Professor Cheng-Yu Chen's team at Taipei Medical University Hospital has developed a "Multimodal AI-Personalized 4D Elderly Health Dementia Prediction Module." This AI-powered tool analyzes brain MRI images to accurately calculate "brain age," predict future dementia risk, and identify potential areas of brain atrophy.

Rising Dementia Risks and Limitations of Traditional Prediction Methods

In Taiwan, the prevalence of dementia among individuals over 65 years old is as high as 8%, and globally, 75% of the population is in a sub-health state, placing them at high risk for dementia. Traditional dementia prediction methods, such as cognitive assessments and blood tests, are limited by their reliance on data from a single point in time, making it difficult to accurately predict future dementia risk and brain atrophy, and hindering early intervention and treatment opportunities. The "Multimodal AI-Personalized 4D Elderly Health Dementia Prediction Module," developed by Professor Chen's team, uses deep learning technology combined with large international dementia databases (ADNI, J-ADNI, NIFD, UK Biobank) and clinical validation. With just a single brain MRI scan, it can generate a comprehensive brain aging trajectory.

Four Integrated Modules for Accurate Brain Aging Trajectories

This prediction module integrates the following four functionalities, compatible with all medical imaging formats, and automatically generates personalized 4D brain age maps:

1. Fully Automated Image Preprocessing: Automatically corrects MRI images and produces voxel-based morphometry (VBM) maps of gray matter, white matter, and cerebrospinal fluid to analyze the degree of brain atrophy.
2. Brain Age Health Prediction: Establishes a normative model of brain atrophy aging to assess the biological brain age of the subject, with an average error of only 2.99 years.
3. Brain Image Dementia Risk Prediction: Combines imaging, age, genetic information, and uses CVAE-GAN technology to generate highly accurate future brain maps, visually presenting the brain's aging trajectory.
4. Brain Cortical Atrophy Region Prediction: Segments the VBM maps according to anatomical functional areas to analyze future brain atrophy distribution and patterns, aiding physicians in differentiating various subtypes of neurodegenerative diseases.

Accuracy of Nearly 90%, Promising Benefits for the Global Elderly Population

Professor Chen stated that this dementia prediction module has undergone prospective clinical validation at four major medical centers in Taiwan, achieving an accuracy rate of 89.7%. It has potential applications in other neurocognitive diseases, such as cognitive impairment post-COVID-19 and concussion. This innovation paves the way for more possibilities in precision medicine. In the future, the team will continue to promote the clinical application of this technology and actively expand into international markets, aiming to bring healthier and higher quality lives to the global elderly population.

Resource (mandarin)：北醫研發失智預測模型 以MRI分析腦部圖譜(VBM)，自動生成4D腦萎縮老化軌跡！