What is the clinical evidence from this study?

Study design: Observational. Population: Mild cognitive impairment (n=564). Intervention: BSWiMS model combining qMRI, cognitive assessments, CSF biomarkers, and APOE4 status vs. Other models evaluated. Primary outcome: Prediction of progression from mild cognitive impairment to Alzheimer's disease (Sensitivity 0.69, 95% CI 0.63-0.76).

What question did this study set out to answer?

This research aims to improve predictions for individuals with mild cognitive impairment transitioning to Alzheimer's disease by integrating various data sources.

May 3, 2026

Interpretable multivariate survival models: Improving predictions for conversion from mild cognitive impairment to Alzheimer's disease via data fusion and machine learning.

Key Result

The BSWiMS model combining qMRI, cognitive assessments, CSF biomarkers, and APOE4 status predicted progression from MCI to AD with a sensitivity of 0.69 (95% CI 0.63-0.76) and specificity of 0.87.

Key Points

This research aims to improve predictions for individuals with mild cognitive impairment transitioning to Alzheimer's disease by integrating various data sources.
Analyzed data from 564 participants who progressed from MCI to AD in the ADNI study.
Utilized machine learning methods to evaluate data sources, select features, and develop Cox survival models.
Implemented a cross-validation framework to optimize model performance.
The optimized model achieved a sensitivity of 0.69 (95% CI [0.63, 0.76]) and specificity of 0.87 (95% CI [0.83, 0.90]).
Combining qMRI, cognitive assessments, CSF biomarkers, and APOE e4 status reached 81% precision and 87% specificity, outperforming other models.
Cognitive scores were identified as the most relevant predictors, followed by CSF and qMRI biomarkers.

Study Design

Type

Observational (n=564)

Structured PICO

Does combining qMRI, cognitive evaluations, APOE ε4, and CSF biomarkers in interpretable multivariate survival models improve predictions for conversion from mild cognitive impairment to Alzheimer's disease?

Population

564 participants in the ADNI study with mild cognitive impairment (MCI) who transitioned to Alzheimer's disease (AD)

Intervention

Interpretable multivariate survival models (BSWiMS model) combining quantitative MRI (qMRI), cognitive evaluations, APOE ε4 status, and cerebrospinal fluid (CSF) biomarkers

Comparator

Other evaluated models

Outcome

Prediction of progression from mild cognitive impairment to Alzheimer's diseasesurrogate

Combining qMRI, cognitive assessments, CSF biomarkers, and APOE ε4 status in interpretable Cox survival models substantially improves the prediction of progression from mild cognitive impairment to Alzheimer's disease.

Main Result

Effect estimate: Sensitivity 0.69 (95% CI 0.63-0.76)

Abstract

Accurately predicting which individuals with mild cognitive impairment (MCI) will progress to Alzheimer's disease (AD) can improve patient care. This study examines the role of quantitative MRI (qMRI), cognitive evaluations, apolipoprotein Formula: see text4 (APOE Formula: see text4), and cerebrospinal fluid (CSF) biomarkers in Cox survival models to predict progression from MCI to AD. Data from 564 participants in the ADNI study, who transitioned from MCI to AD, were analyzed. The data set included 330 features encompassing qMRI, cognitive assessments, CSF biomarkers, and APOE Formula: see text4 status. Advanced machine learning (ML) methods were applied to evaluate the importance of these data sources, select relevant features, and develop interpretable Cox survival models within a cross-validation framework. The top optimized model achieved a sensitivity of 0.69, 95% CI 0.63, 0.76, and a specificity of 0.87, 95% CI 0.83, 0.90, and used all data sources. The results demonstrated that combining qMRI features with cognitive assessments, CSF biomarkers, and APOE Formula: see text4 status, analyzed using the BSWiMS model, resulted in a substantial improvement in the ability to predict progression from MCI to AD, achieving 81% precision and 87% specificity. These results exceed those obtained with other models evaluated. Finally, biomarker analysis showed that cognitive scores are the most relevant features to predict conversion, followed by CSF and qMRI biomarkers. These findings highlight the value of integrating multiple data sources in highly interpretable Cox survival models for the early identification of individuals at risk for AD.

KI fragen

Bookmark