Revolutionizing Cancer Diagnostics and Personalized Treatment: The Transformative Impact of Artificial Neural Networks

Abstract

Artificial Neural Networks (ANNs) have become increasingly influential in cancer diagnostics, particularly for analyzing histopathological images and genomic data. This study explores the use of deep learning models, specifically Convolutional Neural Networks (CNNs), to improve diagnostic accuracy and personalize treatment strategies for cancer patients. By incorporating Artificial Intelligence (AI) technologies, the research aims to optimize the diagnostic process and enhance patient outcomes. CNN models have achieved diagnostic accuracy improvements of up to 99.18% in breast cancer detection compared to traditional pathology methods. The findings demonstrate that AI significantly accelerates the identification of cancerous cells and predicts patient responses to therapies, emphasizing its transformative potential in precision medicine. Personalization was achieved through genomic-based analyses to identify mutation-specific therapies and image-based treatment planning tailored to individual patient profiles. These results highlight the expanding role of AI in advancing cancer diagnostics and personalized treatment.

Keywords: Cancer Diagnostics, Convolutional Neural Networks, Deep learning, Precision medicine

Introduction

Artificial Neural Networks (ANNs) are computational models designed to recognize patterns and make decisions based on data input¹. These networks are pivotal in advanced fields, including cancer research, due to their ability to enhance image recognition and natural language processing capabilities. ANNs are structured into three primary layers: the input layer, where data is introduced into the system; one or more hidden layers, which are responsible for data processing; and the output layer, which generates the results based on the processed information². The complexity and effectiveness of ANNs are primarily dictated by the number and intricacy of the hidden layers. Networks with only one or two hidden layers are generally classified as simple neural networks, while those with a greater number of hidden layers are referred to as deep learning networks. These deep learning models, with their multiple layers, are particularly adept at learning and extracting complex features from data, making them exceptionally valuable for tasks such as image recognition and natural language processing – areas of significant importance in cancer research.

Building on the foundational understanding of Artificial Neural Networks (ANNs) and their applications, it is crucial to explore their specific implementation and impact on cancer research. In their article from Nature Reviews Cancer, Perez-Lopez and colleagues offer a comprehensive analysis of this topic, particularly focusing on deep-learning models used for histopathological analysis³. Figure 1 [adapted from Perez-Lopez et al., Nature Reviews Cancer] illustrates the development process of these models. This figure depicts how convolutional neural networks (CNNs) are trained on large datasets of histopathological images to identify patterns indicative of cancerous cells.

• Figure 1(a): Collection and pre-processing of histopathological images. Large datasets of digital slides, often sourced from biopsy samples, are compiled and prepared for analysis. This step is crucial as the quality and diversity of the training data significantly impact the model’s ability to generalize and accurately identify cancerous patterns.

• Figure 1(b): CNN architecture, wherein layers of neurons are organized to process these images. CNNs are particularly well-suited for image analysis due to their ability to capture spatial hierarchies in data. They use a series of convolutional layers that apply filters to the input image, extracting features such as edges, textures, and more complex structures. This hierarchical feature extraction is what makes CNNs especially effective in visual imagery analysis.

• Figure 1(c): Training process. During training, the CNN learns to associate specific patterns in the images with corresponding diagnostic labels (e.g., cancerous vs. non-cancerous). The model adjusts the weights of its neurons through backpropagation, a process that minimizes the error in its predictions by iteratively updating the weights based on the difference between its output and the actual labels.

• Figure 1(d): Application of the trained model in real-world diagnostic settings. Once the CNN has been trained and validated, it can be used to analyze new histopathological images. The model’s predictions can assist pathologists by highlighting areas of interest on slides, potentially reducing diagnostic errors and speeding up the review.

Methods

Convolutional Neural Networks (CNNs) are a type of deep neural network that excels in visual imagery analysis due to their unique architecture, making them particularly effective in identifying subtle patterns in histopathological images indicative of cancerous cells⁴.

