# Individualized prediction of transition from subjective cognitive decline to mild cognitive impairment based on multimodal MRI: a 10-year follow-up study

## TL;DR

This study uses brain imaging data to predict which people with early cognitive issues will develop more severe memory problems over 10 years.

## Contribution

A new individualized model using multimodal MRI data to predict progression from SCD to MCI is developed and validated.

## Key findings

- The model achieved high accuracy in predicting MCI progression with a C-index of 0.962 in training and 0.911 in testing.
- A nomogram with 10 predictors was created to estimate individual risk at 5, 7, and 10 years.
- Calibration and decision curve analysis confirmed the model's clinical utility and accuracy.

## Abstract

Predicting the transition from subjective cognitive decline (SCD) to mild cognitive impairment (MCI) is critical for dementia prevention.

Comprehensive assessment of MRI-based macro-/micro-structural and functional brain changes in SCD to develop an individualized model predicting transition to MCI.

Patients with SCD were screened from the ADNI, NACC, and OASIS-3 databases. 89 patients met the inclusion criteria and underwent structural magnetic resonance imaging (sMRI) and resting-state functional MRI (rs-fMRI). Over a 10-year follow-up, 49 patients progressed to MCI, while 40 remained stable.

The VB-net automated brain segmentation, extracting hippocampal radiomics and whole brain subregion volume features. Brain functional features were extracted based on rs-fMRI. Cox regression was used to develop predictive models, which were independently validated with the testing set. The nomogram was constructed to estimate the probability of transition to MCI at 5-/7-/10-year. The nomogram’s accuracy was assessed using calibration curves and concordance index (C-index), and clinical utility was evaluated through decision curve analysis.

The model incorporating age, brain volume, functional, and radiomics features demonstrated the highest predictive performance for SCD progression in training (C-index: 0.962; 95 % CI: 0.95–0.98) and testing (C-index: 0.911; 95 % CI: 0.861–0.968) sets. A nomogram comprising 10 predictors was constructed to estimate individualized risk of progression to MCI at 5-/7-/10-year. The calibration curve showed good agreement between predicted and observed values. Decision curve analysis demonstrated the nomogram had substantial clinical value.

This multivariate model and nomogram could accurately predict the individual progression from SCD to MCI.