Cumulus Neuroscience (Cumulus), a global digital health company dedicated to improving neuroscience research and patient care through advanced digital data collection, has unveiled new findings on its Symbol Swap digital cognitive assessment at the Alzheimer's Association International Conference (AAIC) 2026.

The poster, "A Brief Digital Symbol-Coding Task Outperforms Clinical Benchmarks for Alzheimer's Trial Pre-enrichment," presents interim results demonstrating the effectiveness of Symbol Swap—a two-minute, tablet-based adaptation of the traditional Digit Symbol Substitution Test used to assess executive function. The analysis combines data from one completed home-based study and two ongoing large-scale clinic-based studies.

Across all three independent studies, Symbol Swap equaled or surpassed the performance of widely used cognitive screening tools for identifying suitable participants for Alzheimer's clinical trials. Despite requiring only two minutes to complete, the assessment successfully distinguished between healthy individuals, those with mild cognitive impairment (MCI), and patients with Alzheimer's dementia. By significantly reducing both participant burden and site workload, the tool offers a practical alternative for large-scale screening.

The findings also revealed a strong relationship between Symbol Swap performance and Alzheimer's-related pathology measured through blood plasma pTau-217 biomarkers. Notably, the assessment identified subtle cognitive changes even in clinically normal participants who typically perform well on conventional cognitive tests. This suggests the tool could help pinpoint individuals more likely to test positive for amyloid or tau pathology, improving the efficiency of follow-up blood tests, PET imaging, or cerebrospinal fluid (CSF) analysis. Because the assessment is delivered digitally and scored automatically, it can be easily deployed across large, multi-center clinical trials in both home and clinic settings, serving as an effective initial screening step or as part of composite digital endpoints.

"It is remarkable that a patient-friendly test lasting just two minutes can perform as well as—or better than—traditional assessments requiring 10 to 45 minutes of clinician time, while also detecting Alzheimer's pathology in participants who appear cognitively normal using standard methods," said Brian Murphy, PhD, Co-Founder and Chief Scientific Officer of Cumulus. "These findings reinforce Symbol Swap's potential as an efficient first-line screening tool that complements biomarker testing and multimodal digital assessments. We sincerely thank all study participants and collaborators whose contributions made this research possible."

Symbol Swap was assessed across three studies: CNS-101, the first home-based validation study using the NeuLogiq Platform to measure functional neurophysiology in individuals with mild Alzheimer's dementia and healthy controls, benchmarked against the ADAS-Cog assessment; the Fastball i4i clinic-based study, compared with the Montreal Cognitive Assessment (MoCA); and the Global Alzheimer's Platform (GAP) BioHermes-002 trial conducted across 20 sites in the United States, Canada, and Europe, where it was evaluated against the Mini-Mental State Examination (MMSE). Alzheimer's pathology in the CNS-101 and Fastball i4i studies was confirmed using the AlzPath phosphorylated-tau 217 assay. The Fastball i4i and BioHermes-002 findings presented at AAIC represent interim analyses.

"The BioHermes-002 study is evaluating individuals with and without memory concerns to determine how blood-based and digital tools can identify the presence of amyloid plaques and tau tangles, which are defining characteristics of Alzheimer's disease," said Lammert Albers, Chief Commercial Officer of the Global Alzheimer's Platform. "The Cumulus data indicate that digital assessments such as Symbol Swap could help identify participants with underlying Alzheimer's pathology more efficiently, enabling more targeted biomarker testing while reducing screening costs and accelerating trial enrollment."

Cumulus also presented a second poster, "Multi-domain Digital Endpoints for Decentralized Alzheimer's Trials: Experience from the CNS-101 Study (NeuLogiq)," led by Dr. James Rowe, Professor of Cognitive Neurology at the University of Cambridge and Principal Investigator of CNS-101. The research demonstrated that the NeuLogiq Platform's digital cognitive and EEG assessments were well accepted by participants and offered sufficient sensitivity to detect meaningful clinical changes, supporting the development of smaller and more efficient Alzheimer's trials.

"Alzheimer's clinical research has traditionally relied on assessments that place considerable demands on both participants and research sites," said Dr. James Rowe.

"Our CNS-101 findings show that the NeuLogiq Platform's combined cognitive and EEG measures are not only well tolerated by individuals living with Alzheimer's disease but also sensitive enough to capture clinically meaningful changes. This level of sensitivity opens the door to designing smaller, faster, and more cost-effective studies."

According to Tina Sampath, Chief Executive Officer of Cumulus, participant recruitment remains one of the greatest challenges in Alzheimer's drug development. "A low-cost, easy-to-use digital assessment that can be completed either at home or in the clinic, automatically scored, and capable of identifying individuals most likely to be biomarker-positive has tremendous potential to reduce screening failures while easing the burden on patients and study sites," she said. "Together, the two AAIC presentations demonstrate how Symbol Swap can improve participant selection at the start of a study, while NeuLogiq's cognitive and EEG endpoints provide sensitive measures for tracking outcomes. Combined, these capabilities could substantially reduce both the cost and duration of Alzheimer's clinical trials."

