Zhitong Finance App News, Brain Action Aurora-B (06681) issued an announcement. The Group has recently made a number of important breakthroughs in vertical model technology research and development and clinical research. The relevant information is now announced as follows:
1. Won the runner-up in the 3rd Xiong'an Vertical Large Model (Smart) Application Competition
The final of the 3rd Xiong'an Future City Scene Huiong'an Vertical Model (Smart) Application Contest was held at Xiong'an Science and Technology Innovation Park in early July. The competition was hosted by the Xiong'an Future City Scene Conference Organizing Committee. More than 700 artificial intelligence projects from home and abroad participated in the competition, 90 high-quality projects advanced to the finals, and 52 top national teams competed on the same stage.
The “Exploration and Application of Personalized Intervention Techniques Based on the Big Model of Cognitive Deep Reasoning in the Medical Field” project submitted by the Group won the runner-up in the “Artificial Intelligence +” new scenario circuit with the self-developed BraPT cognitive deep reasoning model. Based on the BraPT cognitive deep reasoning model, the team built an industry-unique “cloud-edge-end” intelligent evaluation system and a “thousand people, a thousand square” personalized AI intervention system, relying on three original technologies: fully mining graph structure behavioral data, proposing LLM fine-tuning technology for second-order knowledge expression, Monte Carlo tree search and decoding technology for spatio-temporal uncertainty, and big model decoding technology with cross-layer entropy enhancement to form a closed loop of “perception - reasoning - decision - feedback - optimization”.
At the same time, the “Exploration and Application of Smart Agents in the Medical Field Based on Time Series Knowledge Maps and Large Cognitive Models” project of Beijing Zhizheng Technology Co., Ltd., a wholly-owned subsidiary of the Group, was also successfully selected for the 2026 Global Digital Economy Conference “Typical Case of Artificial Intelligence Empowerment Scenario Application (2026)” this month.
2. The latest clinical research results were published in “Alzheimer's Research & Therapy”
Recently, the clinical study “Neural Changes After Computerized Cognitive Training in Coronary Heart Disease with Mild Cognitive Symptoms: Secondary Analysis of a Randomized” hosted by Professor Zeng Yong's team from Beijing Anzhen Hospital affiliated to Capital Medical University and with the Group's deep participation “Clinical Trial” was published in the authoritative international academic journal “Alzheimer's Research & Therapy” (impact factor 8.9, JCR Q1 area).
The study was a pre-set secondary analysis of a multi-center, double-blind, actively controlled randomized clinical trial (NCT05735041). 185 patients with coronary artery disease complicated with mild cognitive impairment were included. Through multi-modal magnetic resonance imaging, the effects of multi-domain adaptive computerized cognitive training on the structural and functional plasticity of brain networks were systematically investigated. Discuss the mechanism of action of the “brain-heart axis” in non-drug cognitive rehabilitation. After 12 weeks of multi-domain difficulty adaptive computerized cognitive training, the core product of the Group can significantly increase the functional connectivity between the parietal lobe memory network (PMN) (left PMN, p = 0.028) and contextual correlation network (CAN) (left CAN, p = 0.045; right CAN, p = 0.031) in patients with coronary artery disease complicated with mild cognitive impairment. The volume of gray matter in the patient's right anterior cingulate lobe (p = 0.048) and right parahippocampal cortex (p = 0.032) increased significantly. Structural connectivity between the “inferior parietal lobe” (p = 0.016) and the “parahippocampal-suprafrontal gyrus” (p = 0.044) white matter pathways was markedly enhanced. Furthermore, the study revealed a close link between changes in brain network connectivity and increased patients' cognition and decreased blood pressure/pulse pressure. It was found that the increase in functional connectivity between PMN on the right side of patients and network pathways such as cingulate reversus-dominant network (CON) was significantly negatively correlated with changes in systolic pressure and pulse pressure. Mediating effect analysis further showed that changes in the functional connection between PMN and CON on the right side significantly mediated the therapeutic effect of cognitive intervention on the reduction of patients' pulse pressure (indirect effect = 0.62, p = 0.040).
Research conclusion: Multi-domain adaptive computerized cognitive training can promote functional and structural neuroplasticity of memory-related circuits in patients with coronary heart disease and mild cognitive impairment. Changes in functional connectivity are related to cognitive improvement and may also mediate lower blood pressure, suggesting that multi-domain adaptive computerized cognitive training may participate in brain-heart interaction.
III. Significance of the above results to the Group
The above awards and scientific research results highlight the Group's comprehensive competitiveness: the self-developed BraPT cognitive deep reasoning model won the runner-up in the 3rd Xiong'an Vertical Large Model (Smart Body) Application Competition, and was also selected as a typical case of artificial intelligence enabled scenario application at the 2026 Global Digital Economy Conference, marking that the Group's technological innovation and clinical implementation results in the field of medical vertical models were recognized by national platforms; the breakthrough in the “brain-heart axis” theory provided high-level evidence-based evidence for the dual benefits of digital therapy in patients with cardiovascular comorbidities. The therapeutic potential for collaborative intervention at the “brain-cardiac axis” level provides new clinical ideas for active health management and non-pharmacological precision intervention for patients with cardiovascular and cerebrovascular comorbidities.
The board of directors believes that the above results are in line with the Group's long-term development strategy and will further consolidate the Group's leading position in the field of digital therapy and medical vertical models for cognitive impairment, and create long-term value for shareholders.