Can we use artificial intelligence to identify patients with Alzheimer's as defined by its biological markers by analysing brain scans?
Bibliographic record
Abstract
BACKGROUND: Medical advancements have led to an ageing population increasing the prevalence of Alzheimer's disease (AD), the most common form of dementia, poses a significant burden on the healthcare system. The diagnostic process is lengthy and complicated, averaging around 2 years. AD diagnosis is shifting from symptomatic assessment to physiological testing using fluid biomarkers and imaging. Neuroimaging techniques like Positron Emission Tomography (PET) measure amyloid load in dementia-related brain regions. Artificial intelligence (AI) has shown potential for its application in PET scanning to identify patients with AD. METHOD: 541 patients' medical data was obtained from the Alzheimer's Disease Neuroimaging Initiative (ADNI) study, containing normal and cognitively impaired patients, were matched to their PET scans. Input for AI was based on the amyloid load on brain regions of interest obtained by PET. AI models were trained to classify patients into normal or abnormal for each biomarker threshold: amyloid PET (680pg/ml), tau (355pg/ml), and phosphorylated tau (56pg/ml), as defined by CSF measurements. Machine learning algorithms were trained and tested on up to 10 brain regions of interest for each biomarker, grouping patients into their disease states. RESULT: The p-tau biomarker model achieved the highest test accuracy of 72.8% among all models. However, the amyloid-beta (Aβ) biomarker model achieved better overall test metrics, with an area under the curve (AUC) of 0.80, indicating reliable classification performance. The precuneus and left lingual gyrus were significant regions for the Aβ model as per the Shapley importance values. CONCLUSION: Explainable AI shows potential in AD diagnosis from amyloid PET, achieving moderate accuracy using pathologically relevant brain regions. To increase accuracy and reduce dataset imbalances, a larger dataset and utilization of tools like stratified cross validation is necessary.
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 | 0.008 | 0.042 |
| Meta-epidemiology (narrow) | 0.002 | 0.000 |
| Meta-epidemiology (broad) | 0.002 | 0.002 |
| Bibliometrics | 0.004 | 0.002 |
| Science and technology studies | 0.000 | 0.002 |
| Scholarly communication | 0.004 | 0.004 |
| Open science | 0.002 | 0.001 |
| Research integrity | 0.002 | 0.002 |
| 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".