Transformers and capsule networks <i>vs</i> classical ML on clinical data for alzheimer classification
Bibliographic record
Abstract
Alzheimer’s disease (AD) is a progressive neurodegenerative disorder and the leading cause of dementia worldwide. Although clinical examinations and neuroimaging are considered the diagnostic gold standard, their high cost, lengthy acquisition times, and limited accessibility underscore the need for alternative approaches. This study presents a rigorous comparative analysis of traditional machine learning (ML) algorithms and advanced deep learning (DL) architectures that that rely solely on structured clinical data, enabling early, scalable AD detection. We propose a novel hybrid model that integrates a convolutional neural networks (CNNs), DigitCapsule-Net, and a Transformer encoder to classify four disease stages—cognitively normal (CN), early mild cognitive impairment (EMCI), late mild cognitive impairment (LMCI), and AD. Feature selection was carried out on the ADNI cohort with the Boruta algorithm, Elastic Net regularization, and information-gain ranking. To address class imbalance, we applied three oversampling techniques: synthetic minority oversampling technique (SMOTE), oversample using adaptive synthetic (ADASYN), and SMOTE-Tomek. In the three-class setting, the CNN + DigitCapsule-Net hybrid attained 90.58% accuracy, outperforming state-of-the-art baselines that rely only on clinical variables. A tuned gradient boosting (GB) model achieved comparable performance with substantially lower computational requirements. Model interpretability was assessed with SHAP and gradient-weighted class activation map (Grad-CAM), which identified Clinical Dementia Rating-Sum of Boxes (CRD-SB), Logical Memory-Delayed Recall Total Number of Story Units Recalled (LDELTOTAL), and Modified Preclinical Alzheimer Cognitive Composite with Trails B (mPACC-TrailsB) as the most informative clinical features. This combination of predictive strength, computational efficiency, and transparent interpretation positions the proposed approach as a promising open-source tool for facilitating early AD diagnosis in clinical settings.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.002 | 0.005 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.001 |
| Bibliometrics | 0.001 | 0.001 |
| Science and technology studies | 0.000 | 0.001 |
| Scholarly communication | 0.001 | 0.002 |
| Open science | 0.001 | 0.001 |
| Research integrity | 0.001 | 0.001 |
| Insufficient payload (model declined to judge) | 0.002 | 0.001 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".