Artificial intelligence for dementia-Applied models and digital health

Area of Science:

• Computational neuroscience
• Geriatric medicine
• Biostatistics

Background:

• Applied modeling and digital health technologies are increasingly important for dementia risk prediction, diagnosis, and prognosis.
• Deep phenotyping cohorts with multi-omics data are becoming available, offering new opportunities for research.

Purpose of the Study:

• To synthesize the understanding of applied models and digital health technologies for dementia.
• To review their application in risk prediction, diagnostic discrimination, prognosis, and progression.

Main Methods:

• This is a narrative review.
• It synthesizes current understanding of applied models and digital health technologies in dementia research.

Main Results:

• Machine learning models demonstrate superior predictive power for dementia compared to standard clinical risk scores.
• These models can simplify complex datasets by identifying critical predictors.

Conclusions:

• Future advancements will benefit from improved data sharing, integration of multi-omics and electronic health records, and remote data measurement.
• Focus is shifting towards predicting secondary outcomes like treatment response and physical health, alongside primary dementia prediction.