Abstract 504: A comparative analysis of statistical and machine learning approaches to predict drug resistance based on synergistic genetic alterations
Notice bibliographique
Résumé
Introduction: This study evaluates the utility of a machine learning approach to predict drug resistance in breast cancer (BC), based on comparison to an automated statistical model, using data from cBioPortal. Methods: Our in silico approach investigated the association between synergistic genetic alterations in BC and survival data with various treatments based on one study (TCGA, PanCancer Atlas 2018). A set of 491 commonly altered genes in cancer (based on TSO500) was analyzed for this study. We programmed a statistical model to automatically select pairs of genes whose alterations (mutations, structural variants/fusions, copy number alterations) correlated with significantly worse five-year overall survival (OS) upon synergistic alterations, compared to one gene alone (p<0.05; HR≥1). The results were analyzed via subgroup analysis of six treatment groups according to cBioPortal (chemotherapy, ancillary, hormone, immunotherapy, radiation, other) to predict potential drug resistance if HR significantly changed based on treatment. We compared our results with a machine learning approach. The model first computed relevancy/attention scores on treatments based on order, duration and type. This allowed the model to consider chronological order and simultaneous treatment administration. We then fed this into a model that returned a classification score (probability of death before 60 months) with ∼90% accuracy, and another model that returned a regression score (expected survival in months) score, with mean squared error<0.02. Results: A statistical brute force approach tested each gene's influence on another, and each treatment category’s influence on a gene pair. Permutations of pairs from the geneset (240590 total) were generated for each treatment; 570 corresponded with significantly worse OS (HR≥1) with chemotherapy, 392 with hormone therapy, 124 with immunotherapy and 614 with radiation therapy. Less than five patients had ancillary or other treatment, preventing us from generating statistically significant results. Permutations were used since the order of genes matters during statistical analysis for alterations to one gene versus synergistic alterations. The machine learning approach yielded treatment rankings for each gene permutation based on the classification score. The latest test demonstrated that most gene permutations correspond with a high classification score with hormone therapy (48%), and most gene permutations correspond with worse total OS with chemotherapy (86%). Rankings show promise that some marked values from statistical analysis are shared with the machine learning model, but further work is needed to improve stability to make accurate predictions for survival outcomes. Conclusion: Our findings suggest the possible utility of an automated approach to predict gene pairs that confer drug resistance in BC. Citation Format: Rishi Nair, Nicholas R. Mistry, Roy Khalife, Anthony M. Magliocco. A comparative analysis of statistical and machine learning approaches to predict drug resistance based on synergistic genetic alterations [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2025; Part 1 (Regular Abstracts); 2025 Apr 25-30; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2025;85(8_Suppl_1):Abstract nr 504.
Récupéré en direct depuis OpenAlex et désinversé. Les résumés ne sont pas conservés dans cette base de données : les index inversés représentent 8,6 Go des 9,3 Go de texte de la base, et le serveur dispose de 13 Go libres.
Comment cette classification a été obtenuedéplier
Prédiction machine sur la base complète
Imitation des enseignantsNi prévalence calibrée, ni vérité terrain. Validation humaine à venir. Le volet Gemma est une étiquette directe du modèle pour chaque travail de la base, lue sur la notice réduite au titre. Le volet Codex est un classifieur appris des 10 348 étiquettes directes de Codex et calibré sur les taux pondérés de l'échantillon; les champs sans appui suffisant ne portent aucun appel Codex. Le mode candidate est l'union des deux volets; le consensus est leur intersection. Ces sorties portent le statut machine_predicted_unvalidated et ne sont pas des étiquettes humaines.
Scores du classifieur distillé par catégorie (deux têtes)
| Catégorie | Codex | Gemma |
|---|---|---|
| Métarecherche | 0,006 | 0,011 |
| Méta-épidémiologie (sens strict) | 0,001 | 0,000 |
| Méta-épidémiologie (sens large) | 0,001 | 0,002 |
| Bibliométrie | 0,004 | 0,001 |
| Études des sciences et des technologies | 0,000 | 0,000 |
| Communication savante | 0,001 | 0,001 |
| Science ouverte | 0,001 | 0,001 |
| Intégrité de la recherche | 0,001 | 0,000 |
| Charge utile insuffisante (le modèle a refusé de juger) | 0,003 | 0,001 |
Scores machine (provisoires)
Les deux têtes enseignantes du modèle étudiant, lues sur ce travail. Un score ordonne la base pour la relecture; il n'affirme jamais une catégorie, et le statut de validation accompagne chaque rangée tel quel.
Scores de référence d'un modèle non mature (critères de maturité non atteints, 7 itérations). Un score ordonne; il n'affirme jamais une catégorie.
score_only:v0-immature-baseline · tel quel depuis la passe de notation : score_only signifie que le nombre peut ordonner les travaux, et qu'aucune étiquette de catégorie n'en découleClassification
machine, non validéePrédiction automatique; un appel candidat d’une seule source (Gemma direct ou Codex distillé), pas un consensus.
Le détail, modèle par modèle et score par score, se trouve en fin de page sous « Comment cette classification a été obtenue ».