The CNN architecture in this study consisted of five convolutional layers with kernel sizes of 3×3 and ReLU (Rectified Linear Unit) activation functions, followed by max-pooling layers to reduce dimensionality. An Adam optimizer was employed for its adaptive learning capabilities, and categorical cross-entropy was used as the loss function. Pre-trained models, such as ResNet-50, were fine-tuned on cancer-specific datasets to enhance diagnostic performance while reducing the training time. Designed to process and analyze visual data through convolutional layers that adaptively learn spatial hierarchies of features, CNNs provide highly accurate diagnostic predictions⁵.

This capability highlights the significant role of AI in advancing precision medicine, which tailors medical care to individual patients based on specific characteristics like genetic makeup and environmental factors⁶. By automating pathology slide analysis, CNNs reduce the workload on pathologists, enhance diagnostic accuracy and efficiency, and ensure patients receive timely and precise diagnoses, ultimately improving treatment outcomes⁷.

The implementation of AI, particularly CNNs, in histopathological analysis offers numerous benefits⁸. By automating the analysis of pathology slides, CNNs allow pathologists to shift their focus from routine data processing to tasks requiring critical thinking, thereby making better use of their expertise⁹. Additionally, CNNs enhance diagnostic accuracy and efficiency, both of which are crucial for timely and effective cancer treatment. Early and accurate detection of cancerous cells facilitates prompt intervention, which is essential for improving patient outcomes¹⁰.

On a broader scale, Artificial Neural Networks (ANNs) are computational models inspired by the structure and function of the human brain¹¹. These networks consist of interconnected layers of nodes, analogous to neurons in the human brain, which process input data to recognize patterns and make decisions. This architecture allows ANNs to adapt and learn from data, making them versatile tools for a wide range of applications¹². A visual representation of ANNs would effectively illustrate how these layers connect and interact, enhancing understanding of their structure and functionality¹³.

Results

Advancing AI in Cancer Research: From Specialized Models to Generalist Systems

The evolution of Artificial Intelligence (AI) in cancer research has been marked by significant advancements, from simple, rule-based systems relying on manually selected features to sophisticated deep learning models capable of automatic feature extraction and handling diverse data types and medical tasks¹⁴. For example, early CADx systems in the 1990s depended heavily on predefined features like shapes and textures, whereas modern CNN models, such as AlexNet introduced in 2012, automatically learn hierarchical features directly from raw image data. These advancements significantly improved generalizability and diagnostic accuracy, with CNNs achieving up to 99% accuracy in certain cancer imaging tasks¹⁵.  This development can be traced through a timeline of machine learning techniques from 1990 to the envisioned future.

Timeline of AI Model Progression (1990 – Envisioned Future)

• 1990: The early AI models in medical imaging were grounded in supervised machine learning, relying heavily on fully annotated datasets. Each image was meticulously labeled by experts, and the focus was on manually selected features that could distinguish between normal and abnormal tissues. For instance, researchers would identify specific shapes, textures, or patterns in pathology slides to determine whether a sample was cancerous. Classic tools from this period include early CADx (Computer-Aided Diagnosis) systems, which were used to support radiologists in detecting abnormalities like microcalcifications in mammograms. These tools, though innovative, were limited by their reliance on predefined features and lacked adaptability to complex, unseen cases.

• 2012: A significant leap forward occurred with the introduction of supervised deep learning. The advent of Convolutional Neural Networks (CNNs) allowed AI models to learn hierarchical features directly from raw image data. These deep learning models, which required large, labeled datasets, offered improved performance and generalizability compared to earlier methods. The ability to automatically extract relevant features from images without manual intervention revolutionized the field of medical imaging. For example, the AlexNet architecture demonstrated groundbreaking performance on image classification tasks and inspired applications in cancer imaging, such as lymph node metastasis detection in breast cancer pathology slides. However, these systems required extensive computational resources and large datasets, which posed a barrier to widespread adoption during this period.

• 2020: The next major advancement was the emergence of multimodal models. These models integrated information from various data modalities, such as radiology images, pathology slides, and genomic data, to provide a more comprehensive analysis. By combining these diverse sources of information, multimodal models could predict more complex outcomes, such as whether a patient would respond to a specific treatment, thereby enhancing the precision of personalized medicine. An example is the DeepMind AlphaFold system, which combined protein structure data and genomics to predict protein folding, advancing drug discovery for cancer treatment. Despite its potential, multimodal models face challenges in harmonizing disparate data formats and ensuring interoperability between healthcare systems.

