Artificial Intelligence in Cancer Biology: Unlocking the Secrets of Tumor Metastasis

Artificial Intelligence in Cancer Biology: Unlocking the Secrets of Tumor Metastasis

Altaf Ahmad Shap Department of Bioengineering Integral University, Lucknow Lucknow, India altaf.bioinfo321@gmail.com

Salman Akhtar2*

Department of Bioengineering

Integral University, Lucknow Lucknow, India salmanakhtar18@gmail.com

Mohammad Kaleem Ahmad Khan3 Department of Bioengineering Integral University, Lucknow

Lucknow, India

Abstract

Cancer remains a leading global cause of death, with its complexity posing major treatment challenges. Artificial intelligence (AI) is transforming oncology by improving early detection, personalized therapy, drug development, and prognosis. This review examines AIs role in advancing cancer research, its integration into clinical practice, key methodologies, recent breakthroughs, and future trends. It also addresses challenges and ethical concerns in AI adoption. Ultimately, AI-driven innovations are reshaping cancer care, offering promising solutions to enhance outcomes. The article highlights how AI is paving the way for more precise, efficient, and patient-centered oncology.

Keywords: Artificial Intelligence, Convolutional neural networks, Tomur biology, Circulating tumor cells, Generative Adversarial Networks, Histopathology

1. Introduction

   Cancer is not a singular disease but a complex collection of related illnesses characterized by uncontrolled cell growth and the potential to invade or spread to other body parts. Each type of cancer can have different genetic mutations, phenotypic presentations, and responses to treatment, making it one of the most challenging diseases to manage. Traditional research methods, while impactful, often fall short in coping with the scale and complexity of modern biological data. The integration of Artificial Intelligence (AI) into cancer research represents a revolutionary step forward. AI can handle massive datasets, recognize patterns imperceptible to humans, and provide actionable insights. By leveraging machine learning (ML), deep learning (DL), and natural language processing (NLP), AI is contributing to major advances in detection, diagnosis, treatment planning, drug development, and patient monitoring. This article delves into the transformative role AI plays in transitioning from data-rich environments to life-saving cures.

2. AI in Cancer Detection and Diagnosis

   Artificial Intelligence is very helpful in improving the speed, reliability, and accuracy of some cancer screening and detection methods. The FDA has fully authorized the marketing of AI-basedsoftwareto help pathologists identify the areas in prostate biopsy images that possess cancer signals or clues[[1]]. Mammograms, like medical images, are being rapidly processed via AI to help radiologists keep their time focused on other jobs that require technical judgment. The various medical fields that have adopted AI algorithms to enhance cancer treatment options are discussed here in detail.

   Imaging and Radiomics

   After extracting the quantitative features from identified and delineated regions of interest (ROIs), medical images can be analyzed, and this referred to as conventional radiomics. The medical images can also be analyzed by analyzing an entire image or image series. The analyzed medical images from image series are often driven by emerging deep learning (DL) approachesthat are not always dependent on ROIs segmentation[2]. The medical imaging is very important for identifying various cancers, but the manual interpretation by radiologists is considered a subjective, time-consuming, and labor-intensive job. Radiological imaging is fundamental to cancer diagnosis. Traditional image interpretation is subjective and dependent on the radiologists experience. AI, intense learning models like convolutional neural networks (CNNs), have been shown to improve diagnostic accuracy and speed. These models are trained on large datasets of annotated images to learn features associated with malignancy. Radiomics, a subfield of AI, involves the extraction of large amounts of quantitative features from medical images. These features can be analyzed to predict tumor type, aggressiveness, and potential response to treatment. AI-based radiomics has demonstrated

   Table 1. AI and ML models used in the detection and prediction of various cancer types

