# Artificial neural networks fighting real neural decline: a systematic review of AI in Alzheimer’s research

## TL;DR

This review explores how AI is transforming Alzheimer’s research, focusing on early detection, disease modeling, and therapeutic discovery.

## Contribution

The paper introduces a novel Layered Framework and ARIMA-based forecasting to categorize and project AI applications in Alzheimer’s research.

## Key findings

- Multimodal AI approaches combining neuroimaging, genetics, and clinical data show improved accuracy in Alzheimer’s detection.
- Generative models and transformer architectures are the fastest-growing AI methodologies in this field.
- Persistent challenges include limited model generalizability and underexplored clinical implementation.

## Abstract

Alzheimer’s disease (AD) is a major global health challenge, with Artificial Intelligence (AI) increasingly recognized as a transformative tool for early detection, disease progression modeling, and therapeutic discovery. This systematic review, conducted in accordance with PRISMA guidelines, analyzed 156 peer-reviewed studies published between 2010 and 2024, identified from four major databases (Scopus, PubMed, Web of Science, IEEE Xplore). A particular emphasis was placed on multimodal approaches that integrate neuroimaging, genetics, biomarkers, and clinical data to improve accuracy and translational value. To organize this fragmented field, we introduce a novel Layered Framework that categorizes AI applications into four domains: Early Detection, Disease Progression Modeling, Therapeutic Discovery, and Real-World Integration. In addition, we applied ARIMA-based forecasting to project research trajectories through 2030, which revealed generative models and transformer architectures as the fastest-growing and most promising methodologies. The review highlights substantial advances in early detection and multimodal fusion, particularly through deep learning, while also identifying persistent challenges such as limited model generalizability, ethical concerns, and underexplored clinical implementation. Addressing these barriers will require multi-cohort validation, interpretable AI, and equity-driven model development. By consolidating evidence and forecasting future directions, this review provides a roadmap for accelerating precision-driven innovations in Alzheimer’s care.