhancing neuronal activity around the lesion site during the rehabilitation stage, and potentially suppressing hunger sensations for weight loss when depth is sufficient, with future advancements tailored to specific indications.
"Writing to the Brain" is the Watershed: Treatment Requires a Closed Loop, Not Fixed Milliampere Values
Li Luya positioned "writing to the brain" as the watershed for BCIs, representing the current "ceiling-level existence." While there are many methods for reading the brain—whether three mainstream routes or their variants—methods for writing to the brain are rare. Photoelectric and acoustic means can all be used for writing to the brain, but he tends to implement simpler methods, which naturally require clinical refinement and verification.
He used depression as an example to illustrate the value of the "Read-Understand-Write" closed loop: In the past, doctors prescribed medication or applied stimulation, such as 30 minutes daily with one or two milliamperes, lacking closed-loop feedback, making it difficult to judge what constituted successful treatment. Only with BCI technology forming a closed loop can previously unsolvable problems be precisely addressed. "Now 20 minutes is enough, but what are the results? This still requires clinical verification," he said. For instance, the usage density of double microneedle electrodes also needs consideration.
After solving reading and writing to the brain, the next step is integration, feeding back effects into treatment plans. Angelexo currently has products covering shoulders, wrists, hands, etc., across the "Brain-Machine-Body" three levels, and validates whether brain states have recovered through extensive clinical imaging.
Li Luya emphasized that Angelexo's "Brain-Machine-Body" closed-loop technology platform is not merely a superposition of products, but a system-level closed-loop rehabilitation platform.
He also revealed that Angelexo is building a "Brain-Controlled Intelligent Operating System" covering platform technologies for the acquisition layer, decoding layer, and execution layer, welcoming more enterprises to join.
Three Judgments
At the end of the speech, Li Luya offered three judgments regarding the future evolution of BCI technology:
In terms of technological judgment, BCIs are moving from single-signal acquisition to bidirectional neural interaction; the combination of AI and BCI technology is shifting from general models to rehabilitation-specific models.
In terms of clinical implementation judgment, neurological rehabilitation is currently the scenario with the clearest clinical value, strongest willingness to pay, and most defined health pathway for BCIs. Manufacturers need to tackle the most difficult scenarios before extending to consumer-grade applications, which will then follow naturally. This requires perfecting scenarios while combining medical insurance coverage and consumer willingness, truly transforming BCI from national policy into a tool serving the people.
In terms of industry judgment, Li Luya believes that the watershed in the next three to five years will not be determined by whose parameter counts look best, but by whose clinical evidence is most solid, registration barriers are highest, and channel coverage is deepest.
It is worth noting that Angelexo's "Read-Write" dual-modal platform has recently gained industry attention. At the 2026 Global Brain-Computer Interface × Medical Insurance Innovation Scenario Competition, its brain-controlled single-limb intelligent rehabilitation robot project stood out among hundreds of global teams to win the second prize. The judging criteria focused precisely on four dimensions: clinical validation, technical access, pricing, and achievement transformation.
Among Angelexo's seven current NMPA Class II medical device registrations, five have been selected for innovative medical device review, covering multi-joint intelligent rehabilitation training for shoulders, elbows, wrists, hands, hips, knees, and ankles. Cumulative intellectual property rights exceed 200 items.
Editor: ChipInsight - Langkejian