Page 124 - AI Standards for Global Impact: From Governance to Action

AI Standards for Global Impact: From Governance to Action

20�2 Opportunities and challenges of BCI

The following opportunities and challenges of BCI were identified in this session based on the
presentations made:

a) University of Rome showcased BCI’s application in monitoring high-risk professionals
like pilots and surgeons. By analyzing EEG, galvanic skin response, and heart rate via
machine learning, the system real-time quantified cognitive load, attention, and stress to
trigger safety alerts. Scalable lightweight devices are required—using few electrodes while
retaining 80-90% core data—for easy real-world deployment.
b) Neuroscience-Paris Seine-lBPS Laboratory ICNRS INSERM at Sorbonne University warned
of ethical risks in non-medical BCI expansion, particularly in employee monitoring
scenarios that could infringe on cognitive freedom. It was highlighted that there is a need
for robust governance frameworks, citing the OECD’s 2019 recommendations (classifying
neural data as uniquely sensitive) and France’s 2022 Responsible Development of
Neurotechnology Charter – a multistakeholder agreement banning non-voluntary neural
data use and protecting mental privacy, now integrated into the European Brain Initiative.
c) A BCI implant recipient shared firsthand experience of controlling prosthetics with four
implanted electrodes, highlighting needs for longer device lifespan (current maximum is
10 years) and easier maintenance.
d) Institute of Information and Communication Technologies (IICT) at the Bulgarian Academy
of Sciences focused on technical challenges of invasive BCI, such as signal instability
caused by brain deformation. The need for standardized failure analysis frameworks was
emphasized and the presentation also highlighted IEEE’s efforts to unify terminology, data
formats, and metadata compatibility to break resource-sharing barriers.
e) China's Beijing Institute of Technology analysed BCI’s market growth and risks, including
surgical infections from invasive devices which have a 5-10 year implant lifespan, and
ethical issues like neural data privacy leaks. The presentation highlighted global
governance progress, such as China’s BCI Research Ethics Guidelines and UNESCO’s
2025 draft Recommendation on the Ethics of Neurotechnology.

20�3 Industrialization and commercialization barriers of BCI

The following industrialization and commercialization barriers of BCI were identified in this
session based on the presentations made:

a) Araya showcased 79.8% accuracy in decoding 512 common phrases from EEG data,
enabled by a "health data pre-training + patient fine-tuning" framework that slashed
clinical adaptation time – boosting word recognition accuracy from 13.2% (using only
patient data) to 54.5% with minimal patient-specific data. It was highlighted that a rental-
based business model (equipment leasing + pay-per-service) to lower initial hospital
investment barriers, is now piloted in three Japanese rehabilitation facilities to accumulate
real-world data.
b) SceneRay presented dual-target deep brain stimulation (Combo-Stim DBS) therapy for
opioid addiction, which precisely targets the nucleus accumbens and anterior cingulate
bundle to repair reward circuit deficits. Clinical trials (2018–2024) and 10-year follow-ups
showed an 80% 6-month abstinence rate and 69% long-term success, with no adverse
impacts on cognition or daily function. Adhering to ISO 13485 (quality management) and
ISO 14971 (risk management) standards, the technology supports safe device removal
after two years, paving the way for scalable medical applications in addiction treatment.
c) iFLYTEK European Division shared their experience in scaling neurotechnology
applications through "R&D-scenario validation-large-scale promotion" models. For
example, optimizing algorithms for specific scenarios and reducing deployment costs,
which offers insights for BCI’s "technology inclusivity" challenges.

112