• 2023: The development of foundation models introduced large scale, self-supervised models pre-trained on diverse datasets. These models are designed to be fine-tuned for specific tasks with minimal additional data, making them highly adaptable. Foundation models serve as the backbone for more specialized downstream models, enabling AI systems to perform a wide range of tasks in oncology with greater efficiency and accuracy. An example is IBM Watson for Oncology, which uses foundation model principles to process vast medical literature and provide oncologists with evidence-based treatment recommendations. However, these systems are often criticized for lack of transparency in their decision-making processes, which can hinder clinical trust and adoption.

• Envisioned Future: Looking ahead, the envisioned future of AI in cancer research lies in the creation of generalist models. These all-encompassing models would integrate and interpret a wide range of data types, including genomic data, pathology slides, and radiology images, to support diagnosis, treatment recommendations, and other medical decisions. Generalist models aim to be multipurpose tools capable of analyzing complex cases, interacting with both patients and healthcare professionals, and providing comprehensive, interpretable insights across various medical domains. The success of these models will depend on overcoming current limitations, such as the lack of diverse training datasets and the need for stringent regulatory frameworks to ensure ethical and safe use.

Details of the Model Evolution

Machine Learning with Hand-Crafted Features (Early 2000s):

These models relied on manually selected features from medical images, often based on domain expertise. Pathologists would identify specific patterns or markers in histopathology slides to differentiate between cancerous and non-cancerous tissues. While effective, this approach was limited by the need for extensive human input and the inability to generalize beyond the features explicitly selected by experts.

Single-Domain, Single-Purpose Models:

As AI progressed, models were developed to solve specific tasks within a single domain. These models, typically based on supervised learning, could, for example, accurately distinguish between malignant and benign cells in a given dataset. However, their scope was limited to the tasks they were specifically trained on, restricting their applicability to broader medical questions.

Multimodal Models (2020):

The integration of multiple data types in AI models marked a significant breakthrough. Multimodal models leverage information from different sources–such as radiology images, genomic data, and clinical records–to make more informed predictions. For example, by combining genomic data with histopathology images, these models can better predict patient outcomes and guide personalized treatment strategies.

Foundation Models (2023):

These large-scale, self-supervised models are pre-trained on vast, diverse datasets and can be adapted to various tasks with minimal additional training. Foundation models form the basis for specific downstream applications, allowing researchers to fine-tune them for particular tasks such as predicting drug response or identifying cancer subtypes with high accuracy and efficiency.

Generalist Models (Future):

The ultimate goal in AI development for cancer research is the creation of generalist models. These models would be capable of integrating a wide array of data sources and providing comprehensive diagnostic and treatment recommendations. By combining the strengths of various specialized models, generalist models could offer a holistic approach to patient care, addressing multiple aspects of diagnosis and treatment in a unified framework.

AI Workflows in Cancer Research

The integration of AI technologies into cancer research has had a transformative impact, particularly in diagnostics and personalized medicine. AI algorithms, such as those utilized in CNNs, can now analyze complex genomic data with high throughput, accelerating the discovery of new biomarkers¹⁶. This process involves identifying key genomic features that distinguish responders from non-responders to certain therapies. The AI models then use these features to predict treatment outcomes, guiding clinical decisions and ensuring that each patient receives the most effective therapy.In addition to its role in personalized medicine, AI has revolutionized cancer diagnostics. CNNs trained on large datasets of labeled images can identify patterns indicative of cancerous cells with high precision¹⁷. By automating the analysis of these images, AI not only reduces the workload on pathologists but also minimizes human error, allowing pathologists to focus on more complex cases that require their expertise¹⁸.

These dual roles of AI–in both diagnostics and personalized medicine–demonstrate its transformative impact on cancer research¹⁹. The integration of AI technologies has led to more accurate cancer detection and more effective treatment planning, tailored to the genetic profiles of individual patients²⁰. This dual approach not only improves patient outcomes but also reduces the adverse effects of treatments by ensuring that therapies are precisely matched to each patient’s tumor characteristics²¹.