   Cancer Type	AI/ML Models Used	Application	Key Performance Metric (AUC / Accuracy)	Reference / Source
   Breast Cancer	CNN, SVM, ANN, RF	Detection from mammogra ms	AUC: 0.88 0.95	PMC10463622, Sciencedirect, Nature
   Lung Cancer	CNN, RNN, RF, XGBoo st	Nodule detection in CT scans	AUC: 0.94	https://www.ncbi. nlm.nih.gov/pmc/a rticles/PMC69280 88/
   Prostat e Cancer	CNN, SVM, LSTM	MRI-based detection	Accuracy: ~90%	Frontiers in Oncology, 2021
   Colore ctal Cancer	CNN, RF, SVM, Decisio n Tree	Colonoscop y video/imag e analysis	AUC: 0.94	https://www.natur e.com/articles/s41 598-018-29993-2
   Gastric Cancer	DCNN, SVM, ANN, DCCN- LSC	Predicting peritoneal metastasis from CT	AUC: 0.85 0.96	PMC11735730
   Liver Cancer	CNN, XGBoo st, RF, SVM	Tumor detection via MRI/CT	Accuracy: ~91%	IEEE Transactions on Medical Imaging
   Ovaria n Cancer	ANN, SVM, RF	Prediction from ultrasound and biomarkers	AUC: 0.89	https://www.ncbi. nlm.nih.gov/pmc/a rticles/PMC81743 03/
   Brain Tumors	CNN, Deep Belief Networ k, RF	MRI-based classificatio n	AUC: ~0.93	IEEE Xplore, 2020
   Kidney Cancer	Ensem ble Models (RF, XGBoo st, MLP, SVM)	Bone metastasis prediction	AUC: 0.912	https://www.mdpi. com/2072- 6694/16/15/2700
   Non- Small Cell Lung CA	XGBoo st, RF, Decisio n Tree	Bone metastasis prediction	AUC: N/A	PMC11311270
   Pancre atic Cancer	CNN, Deep Neural Networ ks	Histopathol ogy image classificatio n	AUC: 0.90	https://www.natur e.com/articles/s41 591-019-0392-0
   Skin Cancer (Melan oma)	CNN, Mobile Net, ResNet	Image- based lesion classificatio n	AUC: 0.96	Nature, 2017

   Figure1. This figure shows how the cancer cells are released for analysis using the biopsy technique.Reproduced from: https://ars.els-cdn.com/content/image/1-s2.0-S2352551724000921-gr1.jpg)

   potential in distinguishing benign from malignant lesions and in identifying tumor heterogeneity[3].

   Histopathology and Digital Pathology

   The histopathology discipline was founded on the visual interpretation of images of cellular biology. The daw of digital imagery in pathology has propelled this traditional field into a new domain called digital pathology (DP). Teachers and pupils, local hospitals, colleges (second opinion), and homes and workplaces (home office) can all be connected in real- time by sharing digital photos and video streams. To enable spatial correlation between slides and stains, they can be linked or layered beyond what is possible with physical glass slides. Computational pathology (CPATH) works well with digital pictures for both simple counting and measuring activities and more complex machine learning (ML) tasks. Most intriguingly, machine learning (ML) may now analyze photos for characteristics that go beyond the evaluation of conventional histology (artificial intelligence, or AI). For example, it can immediately connect images to clinical data (e.g., mutations, prognosis) [4].

   Pathology remains the gold standard for cancer diagnosis. The digitization of pathology slides allows for AI-driven image analysis. Algorithms can quantify cell morphology, classify cancer subtypes, and even detect rare events such as micro- metastases with higher accuracy and consistency than manual examination.For example, Googles DeepMind developed an AI model that matches or exceeds the performance of pathologists in detecting breast cancer metastasis in lymph node biopsies. Such advances are particularly valuable in regions with a shortage of expert pathologists[5].

