A vision transformer approach for fully automated and scalable dementia screening using clock drawing test images
Bibliographic record
Abstract
INTRODUCTION: The clock drawing test (CDT) screens for dementia but requires trained scorers and lacks standardized criteria. Thus, we developed an automated vision transformer (ViT)-based diagnostic system with convolutional neural network preprocessing for analyzing hand-drawn CDT images. METHODS: = 862; 522 dementia, 340 normal cognition). RESULTS: The ViT approach predicted dementia with 76.5% balanced accuracy, outperforming human-scored features (74.3%) and existing deep learning models (MiniVGG = 73.3%, MobileNetV2 = 72.3%, relevance factor variational autoencoder = 69.1%) on the TDRA dataset. DISCUSSION: This pen-and-paper compatible diagnostic system enables scalable remote cognitive screening through automated CDT image analysis that is competitive with human-scored features, potentially increasing diagnostic accessibility for diverse populations across varied socioeconomic contexts. HIGHLIGHTS: The vision transformer model achieves 76.5% accuracy in dementia detection from clock drawing tests, outperforming human scoring and existing deep learning methods.Novel convolutional neural network-based preprocessing automatically handles challenging image quality issues like shadows, irrelevant markings, and improper cropping.The system requires only a photo of a hand-drawn clock test, enabling scalable remote screening accessible across socioeconomic contexts.A feature-extraction model trained on 54,027 samples demonstrates robust generalization to an independent clinical dataset of 862 patients.This fully automated approach eliminates the need for trained scorers while maintaining diagnostic accuracy above manual methods.
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 | 0.001 | 0.002 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.001 | 0.001 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.001 | 0.001 |
| Open science | 0.001 | 0.001 |
| Research integrity | 0.001 | 0.001 |
| Insufficient payload (model declined to judge) | 0.003 | 0.002 |
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".