However, while the potential of AI in cancer research is immense, it is also important to acknowledge its limitations²². AI models require large, well-labeled datasets to achieve high accuracy, and there is always a risk of bias in the algorithms, which could affect their predictions²³. Future research should focus on addressing these challenges, such as developing methods to improve the robustness and fairness of AI models and exploring ways to integrate AI more effectively into clinical practice²⁴.

Discussion

This study leverages Convolutional Neural Networks (CNNs) and other advanced AI algorithms to analyze both histopathological images and genomic data in the context of cancer research²⁵. The CNN models were meticulously trained on extensive datasets, consisting of thousands of labeled histopathological images. These images underwent a series of preprocessing steps, including normalization, augmentation, and segmentation, to enhance the models’ ability to accurately identify and classify cancerous cells versus healthy tissue²⁶.

The CNN architecture was specifically designed with multiple convolutional layers, pooling layers, and fully connected layers, which enabled the model to capture intricate patterns and features indicative of malignancy²⁷. In the CNN architecture, convolutional layers are responsible for scanning the input images and detecting features like edges and textures²⁸. Pooling layers, which follow the convolutional layers, are used to reduce the spatial dimensions of the feature maps. This reduction helps to decrease the computational load and control overfitting by retaining only the most essential information²⁹. Finally, the fully connected layers integrate the extracted features and make the final prediction, classifying the image as containing cancerous or healthy tissue.

The genomic data processing pipeline involved several preprocessing steps, including sequence alignment using BWA, variant calling with GATK, and annotation through ANNOVAR,  followed by feature extraction and selection to enhance model performance³⁰. Recursive Feature Elimination (RFE) was employed to identify the most informative features by iteratively removing less significant features based on model weights³¹. Variant calling, a crucial step in the analysis, involves the identification and cataloging of genetic variations, such as mutations, within the DNA sequence³². These variations can then be analyzed to determine their significance in the development or progression of cancer, aiding in the identification of potential therapeutic targets³³.

The performance of these models was evaluated using metrics such as accuracy, precision, recall, and the Area Under the Curve of the Receiver Operating Characteristic (AUC-ROC)³⁴. The AUC-ROC is a critical metric for evaluating the performance of classification models, as it illustrates the trade-off between the true positive rate (sensitivity) and the false positive rate (1 minus specificity) at various threshold settings³⁵. The AUC, or Area Under the Curve, represents the likelihood that the model will correctly distinguish between positive and negative classes. An AUC value of 1.0 indicates a perfect model, while a value of 0.5 suggests a model with no discriminative power³⁶. Visual representations, such as ROC curves, are recommended to better illustrate this concept. The CNN models achieved high accuracy in identifying cancerous cells, with precision and recall values exceeding 90%³⁷. The genomic analysis algorithms also demonstrated strong predictive power in identifying actionable biomarkers³⁸. Cross-validation techniques were employed to ensure the robustness and generalizability of the models³⁹.

These findings align with the broader body of literature, which emphasizes the potential of AI to revolutionize cancer research²⁴. By enhancing diagnostic accuracy and accelerating the discovery of new biomarkers, AI is significantly advancing the development of personalized treatment strategies. However, this study also acknowledges the challenges associated with AI, particularly regarding the large volumes of high-quality data required for training and the potential for algorithmic bias, which may impact the generalizability of the models across diverse populations⁴⁰. Addressing these challenges is crucial for the continued advancement of AI-driven cancer research and its translation into clinical practice⁴¹.

Conclusion

This study highlights the transformative impact of Artificial Intelligence (AI), particularly Artificial Neural Networks (ANNs) and Convolutional Neural Networks (CNNs), in cancer research⁴². By significantly improving the accuracy of diagnostics and enabling the development of personalized treatment plans, AI is paving the way for more effective and targeted cancer therapies. For example, FDA-approved tools like PathAI assist pathologists by automating the analysis of biopsy samples, while Viz.ai integrates AI into stroke detection and is expanding into cancer imaging applications⁴³. In Europe, ScreenPoint’s Transpara has been CE-marked and is used to enhance breast cancer screening accuracy. These products are increasingly being adopted by health insurance systems as auxiliary tools to improve clinical decision-making and streamline workflows in oncology⁴⁴.

