P2‐222: Clinical Feature vs Artificial Intelligence Feature: Risk Factor Analysis Based on Deep Learning
Bibliographic record
Abstract
Alzheimer’s disease (AD) is one of the major causes of dementia costing billions of dollars annually, which imposes enormous burden on the health care system. Due to the complexity of AD pathology, less than 50% of the patients are correctly diagnosed. Therefore, an efficient and accurate diagnosis system is of vital importance. In this work, we propose a deep artificial neural network based system. Moreover, we evaluate the heterogeneous medical data in terms of variable impact on diagnosis decision. The dataset used in this analysis is Alzheimer’s Disease Neuroimaging Initiative (ADNI) database. The data include medical imaging data and demographic information. The medical imaging data are processed by FreeSurfer to derive regional brain volumes. Demographic information, such as age, gender, level of education, year of onset, etc., is also included. We designed a deep learning neural network to evaluate the importance of medical variables. The ranking score of various medical variables are defined as how much it affects the system performance. Two ranking lists are presented from both artificial intelligence and clinical points of view. System performance is evaluated based on 2 tasks: diagnosis and prognosis. Our algorithm achieves very promising results, 85.1% for diagnosis and 80.3% for prognosis. Besides, we compare our algorithm against other algorithms: Support Vector Machines (SVM), Random Forest (RF), Decision Tree (DT) and Random Subspace (RS). The results reveal that deep learning neural network achieves advantageous results over other widely used methods. For risk factor analysis, age remains the 1st risk factor to both clinicians and proposed algorithm. However, hippocampal volume is not among the top risk factors for automatic diagnosis. Instead, the regions around hippocampus, e.g. 3rdVentricle, Occipital, etc., impose more impact on the diagnosis decision. We designed a general framework for AD diagnosis and prognosis. The proposed algorithm shows promising performance, indicating great potential for practical applications. Further analysis reveals the difference between artificial intelligence systems in terms of risk factor importance. This may due to the stronger ability of machine learning algorithms in identifying subtle changes in brain structures.
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 | 0.001 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.000 | 0.001 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.000 | 0.000 |
| Open science | 0.000 | 0.000 |
| Research integrity | 0.000 | 0.001 |
| Insufficient payload (model declined to judge) | 0.001 | 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; both teacher heads agree on what is shown here.
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".