   Liquid Biopsy and Omics Data Integration

   A liquid biopsy is the process of analyzing readily accessible bodily fluids, such as blood, urine, or cerebrospinal fluid, to identify circulating tumor cells (CTCs), circulating tumor DNA (ctDNA), exosomes, and other tumor-derived components (Fig. 1). Without requiring repeated intrusive operations, this minimally invasive technology allows for real- time monitoring of tumor progression, offering crucial insights into cancer genetics and molecular alterations. Therefore, liquid biopsy has potential for use in early diagnosis, treatment monitoring, detecting minimal residual disease (MRD), and identifying therapeutic resistance, among other aspects of cancer care.Liquid biopsy offers a minimally invasive approach for cancer detection and monitoring, analyzing biomarkers such as circulating tumor cells (CTCs), ctDNA, and exosomes. AI algorithms can process high-dimensional omics data from liquid biopsies to identify patterns and biomarkers associated with early-stage cancers.

   The goal of ongoing research is to increase the sensitivity, specificity, and cost-effectiveness of liquid biopsy. To provide a more thorough understanding of cancer biology, multi-omic approaches that include information from ctDNA, CTCs, exosomes, and other biomarkers, are being investigated. Furthermore, the potential for liquid biopsy to identify malignancies early, before symptoms appear, is increasing with the discovery of novel markers, especially in high-risk groups. Advances in molecular biology, genetics, and bioinformatics are propelling the rapid evolution of liquid biopsy[6].

   Integration of genomics, transcriptomics, proteomics, and metabolomics through AI enables a comprehensive understanding of tumor biology. Multimodal AI models can correlate molecular profiles with clinical outcomes, aiding in early detection and risk assessment. Second, by combining data from several omics levels, it is possible to identify important regulatory components that serve as crucial checkpoints in cellular pathways, exposing the intricate feedback loops and interactions that regulate cellular functions. Lastly, a biological system can only be partially understood from individual omics datasets. Their combination will improve the forecasting ability of computer models that try to link phenotype and heredity. There are many different kinds of cells in the human body. The arrangement of these cells is hierarchical: tissues are made up of cells, organs are made up of tissues, and organs work together to generate a functional organism. Hormones and neurotransmitters are examples of chemical signals that cells use to communicate. It is now feasible to view and measure heterogeneous cellular processes and cell-cell communications across an organism's hierarchical tiers at single-cell resolution due to recent developments in single-cell and spatial omics approaches [7].

3. AI in Precision Oncology

   Precision medicine is becoming more and more possible because of artificial intelligence (AI), which is also transforming society and healthcare. Similar to this, immunotherapy in oncology (IO) has revolutionized cancer treatment through new therapeutic pathways, but has also produced unusual response patterns that put conventional approaches of response evaluation to the test. This editorial examines how AI might help address these issues by creating novel biomarkers for accurate disease characterization, particularly those based on imaging for the early response evaluation of cancer patients receiving IO treatment. A new era of precision medicine driven by non-invasive, imaging- based disease evaluation could be ushered in by effectively utilizing AI-based methodologies [8].

   Genomic Profiling and Biomarker Discovery

   The study of an entity's genes required to create customized drugs is known as genomics. Clinical professionals can tailor prevention, diagnosis, and treatment strategies based on a patient's genetic background by utilizing genomic data. The situation has altered due to the quick development of high- throughput sequencing technologies, such as whole-genome sequencing (WGS) and RNA sequencing (RNA-seq), which make genomic data and insights on many pathogens and population genetic differences easily accessible. Utilizing genomic information, personalized medicine improves clinical efficacy by predicting the possibility of a disease, enhancing therapeutic responsiveness, and reducing the risk of adverse drug reactions [9]. Precision oncology relies on understanding the genetic makeup of tumors to tailor treatments. AI plays a crucial role in analyzing genomic sequencing data to identify mutations, gene fusions, and copy number variations that may drive cancer. AI algorithms can uncover novel biomarkers by detecting hidden patterns within vast genomic datasets.For instance, AI-based platforms like IBM Watson for Genomics interpret patient-specific genomic data and match mutations with potential therapeutic options based on curated scientific literature and clinical trial databases.