Collaboration in AI-Driven Cancer Research

Interdisciplinary collaboration is essential for the successful development and implementation of AI in cancer research. Teams comprising computer scientists, biologists, medical experts, and statisticians bring together diverse expertise to address the multifaceted challenges of cancer diagnostics and treatment. For instance, biologists contribute insights into molecular mechanisms, while computer scientists optimize algorithms to process vast genomic and imaging datasets. Shared resources like The Cancer Genome Atlas (TCGA) facilitate the integration of data, enabling cross-disciplinary teams to work more effectively⁴⁵. Organizations such as the Alliance for AI in Healthcare (AAIH) promote open data sharing and best practices, fostering an environment of transparency and innovation⁴⁶. These collaborations accelerate the pace of discovery and enhance the translation of AI research into clinical applications, benefiting patients through improved diagnostic precision and personalized treatments.

Regulation of AI-Related Medical Products

AI-related medical products must adhere to stringent regulatory frameworks to ensure patient safety and data security. In the United States, the FDA evaluates AI tools for efficacy and safety through pathways such as 510(k) clearance or De Novo requests. In Europe, CE marking certifies compliance with the Medical Device Regulation (MDR), ensuring safety and performance standards are met. Regulations also prioritize patient privacy, requiring compliance with laws like HIPAA in the United States and GDPR (General Data Protection Regulation) in Europe to protect sensitive health information. Rigorous validation through clinical trials and continuous post-market surveillance are critical to maintaining safety and trust in AI-driven medical applications⁴⁷.

The technical advancements discussed, such as the intricate design of CNN architectures for image analysis and the integration of genomic data through sophisticated algorithms, highlight the potential for AI to push the boundaries of precision medicine.

As AI technologies continue to evolve, their applications in cancer research are likely to expand, offering new opportunities for improving patient care and outcomes. Future work should focus on refining these models to address current limitations, including the need for diverse and representative datasets and the mitigation of algorithmic bias, to ensure that AI-driven solutions are equitable and widely applicable in clinical settings.

