BRAINAURORA-B (06681) has announced significant progress in its vertical large language model technology and clinical research.
Securing Second Place in the Third Xiong'an Vertical Large Model Competition
The final round of the Third Xiong'an Future City Scenarios Vertical Large Model (Agent) Application Competition was held in early July at the Xiong'an Science and Innovation Park. Organized by the Xiong'an Future City Scenarios Committee, the competition attracted over 700 AI projects from around the world, with more than 90 high-quality projects advancing to the finals and 52 top national teams competing. The group's project, "Exploration and Application of Personalized Intervention Technology Based on Cognitive Deep Reasoning Large Models in the Medical Field," won second place in the "AI+" new scenarios track, powered by its self-developed BrainAuGPT cognitive deep reasoning model. The team built an industry-unique "cloud-edge-end" intelligent assessment system and a "thousands of people, thousands of prescriptions" personalized AI intervention system based on BrainAuGPT. This system leverages three original technologies: mining graph-structured behavioral data, a second-order knowledge representation LLM fine-tuning technique, a spatio-temporal uncertainty Monte Carlo tree search decoding technology, and a cross-layer entropy-enhanced model de-hallucination technique, forming a "perception-reasoning-decision-feedback-optimization" closed loop.
Additionally, a project from the group's wholly-owned subsidiary, Beijing Zhijingling Technology Co., Ltd., titled "Exploration and Application of Intelligent Agent Based on Temporal Knowledge Graph and Cognitive Large Model in the Medical Field," was successfully selected this month as a "Typical Case of AI-Enabled Scenario Application (2026)" at the 2026 Global Digital Economy Conference.
Latest Clinical Research Findings Published in Alzheimer's Research & Therapy
Recently, clinical research deeply participated in by the group, led by Professor Zeng Yong's team from Beijing Anzhen Hospital, Capital Medical University, was published in the internationally authoritative journal Alzheimer's Research & Therapy (Impact Factor 8.9, JCR Q1). The study, "Neural changes after computerized cognitive training in coronary heart disease with mild cognitive impairment: secondary analysis of a randomized clinical trial," is a pre-specified secondary analysis of a multicenter, double-blind, active-controlled randomized clinical trial (NCT05735041). It included 185 patients with coronary heart disease and mild cognitive impairment. Using multimodal magnetic resonance imaging, it systematically explored the remodeling effects of multidomain difficulty-adaptive computerized cognitive training on brain network structure and functional plasticity, investigating the mechanism of the "brain-heart axis" in non-pharmacological cognitive rehabilitation.
After a 12-week intervention using the group's core product for multidomain difficulty-adaptive computerized cognitive training, patients showed a significant increase in functional connectivity within the parietal memory network (left PMN, p = 0.028) and the contextual association network (left CAN, p = 0.045; right CAN, p = 0.031). Patients also exhibited a significant increase in grey matter volume in the right precuneus (p = 0.048) and right parahippocampal cortex (p = 0.032), along with significantly enhanced structural connectivity in white matter pathways between the right "precuneus-inferior parietal lobe" (p = 0.016) and "parahippocampal gyrus-superior frontal gyrus" (p = 0.044).
Furthermore, the study revealed a close link between changes in brain network connectivity and improvements in patient cognition and reductions in blood pressure/pulse pressure. It found that increases in functional connectivity in pathways such as the right PMN and the cingulo-opercular network were significantly negatively correlated with changes in systolic and pulse pressure. Mediation analysis further indicated that changes in functional connectivity between the right PMN and CON significantly mediated the therapeutic effect of cognitive intervention on reducing patient pulse pressure (indirect effect = 0.62, p = 0.040).
The study concluded that multidomain adaptive computerized cognitive training can promote functional and structural neuroplasticity in memory-related circuits in patients with coronary heart disease and mild cognitive impairment. Changes in functional connectivity are associated with cognitive improvement and may mediate blood pressure reduction, suggesting that such training may be involved in brain-heart axis interactions.
Significance of These Achievements for the Group
These awards and research outcomes collectively demonstrate the group's comprehensive competitiveness. The second-place win for the self-developed BrainAuGPT model in a national competition, coupled with selection as a typical AI application case at a major global conference, signifies authoritative recognition from national platforms for the group's technological innovation and clinical implementation in the medical vertical large model field. The breakthrough in "brain-heart axis" theory provides high-level evidence for the dual benefits of digital therapeutics in patients with cardiocerebrovascular comorbidities, confirming their potential for synergistic intervention at the brain-heart axis level and offering new clinical insights for proactive health management and non-pharmacological precision intervention in these patients.
The board believes these achievements align with the group's long-term development strategy, will further solidify its leading industry position in cognitive disorder digital therapeutics and medical vertical large models, and create long-term value for shareholders.