   Predictive Modeling for Therapy Response

   Accurate patient outcome prediction is essential in the healthcare industry to deliver prompt and efficient interventions. Traditional risk assessment techniques

   Figure 2. Role of AI in various domains of drug discovery and development. Reproduced from: https://chemistry- europe.onlinelibrary.wiley.com/doi/10.1002/cbic.202300816

   frequently fail to adequately capture the dynamic and intricate character of patient circumstances. Predictive analytics powered by artificial intelligence (AI) offers a promising way to improve patient outcomes and prognosis accuracy. Notwithstanding the increased interest in AI applications in healthcare, a thorough understanding of how they affect patient outcomes and the identification of areas for development are still necessary [10].

   AI helps predict which patients are likely to respond to certain treatments, thereby avoiding ineffective therapies and reducing side effects. Supervised learning models use historical patient data to train predictive models that anticipate treatment outcomes.In immunotherapy, AI can analyze tumor mutational burden, neoantigen load, and immune infiltration to forecast responsiveness. These models are being integrated into clinical practice to guide treatment decisions in melanoma, lung, and bladder cancers, among others.

   Clinical Decision Support Systems (CDSS)

   AI has just lately become a disruptive force in healthcare, especially in the field of CDSS. AI iproves patient care delivery and results by utilizing sophisticated computational approaches to support healthcare professionals' decision- making. The three main AI technologies driving innovation in CDSS includes machine learning algorithms, natural language processing, and deep learning models. These are extremely intelligent helpers that evaluate enormous volumes of patient data to forecast future health problems and recommend the best course of action. Machine learning techniques enable CDSS to assess patient data and forecast outcomes, supporting diagnosis and treatment planning.

   By extracting insights from clinical literature, NLP enables CDSS to expedite documentation and make data retrieval easier. In order to improve diagnostic precision and individualized patient care, deep learning modelsan even more sophisticated type of machine learning that is particularly good at identifying intricate patterns automatically extract intricate patterns from a variety of medical data, such as pictures and sequential data like ECGs. With the goal of enhancing healthcare decision-making and patient outcomes, each of these technologies offers distinct capabilities, ranging from text analysis using natural language processing (NLP) to predictive analytics using machine learning and complex data pattern recognition using deep learning [11].

4. AI in Drug Discovery and Development

   Artificial Intelligence (AI) has emerged as a transformative force in drug discovery and development, offering significant improvements in speed, accuracy, cost-effectiveness, and innovation. Its integration spans across all stages of the drug development pipelinefrom target identification to post- market surveillance.

   With over 1060 compounds, the enormous chemical space encourages the creation of numerous pharmacological molecules. However, the medication development process is limited by a lack of sophisticated technologies, which makes it a costly and time-consuming operation that AI can help with. AI can identify hit and lead compounds, validate drug targets more quickly, and optimize drug structure design. Figure 2 illustrates many AI uses in drug discovery [12].

   Virtual Screening and Drug Repurposing

   Drug discovery is a lengthy and expensive process. AI accelerates this by enabling virtual screening of chemical libraries to identify compounds with potential anticancer properties. Machine learning models predict interactions between drugs and targets, significantly reducing the number of compounds that need to be tested experimentally.AI is also powerful in drug repurposingidentifying existing drugs that could be effective against different cancer types. For example, deep learning has identified metformin and propranolol as potential adjuvants in cancer therapy.

   De Novo Drug Design

   A computational method known as "de novo drug design" creates new molecular structures from atomic building pieces that don't have any preexisting associations. Structure-based and ligand-based design are examples of conventional techniques that rely on the characteristics of a biological target's active site or its known active binders, respectively. The drug discovery process has benefited from the new discipline of artificial intelligence, which includes machine learning. A subfield of machine learning called "deep reinforcement learning" blends reinforcement-learning architectures with artificial neural networks. Using a range of artificial networks, such as generative adversarial networks, recurrent neural networks, convolutional neural networks, and autoencoders, this technique has been effectively used to create innovative de novo drug discovery methodologies. Generative models such as Generative Adversarial Networks (GANs) and Variational Autoencoders (VAEs) are being used to design new drug molecules from scratch. These AI systems can generate molecules with specific properties like bioavailability, stability, and minimal toxicity.By simulating and optimizing molecules in silico, researchers can streamline lead identification and optimization, saving both time and resources[13].