References

1. Silva, Helbert Eustáquio Cardoso da et al. “The use of artificial intelligence tools in cancer detection compared to the traditional diagnostic imaging methods: An overview of the systematic reviews.” PloS one vol. 18,10 e0292063. 5 Oct. 2023, doi:10.1371/journal.pone.0292063 [↩]
2. “Layers in Artificial Neural Networks (ANN).” GeeksforGeeks, GeeksforGeeks, 19 July 2024, www.geeksforgeeks.org/layers-in-artificial-neural-networks-ann/. [↩]
3. R. Perez-Lopez, N. G. Laleh, F. Mahmood, J. N. Kather, “A guide to artificial intelligence for cancer researchers,” Nature Reviews Cancer, 427-441 (2024). [↩]
4. Priya C V, Lakshmi et al. “Deep learning approaches for breast cancer detection in histopathology images: A review.” Cancer biomarkers : section A of Disease markers vol. 40,1 (2024): 1-25. doi:10.3233/CBM-230251 [↩]
5. Yamashita, R., Nishio, M., Do, R.K.G. et al. Convolutional neural networks: an overview and application in radiology. Insights Imaging 9, 611–629 (2018). https://doi.org/10.1007/s13244-018-0639-9 [↩]
6. Johnson, Kevin B et al. “Precision Medicine, AI, and the Future of Personalized Health Care.” Clinical and translational science vol. 14,1 (2021): 86-93. doi:10.1111/cts.12884 [↩]
7. Thakur, Gopal Kumar et al. “Deep Learning Approaches for Medical Image Analysis and Diagnosis.” Cureus vol. 16,5 e59507. 2 May. 2024, doi:10.7759/cureus.59507 [↩]
8. Abdelsamea, Mohammed M., et al. “A survey on artificial intelligence in histopathology image analysis.” Wiley Interdisciplinary Reviews: Data Mining and Knowledge Discovery 12.6 (2022): e1474. [↩]
9. Van der Laak, Jeroen, Geert Litjens, and Francesco Ciompi. “Deep learning in histopathology: the path to the clinic.” Nature medicine 27.5 (2021): 775-784. [↩]
10. Gandhi, Zainab, et al. “Artificial intelligence and lung cancer: impact on improving patient outcomes.” Cancers 15.21 (2023): 5236. [↩]
11. Prieto, Alberto, et al. “Neural networks: An overview of early research, current frameworks and new challenges.” Neurocomputing 214 (2016): 242-268. [↩]
12. Ahad, Nauman, Junaid Qadir, and Nasir Ahsan. “Neural networks in wireless networks: Techniques, applications and guidelines.” Journal of network and computer applications 68 (2016): 1-27. [↩]
13. Agatonovic-Kustrin, S., and Rosemary Beresford. “Basic concepts of artificial neural network (ANN) modeling and its application in pharmaceutical research.” Journal of pharmaceutical and biomedical analysis 22.5 (2000): 717-727. [↩]
14. Iqbal, Muhammad Javed, et al. “Clinical applications of artificial intelligence and machine learning in cancer diagnosis: looking into the future.” Cancer cell international 21.1 (2021): 270. [↩]
15. Azhagiri, M., Rajesh, P. EAN: enhanced AlexNet deep learning model to detect brain tumor using magnetic resonance images. Multimed Tools Appl 83, 66925–66941 (2024). [↩]
16. Mann, Matthias, et al. “Artificial intelligence for proteomics and biomarker discovery.” Cell systems 12.8 (2021): 759-770. [↩]
17. Ahn, Jong Seok, et al. “Artificial intelligence in breast cancer diagnosis and personalized medicine.” Journal of Breast Cancer 26.5 (2023): 405. [↩]
18. Nakagawa, Keisuke, et al. “AI in pathology: what could possibly go wrong?.” Seminars in Diagnostic Pathology. Vol. 40. No. 2. WB Saunders, 2023. [↩]
19. Sherani, Abdul Mannan Khan, et al. “Synergizing AI and Healthcare: Pioneering Advances in Cancer Medicine for Personalized Treatment.” International Journal of Multidisciplinary Sciences and Arts 3.1 (2024): 270-277. [↩]
20. Liao, Jinzhuang, et al. “Artificial intelligence assists precision medicine in cancer treatment.” Frontiers in oncology 12 (2023): 998222. [↩]
21. Sicklick, Jason K., et al. “Molecular profiling of cancer patients enables personalized combination therapy: the I-PREDICT study.” Nature medicine 25.5 (2019): 744-750. [↩]
22. Sharpless, Norman E., and Anthony R. Kerlavage. “The potential of AI in cancer care and research.” Biochimica et Biophysica Acta (BBA)-Reviews on Cancer 1876.1 (2021): 188573. [↩]
23. Zhu, Ming, et al. “Artificial intelligence in the radiomic analysis of glioblastomas: A review, taxonomy, and perspective.” Frontiers in Oncology 12 (2022): 924245. [↩]
24. Kelly, Christopher J., et al. “Key challenges for delivering clinical impact with artificial intelligence.” BMC medicine 17 (2019): 1-9. [↩] [↩]
25. Khosravi, Pegah, et al. “Deep convolutional neural networks enable discrimination of heterogeneous digital pathology images.” EBioMedicine 27 (2018): 317-328. [↩]
