Background & Aims
Pain due to cancer is one of the biggest challenges in patient care,
so it requires innovative approaches in its management. Research has showed that artificial
intelligence (AI) enhances prediction and decision-making in pain management (1-3). However,
several studies are not specific to cancer pain, indicating the need for more specialized models and
better integration for a holistic approach to cancer care.
Methods
This systematic review followed the Preferred Reporting Items for Systematic Reviews
and Meta-Analyses (PRISMA) guideline and was registered in PROSPERO (CRD42023429925)
on June 6, 2023. We searched articles published in 2013-2023 from seven databases (PubMed,
Medline EBSCO, Scopus, IEEE Xplore, ProQuest, ScienceDirect, and Cochrane). The article
selections used the PICOD framework: Population (P): cancer patients; Intervention (I): Artificial
Intelligence; Comparator (C): traditional or standard cancer pain methods; Outcome (O): Its
primary outcome was to determine how AI has been used in cancer pain; Design (D): all designs
except reviews, commentaries, and editorials. We used MI-CLAIM, STROBE, and CONSORTAI
checklists to assess the risks of bias and study quality. We used Rayyan Systematic Review to
organize the data. In this study, data synthesis was conducted using the Synthesis Without Metaanalysis
(SWiM) approach.
Results
Out of 943 abstracts, 14 articles were included, and most studies have good quality.
Trends showed that 50% of articles were published in 2018-2019. The study designs are
heterogeneous, e.g. cohort (14.2%), cross-sectional/observational/longitudinal (14.2%), and
secondary analyses (7.1%); therefore, we could not conduct a meta-analysis. The characteristic
studies mostly focused on lung, breast, pancreatic, and colorectal cancer. All studies utilized
Machine Learning (ML) techniques with varying accuracies: Measure of Performance (72-73%),
Random Forest, Support Vector Machine (71.7%), and Artificial Neural Networks (94%). The
challenges in ML use, including variable model performance, overfitting risks, and limitations in
NLP documentation, emphasize the need for diverse data and model customization. We mapped
the findings into the Cancer Care Continuum (CoCC) framework from diagnosis, treatment, and
survivorship to end-of-life care. AI has been implemented in two areas: pain assessment (pain
detection, classification, prediction) and pain management (prediction and classification of drug
response or outcome and teleconsultation).
Conclusions
This review revealed that ML is predominantly used in cancer-related pain (8.9). Access to
advanced technology and expertise in oncology, as well as integrated health infrastructure and
extensive electronic databases, supports this research innovation (10). Several challenges with the
use of ML require rigorous testing for calibration and external clinical validation before
implementation in healthcare settings (9,10). We found that AI is widely used in all stages of CoCC
cancer care, from diagnosis to end of life, particularly to improve the accuracy of pain assessment
and management (2). The holistic CoCC model is proven to improve quality of life and patient
satisfaction, reduce readmission rates, and increase cost-effectiveness.
AI has the potential to enhance cancer pain management. Existing studies showed heterogeneity
in study designs and participants’ characteristics. Challenges include using cross-validation with
large datasets and techniques to reduce overfitting. Future research should focus on developing
stage-specific, sensitive, and accurate algorithms across the Cancer Care Continuum,
accommodating complexities in treatment, integrating broader data in survivorship, and
responding sensitively to changing needs in end-of-life care.
References
Shao D, Dai Y, Li N, et al. Artificial intelligence in clinical research of cancers. Brief
Bioinform. 2022;23(1). doi:10.1093/BIB/BBAB523
2. O’Connor S, Vercell A, Wong D, et al. The application and use of artificial intelligence in
cancer nursing: A systematic review. European Journal of Oncology Nursing.
2024;68:102510. doi:10.1016/J.EJON.2024.102510
3. Hagedorn JM, George TK, Aiyer R, Schmidt K, Halamka J, D’souza RS. Artificial
Intelligence and Pain Medicine: An Introduction. Published online 2024.
doi:10.2147/JPR.S42959.
4. Page MJ, McKenzie JE, Bossuyt PM, et al. The PRISMA 2020 statement: An updated
guideline for reporting systematic reviews. The BMJ. 2021;372. doi:10.1136/BMJ.N71
5. Ferrell BR, Smith TJ, Levit L, Balogh E. Improving the quality of cancer care: implications
for palliative care. J Palliat Med. 2014;17(4):393-399. doi:10.1089/JPM.2013.053
6. Levit L, Balogh E, Nass S. Introduction – Delivering High-Quality Cancer Care – NCBI
Bookshelf. National Academies Press (US); 2013. Accessed May 11, 2024.
7. Tong EK, Wolf T, Cooke DT, Fairman N, Chen MS. The Emergence of a Sustainable
Tobacco Treatment Program across the Cancer Care Continuum: A Systems Approach for
Implementation at the University of California Davis Comprehensive Cancer Center. Int J
Environ Res Public Health. 2020;17(9). doi:10.3390/IJERPH17093241
8. Matsangidou, M., Liampas, A., Pittara, M., Pattichi, C. S., & Zis, P. (2021). Machine
learning in pain medicine: an up-to-date systematic review. Pain and therapy, 1-18.
9. Kamdar, M., Centi, A. J., Agboola, S., Fischer, N., Rinaldi, S., Strand, J. J., … & Jethwani,
K. (2019). A randomized controlled trial of a novel artificial intelligence-based smartphone
application to optimize the management of cancer-related pain
10. Wu, T., Duan, Y., Zhang, T., Tian, W., Liu, H., & Deng, Y. (2022). Research trends in the
application of artificial intelligence in oncology: A bibliometric and network visualization
study. Frontiers in Bioscience-Landmark, 27(9), 254.
Presenting Author
Yayu Nidaul Fithriyyah
Poster Authors
Topics
- Specific Pain Conditions/Pain in Specific Populations: Cancer Pain & Palliative Care