   Preclinical and Clinical Trial Optimization

   AI aids in predicting drug efficacy and toxicity during preclinical stages, reducing animal use and failure rates. In clinical trials, AI supports patient recruitment by identifying eligible participants based on electronic health records and molecular profiles.

   AI also facilitates adaptive trial designs by continuously analyzing incoming data to modify trial protocols in real time, improving efficiency and success rates.

5. AI in Cancer Prognosis and Survivorship

   Prognostic Modeling

   Cancer research has long used machine learning. For almost 20 years, decision trees (DTs) and artificial neural networks (ANNs) have been utilized in the diagnosis and detection of cancer. These days, a variety of applications employ machine learning techniques, from the detection and classification of tumors using X-ray and CRT images to the classification of malignancies using proteomic and genomic (microarray) assays. The main application of machine learning has been in the diagnosis and detection of cancer. Researchers studying cancer have recently tried to use machine learning to predict and prognosticate the disease[14].AI algorithms are revolutionizing prognostic predictions by uncovering complex patterns from multi-dimensional data that are often imperceptible to human clinicians. AI can analyze high- throughput genomic data to identify prognostic biomarkers and gene signatures predictive of patient outcomes. For instance, ML-based tools have been used to classify breast cancer subtypes and predict survival based on gene expression profiles[15]. Deep learning models applied to radiographic images (e.g., CT, MRI, PET) can assess tumor heterogeneity, size, and morphology to stratify patients by risk. For example, convolutional neural networks (CNNs) have been trained to predict the likelihood of non-small cell lung cancer (NSCLC) recurrence from CT images [16]. Post-treatment, cancer survivors face long-term health challenges. AI enhances survivorship care through risk assessment, monitoring, and intervention planning. Survivors often suffer from treatment- induced comorbidities such as cardiovascular disease, fatigue, and secondary cancers. AI can predict these risks based on treatment history and patient characteristics, enabling early interventions[17]. AI systems analyze patient-specific data to generate tailored follow-up schedules and preventive care plans. This ensures timely surveillance while minimizing unnecessary procedures. Natural Language Processing (NLP) and mobile health (mHealth) apps powered by AI can monitor patient-reported outcomes (PROs), such as depression, anxiety, and quality of life, and alert clinicians when intervention is needed. AI-powered clinical decision support

   systems (CDSS) help oncologists and primary care providers manage survivorship plans and medications, improving coordination and continuity of care[18].

   Post-treatment surveillance is vital for cancer survivors. AI- powered mobile applications and wearable devices enable continuous monitoring of physiological and behavioral data. These tools can detect early signs of recurrence or complications and alert healthcare providers in real time.

   Natural language processing can also mine patient-reported outcomes and clinical notes to provide insights into quality of life and treatment side effects.

6. Integration into Clinical Workflows

   In order for AI algorithms to be clinically beneficial, they must be included into a radiologist's regular, everyday clinical workflow without creating extra "clicks" or confusion. Any level of integration could be challenging to accomplish given the complexity of contemporary clinical workflows. Researchers working on AI algorithms might not be familiar with technologies lie DICOM, PACS, and VNA, and they might test algorithms offline. A distorted view of performance could arise from offline evaluation since it does not represent how an algorithm would be applied in a genuine clinical workflow[19],[11].Effective use of AI requires harmonized data from disparate sources. Standardization efforts like Fast Healthcare Interoperability Resources (FHIR) and Digital Imaging and Communications in Medicine (DICOM) facilitate data integration and sharing across platforms.Interoperability ensures that AI systems can access and interpret data from electronic health records, imaging databases, and genomic repositories seamlessly. Despite the potential of AI, its clinical adoption depends on user acceptance. Educating healthcare professionals about AI's capabilities and limitations is essential. Transparent and interpretable models help build trust among clinicians.User-friendly interfaces, co-development with clinicians, and integration into existing workflows are critical for successful implementation[20].

