Abstract A021: Improving cardiotoxicity prediction for oncology drugs via domain-specific adaptation
Bibliographic record
Abstract
Abstract Background: Antineoplastic drugs have significantly improved outcomes for cancer patients. However, the cardiotoxicity of antineoplastic drugs, such as anthracyclines and anti-HER2 agents, poses serious challenges in cancer care, increasing patients’ susceptibility to cardiac complications, including heart failure, hypertension, arrhythmias and coronary disease. While predictive models have recently been developed for predicting drug-induced cardiotoxicity (DICT), performance on antineoplastic drugs is less sensitive and specific than on general drugs. Methods: We developed a domain adaptation approach to improve DICT prediction, using data from the recently released FDA DICTrank dataset and the CancerDrug_DB database to identify relevant antineoplastic drugs. We used the latest publicly available cardiotoxicity models as baselines, and constructed oncology-specific models that incorporated novel domain-relevant features, meta-modelling and adaptive strategies to enhance predictive performance on antineoplastic agents, leveraging public structural, physicochemical, mechanism of action, molecular target and morphological (Cell Painting) data. Results: Preliminary results demonstrate that our domain-specific adaptation approach conferred significant increases in performance on antineoplastic drugs (holdout AUC = 0.861), compared to the best performing general DICT predictors (holdout AUC = 0.735). Domain-specific adaptation from structural features alone performed better (holdout AUC = 0.975) on antineoplastic drugs compared to the general model (holdout AUC = 0.548). Model interpretability analyses revealed unique structural, physicochemical and protein targets as key features for predicting antineoplastic DICT. Conclusion: Domain-specific adaptation is a promising approach to DICT prediction for antineoplastic drugs. The strong performance of DICT predictors developed from both in-silico and biological data can enhance the safety and efficiency of the oncology drug development process, especially in early-stage compound screening. Citation Format: Daniel Nguyen, David L. Nguyen, Richard Zhang, Jesse T. Chao. Improving cardiotoxicity prediction for oncology drugs via domain-specific adaptation [abstract]. In: Proceedings of the AACR Special Conference in Cancer Research: Artificial Intelligence and Machine Learning; 2025 Jul 10-12; Montreal, QC, Canada. Philadelphia (PA): AACR; Clin Cancer Res 2025;31(13_Suppl):Abstract nr A021.
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) | 0.002 | 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".