Evaluating IBM Watson’s Role in Oncology Treatment Decisions: A Case Study on AI-Based Clinical Decision Support
DOI:
https://doi.org/10.64229/c869wh33Keywords:
IBM Watson for Oncology, Clinical Decision Support, Artificial Intelligence in Healthcare, Oncology Treatment Recommendations, Machine Learning, NLPAbstract
Cancer is still a leading world cause of death, being responsible for some ten million annual fatalities, as indicated by World Health Organization (WHO). The complex nature of oncology is a consequence of needing to inter-relate several sources of information, like genomic signatures, pathological reports, and clinical studies, to generate individualized therapy. Classical approaches to decision-making, based nearly solely on clinical expertise and labour-intensive scanning of published clinical work, are correspondingly threatened by increasing numbers of data on oncology and fluid nature of recommendations for treatment.
As a result of all these constraints, a cognitive information system, called IBM Watson for Oncology, was constructed for supporting clinical decisions, based on evidence, for oncologists. With Natural Language Processing (NLP) and Machine Learning (ML) techniques, Watson analyses both structured and unstructured clinical information for recommending treatment procedures, following accepted protocols, such as those of National Comprehensive Cancer Network (NCCN) and American Society of Clinical Oncology (ASCO).
This case study evaluates the architecture, modus operandi, and performance of IBM Watson for Oncology as a clinical decision-support system. The evaluation dwells on its capacity for delivering accurate, interpretable, and personal treatment recommendations and its capacity for concordance with decisions of oncologists for breast, lung, and colorectal cancer. Several studies report their results as having a high degree of concordance between Watson's recommendations and experts' judgment ranging between 70% and 96%. The results suggest the promise of AI-based systems like Watson for considerable improvement of consistency and efficiency of decision-making procedures across oncology. Concerns regarding data localization, interpretability, and constant learning are, however, critical areas requiring further improvement.
References
[1]World Health Organization (WHO), “Cancer Fact Sheet,” 2023.
[2]R. L. Siegel, K. D. Miller, and A. Jemal, “Cancer statistics, 2023,” CA: A Cancer Journal for Clinicians, vol. 73, pp. 17–48, 2023.
[3]Memorial Sloan Kettering Cancer Center (MSKCC), “Collaborating with IBM Watson for Oncology,” MSKCC Press Release, 2022.
[4]IBM Corporation, “IBM Watson for Oncology Overview,” IBM Technical Whitepaper, 2021.
[5]IBM Watson Health, “Watson for Oncology Architecture Overview,” IBM White Paper, 2021.
[6]S. P. Somashekhar, P. Sepúlveda, A. Puglielli, et al., “Watson for Oncology and breast cancer treatment recommendations: Concordance study in India,” Journal of Oncology Practice, vol. 14, no. 3, pp. e338–e345, 2018.
[7]A. Esteva, B. Kuprel, R. A. Novoa, J. Ko, S. M. Swetter, H. M. Blau, and S. Thrun, “Dermatologist-level classification of skin cancer with deep neural networks,” Nature, vol. 542, pp. 115–118, 2017.
[8]P. Rajkomar, J. Dean, and I. Kohane, “Machine learning in medicine,” New England Journal of Medicine, vol. 380, no. 14, pp. 1347–1358, 2019.
[9]E. H. Shortliffe, “AI and clinical decision support: The emergence of a new discipline,” Artificial Intelligence in Medicine, vol. 80, pp. 1–13, 2017.
[10]K. Kim, J. H. Kim, and Y. Lee, “Evaluation of IBM Watson for Oncology in lung cancer treatment decisions,” Cancer Research and Treatment, vol. 51, no. 2, pp. 631–639, 2019.
[11]IBM Research, “Watson Discovery Service Overview,” IBM Developer Portal, 2022.
[12]Merative (IBM Watson Health), “AI in Healthcare: Lessons from Watson Health,” 2023.
[13]C. D. Liang, Y. Xu, and P. Zhang, “Explainable artificial intelligence in healthcare: Challenges and opportunities,” IEEE Transactions on Medical Imaging, vol. 41, no. 10, pp. 2808–2823, 2022.
[14]M. McKinney et al., “International evaluation of an AI system for breast cancer screening,” Nature, vol. 577, pp. 89–94, 2020.
[15]M. L. Giger, “Machine learning in medical imaging,” Journal of the American College of Radiology, vol. 15, no. 3, pp. 512–520, 2018.
[16]H. Chen, R. Engkvist, Y. Wang, M. Olivecrona, and T. Blaschke, “The rise of deep learning in drug discovery,” Drug Discovery Today, vol. 23, no. 6, pp. 1241–1250, 2018.
[17]J. H. Topol, “High-performance medicine: The convergence of human and artificial intelligence,” Nature Medicine, vol. 25, pp. 44–56, 2019.
[18]National Comprehensive Cancer Network (NCCN), “Clinical Practice Guidelines in Oncology,” 2023.
[19]American Society of Clinical Oncology (ASCO), “Cancer Treatment Guidelines,” 2023.
[20]Z. Zhou, H. Yang, J. Xu, and L. Chen, “Assessment of Watson for Oncology in colorectal cancer care,” Journal of Cancer Research and Clinical Oncology, vol. 145, no. 12, pp. 3055–3064, 2019.
[21]Z. Zhou, H. Yang, J. Xu, and L. Chen, “Assessment of Watson for Oncology in colorectal cancer care,” Journal of Cancer Research and Clinical Oncology, vol. 145, no. 12, pp. 3055–3064, 2019.
[22]NVIDIA Corporation, “AI Infrastructure Solutions for Healthcare,” Technical Whitepaper, 2022.
[23]IBM Corporation, “IBM Power Systems for AI Workloads,” IBM Cloud Technical Documentation, 2022.
[24]NVIDIA Corporation, “NVIDIA A100 Tensor Core GPU: Performance for AI and HPC,” Technical Brief, 2023.
J. G. Dorsey, E. M. Topol, and L. A. Adams, “Telemedicine and AI: Shaping the future of remote healthcare,” Lancet Digital Health, vol. 2, pp. e60–e70, 2020.
Downloads
Published
Issue
Section
License
Copyright (c) 2025 Rishi Raj Singh, Ankit Garg (Author)
This work is licensed under a Creative Commons Attribution 4.0 International License.