7. Challenges and Ethical Considerations

   Data Privacy and Security

   AI systems frequently depend on confidential patient information. Safeguarding data privacy via anonymization, encryption, and secure storage is essential. Adherence to regulations such as HIPAA and GDPR is obligatory. Federated learning and homomorphic encryption are innovative methodologies that facilitate AI model training while safeguarding patient privacy.

   Algorithmic Bias and Fairness

   Training data bias might produce inconsistent results. Preventing inequities in cancer care requires using fairness- aware algorithms and making sure datasets are diverse. To guarantee generalizability and equity, ongoing assessment and validation across various populations are necessary.

   Regulatory and Legal Frameworks

   Clear and consistent regulatory frameworks are needed to approve and monitor AI tools in healthcare. Agencies like the FDA are developing guidelines for AI-based medical devices. Legal considerations include accountability for AI decisions and informed consent when AI is used in clinical decision- making.

8. Future Directions Explainable AI (XAI)

   Explainable AI enhances transparency by making model predictions understandable to users. XAI is crucial in high- stakes fields like oncology, where clinicians need to trust and verify AI recommendations.Techniques such as SHAP (SHapley Additive exPlanations) and LIME (Local Interpretable Model-agnostic Explanations) are gaining traction in cancer research.

   Multi-Modal Data Integration

   Future AI models will integrate data from imaging, genomics, clinical records, pathology, and wearable sensors to provide a comprehensive view of the patient. Such integration enhances predictive accuracy and enables truly personalized care.

   Global Collaboration and Open Science

   International collaboration and open data sharing accelerate AI development and validation. Initiatives like The Cancer Genome Atlas (TCGA) and the International Cancer Genome Consortium (ICGC) provide invaluable resources for AI researchers.Open-source AI platforms encourage innovation and reproducibility, fostering a collaborative research environment.

9. Conclusion

AI has the potential to revolutionize clinical practice and cancer research by facilitating more precise diagnosis, individualized treatments, early detection, and effective medication discovery. Ongoing developments in AI techniques, together with ethical and regulatory supervision, are ushering in a new age in cancer, despite the obstacles that still exist. The idea that data can be used to cure is becoming a reality thanks to teamwork. AI broadens the scope of what is feasible from a scientific and therapeutic standpoint in the

fight against cancer, in addition to complementing human competence.

Acknowledgment

We are thankful to the Department of Bioengineering, Integral University, for supporting us in conducting the work.

References

1. N. E. Sharpless and A. R. Kerlavage, The potential of AI in cancer care and research, Biochim. Biophys. Acta – Rev. Cancer, vol. 1876, no. 1, Aug. 2021, doi: 10.1016/j.bbcan.2021.188573.

2. I. Tunali, R. J. Gillies, and M. B. Schabath, Application of Radiomics and Artificial Intelligence for Lung Cancer Precision Medicine, Cold Spring Harb. Perspect. Med., vol. 11, no. 8, p. a039537, Aug. 2021, doi: 10.1101/CSHPERSPECT.A039537.

3. I. Ahmad and F. Alqurashi, Early cancer detection using deep learning and medical imaging: A survey, Crit. Rev. Oncol. Hematol., vol. 204, p. 104528, Dec. 2024, doi: 10.1016/J.CRITREVONC.2024.104528.

4. S. W. Jahn, M. Plass, and F. Moinfar, Digital Pathology: Advantages, Limitations and Emerging Perspectives, J. Clin. Med., vol. 9, no. 11, p. 3697, Nov. 2020, doi: 10.3390/JCM9113697.

5. A. H. Dur Karasayar, . Kulaç, and N. Kapucuolu, Advances in Breast Cancer Care: The Role of Artificial Intelligence and Digital Pathology in Precision Medicine, Eur. J. Breast Heal., vol. 21, no. 2, pp. 93 100, Apr. 2025, doi:

   10.4274/EJBH.GALENOS.2025.2024-12-8.

6. S. Pandey and P. Yadav, Liquid biopsy in cancer management: Integrating diagnostics and clinical applications, Pract. Lab. Med., vol. 43, p. e00446, Jan. 2025, doi: 10.1016/J.PLABM.2024.E00446.

