Using an agent-based approach with deep learning models to process text and tabular data in diagnosing thyroid diseases
Abstract
The purpose of this work is to study systems for converting tabular data and algorithms for generating doctor’s report labels based on nested data formats. As a result of studying data from the biomedical domain, a pipeline system was obtained for step-by-step conversion of tabular data into hidden attachments and generation of classification labels according to the Bethesda system. For model design, an agent-based approach and transformative methods were used based on boosting the output responses of solvers to form the resulting ensemble of machine learning models. The paper proposes methods for generating and sampling final data sets based on algorithms for generating medical data and conclusions according to the Bethesda Thyroid classification. The main result is a pipeline for generating label classes using the Bethesda system. When solving problems, approaches were chosen based on the ideas of autoencoders and their modifications for distilling knowledge based on the teacher-student approach; the auxiliary architecture is based on boosting and highlighting the most important features for constructing decision trees. The proposed solution is used within the framework of a smart medical assistant system to minimize decision-making time for high-level specialists and act as an assistant for doctors starting their careers. The system automates routine tasks and improves the quality of diagnostics in the cytological domain.
Full Text:
PDF (Russian)References
Mikolov T. Efficient estimation of word representations in vector space //arXiv preprint arXiv:1301.3781. – 2013. – Т. 3781.
Juhlin C. C., Baloch Z. W. The 3rd edition of Bethesda system for reporting thyroid cytopathology: Highlights and comments // Endocrine Pathology. – 2024. – Т. 35. – №. 1. – С. 77-79.
Prokhorenkova L. et al. CatBoost: unbiased boosting with categorical features //Advances in neural information processing systems. – 2018. – Т. 31.
Kingma D. P. Auto-encoding variational bayes //arXiv preprint arXiv:1312.6114. – 2013.
Генерация врачебных заключений и классификация по Bethesda с использованием глубокого обучения / Е. В. Боброва, А. Ж. Маканов, С. С. Основин [и др.] // International Journal of Open Information Technologies. – 2023. – Т. 11, № 10. – С. 119-129. – EDN WAVOVQ.
Fuhrer B., Tessler C., Dalal G. Gradient Boosting Reinforcement Learning //arXiv preprint arXiv:2407.08250. – 2024.
Louppe G. Understanding random forests: From theory to practice //arXiv preprint arXiv:1407.7502. – 2014.
Sheridan R. P., Liaw A., Tudor M. Light gradient boosting machine as a regression method for quantitative structure-activity relationships //arXiv preprint arXiv:2105.08626. – 2021.
Chen T., Guestrin C. Xgboost: A scalable tree boosting system //Proceedings of the 22nd acm sigkdd international conference on knowledge discovery and data mining. – 2016. – С. 785-794.
Lundberg S. A unified approach to interpreting model predictions //arXiv preprint arXiv:1705.07874. – 2017.
Wang C. Calibration in deep learning: A survey of the state-of-the-art //arXiv preprint arXiv:2308.01222. – 2023.
Vasilev R., D'yakonov A. Calibration of neural networks //arXiv preprint arXiv:2303.10761. – 2023.
Niculescu-Mizil A., Caruana R. Obtaining Calibrated Probabilities from Boosting //UAI. – 2005. – Т. 5. – С. 413-20.
Математическое ожидание https://ru.wikipedia.org/wiki/Математическое ожидание
Градиентный спуск https://ru.wikipedia.org/wiki/Градиентный_спуск
Refbacks
- There are currently no refbacks.
Abava Кибербезопасность ИБП для ЦОД СНЭ
ISSN: 2307-8162