Advancing Precision Oncology Using Data-Driven Machine Learning Approaches
Tatli, Özge LU and Kazi, Julhash U. LU
(2026)
In Cancer Biome and Targeted Therapy
1(1).
p.170-196
- Abstract
- Precision oncology is being transformed by the integration of advanced machine learning (ML) methods and extensive biomedical data from genomics, imaging, proteomics, and clinical records. ML techniques, including supervised, unsupervised, deep learning, and reinforcement learning, have progressed from experimental tools to robust systems that identify clinically actionable biomarkers, refine prognosis, and guide personalized therapies. Deep learning models now achieve expert-level performance in tumor detection, grading, and outcome prediction from digital pathology and radiological images, improving diagnostic precision and therapeutic decision-making. Multi-modal and graph-based fusion networks enable the creation of patient-specific... (More)
- Precision oncology is being transformed by the integration of advanced machine learning (ML) methods and extensive biomedical data from genomics, imaging, proteomics, and clinical records. ML techniques, including supervised, unsupervised, deep learning, and reinforcement learning, have progressed from experimental tools to robust systems that identify clinically actionable biomarkers, refine prognosis, and guide personalized therapies. Deep learning models now achieve expert-level performance in tumor detection, grading, and outcome prediction from digital pathology and radiological images, improving diagnostic precision and therapeutic decision-making. Multi-modal and graph-based fusion networks enable the creation of patient-specific digital twins that simulate treatment responses and optimize therapeutic strategies. Data-centric methodologies such as federated learning, differential privacy, and synthetic data generation address challenges related to data sharing and patient privacy. Additionally, large language models trained on biomedical literature are increasingly integrating structured and unstructured clinical data, thereby fostering hypothesis generation and natural language–based decision support. However, challenges, including data heterogeneity, interpretability, algorithmic bias, and regulatory and ethical constraints, remain. Rigorous benchmarking, explainable AI methods, and prospective multi-center trials are essential for validating ML tools and establishing clinician trust. This review discusses recent developments in next-generation ML for precision oncology. (Less)
Please use this url to cite or link to this publication:
https://lup.lub.lu.se/record/271c3c23-74a5-4051-8f5a-a6fa0371a1ab
- author
- Tatli, Özge
LU
and Kazi, Julhash U.
LU
- organization
- publishing date
- 2026-01-24
- type
- Contribution to journal
- publication status
- published
- subject
- in
- Cancer Biome and Targeted Therapy
- volume
- 1
- issue
- 1
- pages
- 170 - 196
- language
- English
- LU publication?
- yes
- id
- 271c3c23-74a5-4051-8f5a-a6fa0371a1ab
- alternative location
- https://cancerbiometherapy.com/index.php/cbtt/article/view/18/97
- date added to LUP
- 2026-02-06 21:04:54
- date last changed
- 2026-02-09 07:49:02
@article{271c3c23-74a5-4051-8f5a-a6fa0371a1ab,
abstract = {{Precision oncology is being transformed by the integration of advanced machine learning (ML) methods and extensive biomedical data from genomics, imaging, proteomics, and clinical records. ML techniques, including supervised, unsupervised, deep learning, and reinforcement learning, have progressed from experimental tools to robust systems that identify clinically actionable biomarkers, refine prognosis, and guide personalized therapies. Deep learning models now achieve expert-level performance in tumor detection, grading, and outcome prediction from digital pathology and radiological images, improving diagnostic precision and therapeutic decision-making. Multi-modal and graph-based fusion networks enable the creation of patient-specific digital twins that simulate treatment responses and optimize therapeutic strategies. Data-centric methodologies such as federated learning, differential privacy, and synthetic data generation address challenges related to data sharing and patient privacy. Additionally, large language models trained on biomedical literature are increasingly integrating structured and unstructured clinical data, thereby fostering hypothesis generation and natural language–based decision support. However, challenges, including data heterogeneity, interpretability, algorithmic bias, and regulatory and ethical constraints, remain. Rigorous benchmarking, explainable AI methods, and prospective multi-center trials are essential for validating ML tools and establishing clinician trust. This review discusses recent developments in next-generation ML for precision oncology.}},
author = {{Tatli, Özge and Kazi, Julhash U.}},
language = {{eng}},
month = {{01}},
number = {{1}},
pages = {{170--196}},
series = {{Cancer Biome and Targeted Therapy}},
title = {{Advancing Precision Oncology Using Data-Driven Machine Learning Approaches}},
url = {{https://cancerbiometherapy.com/index.php/cbtt/article/view/18/97}},
volume = {{1}},
year = {{2026}},
}