Prediction of AL Amyloidosis Using Deep Learning
Bibliographic record
Abstract
AL amyloidosis (amyloid light chain or primary amyloidosis) is a rare protein disorder that can be potentially fatal or can cause permanent damage to the organs in the body, especially in cases where the diagnosis does not arrive early enough or where the treatment does not begin on time. It is a type of amyloidosis, which occurs when abnormal immunoglobulin light chain (LC) proteins in the body misfold and accumulate on the heart, the kidneys, and the other organs. In order to facilitate timely diagnosis of the disease before the symptoms start fully exhibiting themselves and before the damage to the organs becomes significant, we present a computational solution in this thesis, called "DALAD", which is based on (convolutional) deep learning networks and takes in an LC sequence from a patient as the input, and determines with high confidence whether the patient has the disease or not. We develop and test multiple versions of DALAD, which are characterized by the type of sequences they have been trained on and by the types of features they incorporate to make the predictions, in order to have high performance in each of these scenarios. We establish the following for DALAD. \n \n1. DALAD is the first computational learning model to be able to accurately predict the onset of AL amyloidosis on both lambda and kappa LC sequences. \n \n2. DALAD comfortably beats the state-of-the-art for lambda sequences in terms of accuracy measures, such as AUC score, sensitivity, and specificity. Our numbers for these three metrics are 0.89, 0.81, and 0.83, respectively, while for LICTOR, they are 0.87, 0.76, and 0.82, respectively. \n \n3. DALAD is able to utilize the features from both V and J gene segments of the LC sequences to make more accurate predictions. We additionally show via the pairwise t-test that the J gene segments do improve our performance against both lambda and kappa sequences. \n \n4. We provide aggregate statistics over multiple runs for each version of DALAD, along with the accuracy results for the best trained model corresponding to each version. All our findings indicate high prediction accuracy for both lambda and kappa sequences.
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 | 0.003 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.001 |
| Bibliometrics | 0.001 | 0.000 |
| 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) | 0.002 | 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".