7. Y. Wu and L. Xie, AI-driven multi-omics integration for multi-scale predictive modeling of genotype- environment-phenotype relationships, Comput. Struct. Biotechnol. J., vol. 27, pp. 265277, Jan. 2025, doi: 10.1016/J.CSBJ.2024.12.030.

8. J. McGale, M. J. Liao, E. Lopci, A. Marabelle, and L. Dercle, Artificial intelligence: A transformative tool in precision oncology, Oncotarget, vol. 15, pp. 588 589, 2024, doi: 10.18632/ONCOTARGET.28639.

9. B. Shickel, P. J. Tighe, A. Bihorac, and P. Rashidi, Deep EHR: A Survey of Recent Advances in Deep Learning Techniques for Electronic Health Record (EHR) Analysis, IEEE J. Biomed. Heal. informatics, vol. 22, no. 5, p. 1589, Sep. 2017, doi: 10.1109/JBHI.2017.2767063.

10. D. Dixon et al., Unveiling the Influence of AI Predictive Analytics on Patient Outcomes: A Comprehensive Narrative Review, Cureus, vol. 16, no. 5, p. e59954, May 2024, doi:

    10.7759/CUREUS.59954.

11. M. Elhaddad, S. Hamam, M. Elhaddad, and S. Hamam, AI-Driven Clinical Decision Support Systems: An Ongoing Pursuit of Potential, Cureus, vol. 16, no. 4, Apr. 2024, doi: 10.7759/CUREUS.57728.

12. D. Paul, G. Sanap, S. Shenoy, D. Kalyane, K. Kalia, and R. K. Tekade, Artificial intelligence in drug discovery and development, Drug Discov. Today, vol. 26, no. 1, p. 80, Jan. 2020, doi: 10.1016/J.DRUDIS.2020.10.010.

13. V. D. Mouchlis et al., Advances in De Novo Drug Design: From Conventional to Machine Learning Methods, Int. J. Mol. Sci., vol. 22, no. 4, p. 1676, Feb. 2021, doi: 10.3390/IJMS22041676.

14. J. A. Cruz and D. S. Wishart, Applications of Machine Learning in Cancer Prediction and Prognosis, Cancer Inform., vol. 2, p. 59, 2007, doi: 10.1177/117693510600200030.

15. K. Kourou, T. P. Exarchos, K. P. Exarchos, M. V. Karamouzis, and D. I. Fotiadis, Machine learning applications in cancer prognosis and prediction, Comput. Struct. Biotechnol. J., vol. 13, pp. 817, 2015, doi: 10.1016/j.csbj.2014.11.005.

16. R. Paudyal et al., Artificial Intelligence in CT and MR Imaging for Oncological Applications, Cancers (Basel)., vol. 15, no. 9, p. 2573, May 2023, doi: 10.3390/CANCER15092573.

17. R. Miotto, L. Li, B. A. Kidd, and J. T. Dudley, Deep Patient: An Unsupervised Representation to Predict the Future of Patients from the Electronic Health Records, Sci. Reports 2016 61, vol. 6, no. 1, pp. 1

    10, May 2016, doi: 10.1038/srep26094.

18. F. Jiang et al., Artificial intelligence in healthcare: past, present and future, Stroke Vasc. Neurol., vol. 2, no. 4, pp. 230243, Dec. 2017, doi: 10.1136/SVN-

    2017-000101.

19. D. J. Blezek, L. Olson-Williams, A. Missert, and P. Korfiatis, AI Integration in the Clinical Workflow,

    J. Digit. Imaging, vol. 34, no. 6, p. 1435, Dec. 2021, doi: 10.1007/S10278-021-00525-3.

20. B. A. Jantos, M. P. Wierzbicki, and M. Tomaszewski, Standardization Processes for the Implementation of AI in Healthcare System: Challenges and Limitations, pp. 6174, 2025, doi: 10.1007/978- 3-031-92734-8_7.