Predictive Models for Cognitive Decline Assessment

Topic: Brain imaging, connectivity and assessment

Background: Cognitive decline has become a common sign of mental ability loss in mild cognitive impairment (MCI) and dementia-related diseases, such as Alzheimer’s disease (AD). However, early detection is difficult because traditional diagnostic tools such as brain scans are expensive, time-consuming, and depend on experts. Existing machine learning models show high prediction scores; however, they are uninterpretable because their logic is hidden, which limits trust. Therefore, this study aimed to develop a reliable and interpretable tool named COGNEUTEST, a clinical support cognitive decline assessment system. Methods This study utilized a tuned Random Forest classification model trained on comprehensive clinical features from the National Alzheimer’s Coordinating Center (NACC) cohort dataset. This framework focuses on the top 26 features, including patient demographics, medical conditions, lifestyle, and influential cognitive variables, to predict the objective cognitive status results. To ensure clinical accountability, a state-of-the-art SHapley Additive exPlanations (SHAP) technique was integrated to interpret each contribution. Results Experimental findings proved that tuned Random Forest showed the highest performance in identifying three cognitive classes, Normal Cognition, MCI and Alzheimer’s disease dementia, on the NACC dataset, which achieved 86.05% accuracy and an Area Under the Curve (AUC) of 0.9534. The system efficiently runs machine learning predictions and PDF report creation in an average of 13 s, and database lookups require less than 1 s. Conclusions This study contributes to the healthcare field by demonstrating the potential of providing a fast, transparent, and affordable assessment system that is accessible to low-income populations and clinics with limited budget. Limitations include reliance on a single cohort dataset and dependence on informant availability for Functional Activity Questionnaire (FAQ) variables. Future work should involve testing the model on other databases, exploring some deep learning models, and developing a separate model for cases in which a patient can perform self-testing at home.

Key findings

• This framework focuses on the top 26 features, including patient demographics, medical conditions, lifestyle, and influential cognitive variables, to predict the objective cognitive status results.
• Results Experimental findings proved that tuned Random Forest showed the highest performance in identifying three cognitive classes, Normal Cognition, MCI and Alzheimer’s disease dementia, on the NACC dataset, which achieved 86.05% accuracy and an Area Under the Curve (AUC) of 0.9534.

Study details

• Effect size86.05%
• Evidence strengthModerate

Read the original

• DOI: 10.12688/f1000research.185879.1
• Original source →

Free copy unknown: no free-copy check has an answer for this record yet. That does not mean it is paywalled.

@article{tan2026-3b7537e3,
  author  = {Tan, Y. S. and Lew, S. L.},
  title   = {{Predictive Models for Cognitive Decline Assessment}},
  journal = {F1000Res},
  year    = {2026},
  doi     = {10.12688/f1000research.185879.1},
  url     = {https://doi.org/10.12688/f1000research.185879.1}
}

Related studies in Long-term cohort studies

• Subjective Cognitive Decline Plus and Longitudinal Assessment and Risk for Cognitive Impairment JAMA psychiatry · 2024-10-01 · usefulness score 68/100
• Effects of Normative Adjustments to the Montreal Cognitive Assessment The American journal of geriatric psychiatry : official journal of the American Association for Geriatric Psychiatry · 2018-09-21 · usefulness score 45/100
• Prediction Models for Conversion From Mild Cognitive Impairment to Alzheimer's Disease: A Systematic Review and Meta-Analysis Frontiers in aging neuroscience · 2022-04-07 · usefulness score 63/100
• MRI-Based Classification Models in Prediction of Mild Cognitive Impairment and Dementia in Late-Life Depression Frontiers in aging neuroscience · 2017-02-02 · usefulness score 49/100
• The predictive validity of a Brain Care Score for dementia and stroke: data from the UK Biobank cohort Frontiers in neurology · 2023-12-01 · usefulness score 53/100
• Assessment of Advanced Glycation End Products and Receptors and the Risk of Dementia JAMA network open · 2021-01-04 · usefulness score 64/100

Similar studies

• How Reliable and Explainable Are Machine Learning Models for Dementia Detection? A Leakage-Aware Evaluation 2026 · Cognitive tests and assessment
• A Machine Learning-Based Framework for differential screening of Neurodegenerative Dementias using routine clinical data 2026 · Getting a diagnosis
• Machine learning framework for early MRI-based Alzheimer's conversion prediction in MCI subjects 2015 · Digital health tools
• Boosting power for clinical trials using classifiers based on multiple biomarkers 2010 · Digital health tools
• Forecasting Alzheimer’s disease progression using non-imaging clinical features and XGBoost 2026 · Digital health tools
• Machine learning-based estimation of the mild cognitive impairment stage using multimodal physical and behavioral measures 2025 · Cognitive tests and assessment

More on Long-term cohort studies

• A review that covers this ground External validation of four dementia prediction models for use in the general community-dwelling population: a comparative analysis from the Rotterdam Study European journal of epidemiology · 2018-05-08 · usefulness score 67/100
• Newest in this subject Ultra-sensitive Mapping of Cell-free DNA-derived Methylomes Reveals Brain-specific Epigenetic Alterations Implicated in Alzheimer’s Disease medRxiv · 2026-10-26 · Preprint — not yet peer-reviewed · usefulness score 16/100
• Newest in this subject Interpretable Phenotype Discovery Using Temporal Risk-Aware Clustering medRxiv · 2026-10-20 · Preprint — not yet peer-reviewed · usefulness score 16/100

See all studies in Long-term cohort studies →