26. Ahmed, Luqman, et al. “Images data practices for semantic segmentation of breast cancer using deep neural network.” Journal of Ambient Intelligence and Humanized Computing 14.11 (2023): 15227-15243. [↩]
27. Alakwaa, Wafaa, Mohammad Nassef, and Amr Badr. “Lung cancer detection and classification with 3D convolutional neural network (3D-CNN).” International Journal of Advanced Computer Science and Applications 8.8 (2017). [↩]
28. Chaiyasarn, Krisada, et al. “Integrated pixel-level CNN-FCN crack detection via photogrammetric 3D texture mapping of concrete structures.” Automation in Construction 140 (2022): 104388. [↩]
29. Gholamalinezhad, Hossein, and Hossein Khosravi. “Pooling methods in deep neural networks, a review.” arXiv preprint arXiv:2009.07485 (2020). [↩]
30. Bao, Riyue, et al. “Review of current methods, applications, and data management for the bioinformatics analysis of whole exome sequencing.” Cancer informatics 13 (2014): CIN-S13779. [↩]
31. X. -w. Chen and J. C. Jeong, “Enhanced recursive feature elimination,” Sixth International Conference on Machine Learning and Applications (ICMLA 2007), Cincinnati, OH, USA, 2007, pp. 429-435, doi: 10.1109/ICMLA.2007.35. keywords: {Machine learning;Support vector machines;Support vector machine classification;Filters;Cancer;Application software;Degradation;Target recognition;Hyperspectral sensors;Training data} [↩]
32. Pabinger, Stephan, et al. “A survey of tools for variant analysis of next-generation genome sequencing data.” Briefings in bioinformatics 15.2 (2014): 256-278. [↩]
33. Lapenna, Silvia, and Antonio Giordano. “Cell cycle kinases as therapeutic targets for cancer.” Nature reviews Drug discovery 8.7 (2009): 547-566. [↩]
34. Khan, Shahzad Ali, and Zeeshan Ali Rana. “Evaluating performance of software defect prediction models using area under precision-Recall curve (AUC-PR).” 2019 2nd International Conference on Advancements in Computational Sciences (ICACS). IEEE, 2019. [↩]
35. Sofaer, Helen R., Jennifer A. Hoeting, and Catherine S. Jarnevich. “The area under the precision‐recall curve as a performance metric for rare binary events.” Methods in Ecology and Evolution 10.4 (2019): 565-577. [↩]
36. Jiménez-Valverde, Alberto. “Threshold-dependence as a desirable attribute for discrimination assessment: implications for the evaluation of species distribution models.” Biodiversity and Conservation 23 (2014): 369-385. [↩]
37. Quazi, Sameer. “Artificial intelligence and machine learning in precision and genomic medicine.” Medical Oncology 39.8 (2022): 120. [↩]
38. Roberts, David R., et al. “Cross‐validation strategies for data with temporal, spatial, hierarchical, or phylogenetic structure.” Ecography 40.8 (2017): 913-929. [↩]
39. Frasca, Maria, et al. “Artificial intelligence applications to genomic data in cancer research: a review of recent trends and emerging areas.” Discover Analytics 2.1 (2024): 10. [↩]
40. Elemento, Olivier, et al. “Artificial intelligence in cancer research, diagnosis and therapy.” Nature Reviews Cancer 21.12 (2021): 747-752. [↩]
41. Kourounis, Georgios, et al. “Computer image analysis with artificial intelligence: a practical introduction to convolutional neural networks for medical professionals.” Postgraduate medical journal 99.1178 (2023): 1287-1294. [↩]
42. Ali, Umar, Sameer Ali, and Muhammad Tabraiz Ali. “Synergies of AI and smart technology: Transforming cancer medicine, vaccine development, and patient care.” International Journal of Multidisciplinary Research and Growth Evaluation (2024). [↩]
43. Kurian M, Adashek JJ, West H. Cancer Care in the Era of Artificial Intelligence. JAMA Oncol. 2024;10(5):683. doi:10.1001/jamaoncol.2023.7263 [↩]
44. Singh, G., Kamalja, A., Patil, R. et al. A comprehensive assessment of artificial intelligence applications for cancer diagnosis. Artif Intell Rev 57, 179 (2024). [↩]
45. Wizemann, Theresa, and Sharyl J. Nass, eds. “Informatics needs and challenges in cancer research: workshop summary.” (2012). [↩]
46. Ullagaddi, Pravin. “Digital transformation in the pharmaceutical industry: Enhancing quality management systems and regulatory compliance.” International Journal of Health Sciences 12.1 (2024): 31-43. [↩]
47. Daria Onitiu, Sandra Wachter, Brent Mittelstadt, How AI challenges the medical device regulation: patient safety, benefits, and intended uses, Journal of Law and the Biosciences, 2024; Murdoch, B. Privacy and artificial intelligence: challenges for protecting health information in a new era. BMC Med Ethics 22, 122 (2021). [↩]