# Exploration of biomarkers of Alzheimer’s disease based on orthogonal multi-task canonical correlation analysis

## TL;DR

This paper introduces a new algorithm to find Alzheimer's disease biomarkers by combining brain imaging, genetics, and gene expression data.

## Contribution

The novel contribution is the orthogonal multi-task sparse canonical correlation analysis (MTOSCCA) algorithm with orthogonal constraints to reduce redundant features.

## Key findings

- The MTOSCCA algorithm showed better correlation performance in integrating AD data types.
- The identified biomarkers have diagnostic significance for Alzheimer’s disease and mild cognitive impairment.

## Abstract

As a neurodegenerative disease, Alzheimer’s disease (AD) has many symptoms, such as memory impairment, cognitive decline, and personality change. Image genetics is the correlation analysis between imageology and genetics, and image genetics research can effectively detect the biomarkers of AD. This paper proposed an orthogonal multi-task sparse canonical correlation analysis (MTOSCCA) algorithm. Based on the multi-task canonical correlation analysis, this algorithm added orthogonal constraints to the weight vectors U and V, which can effectively prevent the influence of redundant features on the results. In this paper, the MTOSCCA algorithm was applied to structural magnetic resonance imaging, single nucleotide polymorphism, and gene expression data integration. The results showed that the proposed algorithm has better correlation performance, and the obtained markers have diagnostic significance for AD and mild cognitive impairment (MCI).

The online version contains supplementary material available at 10.1186/s12880-025-01782-2.