Generating <scp>MRI</scp> ‐Derived <scp>CSF</scp> Proxy‐Markers to Predict and Visualize Alzheimer's Disease Progression
Bibliographic record
Abstract
Preclinical detection of Alzheimer's disease (AD) is crucial to efficiently recruit clinical trial participants for examining AD-modifying drugs and ultimately yield clinical benefits for at-risk individuals. Cerebral amyloidosis precedes synaptic dysfunction and neurodegeneration markers, followed by the onset of AD-related cognitive impairment. To improve early AD-biomarker detection accuracy, patient data is, however, often collected via invasive procedures such as a lumbar puncture or intravenous injection of active radiopharmaceuticals. This coupled health risk is small yet significant and can be avoided by generating equally predictive or superior AD-risk staging proxy biomarkers derived from noninvasive neuroimaging modalities. In addition, using neuroimaging can provide richer insights into regional distributions of brain biomarkers of AD. Motivated by that, here we train neural networks to optimally generate latent structural MRI (sMRI) representations as proxies for cerebrospinal fluid (CSF) biomarker status on multiple classification and prediction contexts, an approach that we demonstrate has the potential to be clinically useful in screening and diagnosing AD and predicting AD progression. We found that the amygdala, hippocampus, parahippocampus, posterior and middle cingulate gyrus, middle and inferior temporal gyrus, angular gyrus, precuneus, and inferior parietal lobe regions revealed maximum attribution, thereby implying the highest prognostic value for AD risk. The proposed approach of predicting amyloid and/or tau pathology biomarkers from MRI data and subsequently transferring the MRI-derived amyloid and/or tau pathology models to predict future risk of AD progression may be useful to assist in disease screening, triage of patients for invasive testing, and efficiently determining suitability for clinical trial recruitment.
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 distilled prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.
Codex and Gemma teacher scores by category
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.001 | 0.002 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.001 | 0.001 |
| Science and technology studies | 0.001 | 0.000 |
| Scholarly communication | 0.000 | 0.000 |
| Open science | 0.000 | 0.000 |
| Research integrity | 0.000 | 0.000 |
| Insufficient payload (model declined to judge) | 0.000 | 0.000 |
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 teacher head, 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".