Cumulus Neuroscience continues to support pharmaceutical and biotechnology partners by providing remote, multi-domain monitoring of brain function, helping advance more precise and efficient CNS clinical trials.

Click here to know more.

At the 17th Annual Meeting of the New Champions (Summer Davos 2026), organized by the World Economic Forum in Dalian on June 23, Gong Yingying, Founder and Chairwoman of Yidu Tech, joined a panel titled “Faster Drugs, Better Access?” to discuss the evolving landscape of pharmaceutical innovation. Her remarks focused on the transformative role of artificial intelligence and healthcare data infrastructure in accelerating drug discovery and improving clinical outcomes worldwide.

The discussion, moderated by Li Xin, Managing Editor of Caixin Global, also included Giovanni Caforio, Chairman of Novartis, Ren Minghui, Professor at Peking University’s School of Public Health, and Eric Tse S. Y., Chief Executive Officer of SBP Group.

Building the Foundation for AI-Driven Healthcare
China has emerged as the second-largest market for innovative drug development, with clinical research activity growing rapidly. Despite this momentum, the industry continues to face obstacles including fragmented data systems, lengthy research cycles, recruitment challenges for clinical trials, and difficulties in generating reliable evidence that supports broader patient access to new treatments.

Gong emphasized that these issues extend beyond operational inefficiencies and are rooted in the fragmented nature of healthcare data itself. She explained that data often exists in isolated systems, lacks consistency, and remains disconnected across institutions, limiting the ability of AI technologies to deliver meaningful impact.

To address this challenge, Yidu Tech has focused on developing healthcare data infrastructure as a core strategic priority.

According to Gong, an effective medical AI ecosystem consists of four key layers: computing power, foundation models, healthcare-specific foundation models, and a robust high-quality data layer. She described this final layer as an “evidence foundation,” where raw healthcare data is transformed through standardization, governance, and validation into structured, traceable, and clinically reliable information.

With this evidence-based framework in place, healthcare providers can deploy AI systems that are safe, transparent, and compliant with regulatory requirements. Gong noted that such infrastructure has become essential for improving both the speed and quality of pharmaceutical research.

A central component of this strategy is Yidu Tech’s proprietary AI platform, YiduCore. The system converts fragmented healthcare information into structured medical knowledge and research intelligence. By September 30, 2025, YiduCore had processed nearly seven billion authorized medical records, connected over 10,000 healthcare institutions, and established a disease knowledge graph encompassing virtually every known disease.

Enhancing Drug Development Through Data Infrastructure
Drawing on partnerships with leading hospitals and research organizations throughout China, Yidu Tech has expanded its data infrastructure capabilities beyond clinical settings to support the entire drug development lifecycle.

During the early stages of research, the company’s disease knowledge graph assists scientists in identifying promising therapeutic areas, evaluating study feasibility, and designing more effective clinical trial protocols. These capabilities help reduce costly missteps, improve resource allocation, and increase the likelihood of successful outcomes.

As clinical trials move into execution, the platform enables intelligent patient screening and matching across multiple locations and research sites. This approach improves recruitment efficiency, shortens enrollment timelines, and helps address common issues such as low matching accuracy and limited patient diversity.

The platform also supports AI-powered quality assurance, continuous data verification, and proactive monitoring of adverse events throughout the trial process. These features strengthen compliance, improve data integrity, and reduce operational risks.

Following drug development, Yidu Tech’s standardized evidence framework facilitates the organization of real-world data, generation of clinical evidence, and analysis of patient populations. These capabilities support post-market research, expansion into additional treatment indications, reimbursement evaluations, and broader clinical adoption, creating a continuous value chain from innovation to patient access.

Gong highlighted that while clinical trial optimization traditionally depended heavily on human expertise, the combination of standardized healthcare data and AI technologies has now become a fundamental requirement for conducting high-quality research and accelerating pharmaceutical innovation.

Expanding Medical AI Internationally While Respecting Data Sovereignty
Discussing international cooperation and healthcare data governance, Gong stressed the importance of prioritizing security, privacy, and regulatory compliance.

She noted that healthcare information is typically stored within sovereign cloud environments governed by national regulations. Independent organizations oversee data protection, access control, auditing, and compliance to ensure information remains secure and properly managed.

Gong also pointed out that healthcare differs significantly from industries that can deploy standardized global products. Medical systems are shaped by local regulations, healthcare policies, and population needs, making localization a critical factor for successful international expansion.

As an example, she cited Yidu Tech’s work in Brunei. Through a joint venture established with the Bruneian government and supported by a shared technical team, the company co-operates the country’s national digital health platform, BruHealth. The collaboration allows Yidu Tech to introduce proven AI healthcare capabilities while fully adhering to local requirements regarding data sovereignty and privacy.

Concluding the discussion, Gong reaffirmed that Yidu Tech’s continued investment in healthcare AI infrastructure is guided by a clear objective: ensuring that technological innovation ultimately delivers meaningful benefits to patients. She emphasized that technology itself is not the end goal; rather, it serves as a means to create healthcare systems that are more efficient, reliable, and accessible for everyone.