Page 17 - AIH-1-1

Artificial Intelligence in Health AI in prostate cancer detection

ANN. Fifteen publications used ANN in various aspects of has delved into the integration of AI/ML tools, combining
prostate cancer research, encompassing detection, grading, histopathology and MRI signals. It is our belief that future
prediction, and classification. SVMs, a supervised learning research will leverage a combination of demographic
algorithm commonly implemented in classification or features, clinical data, serological markers, pathological
regression problems, stands out for its enhanced accuracy grading, radiological factors, and genomic data, such as
with smaller datasets compared to other algorithms. cell-free DNA. This combination aims to propose accurate,
Researchers commonly use SVM, as evidenced by its non-invasive diagnoses of clinically significant prostate
presence in approximately 20% of the papers focused on cancer using AI/ML tools. This advancement has the
the classification of prostate cancer. Other techniques, potential to obviate the need for routine serum PSA testing
including random forest, regression, and fuzzy clustering, in early detection, thereby mitigating the subjectivity
have also found application in prostate cancer research. associated with the interpretation of clinically significant
Similarly, Table 6 delineates the specific application areas tumors.
in which these ML tools are utilized.
Acknowledgments
8. Conclusions and future research
None.
In this review, we conducted searches on three research
databases: SCOPUS, WoS, and Google Scholar for data Funding
extraction. While some other studies also employed None.
databases such as PubMed, ProQuest, and others, we opted
for SCOPUS due to its extensive range of articles, complete Conflict of interest
reference lists, and consistent and reliable formatting . The authors declare that they have no competing interests.
[80]
Similarly, WoS and Google Scholar were chosen for their
comprehensive coverage of published articles related Author contributions
to medical and data science. In addition, the keyword
analysis relied on the keywords provided by the authors. Conceptualization: Syed A. Raza, Nadeem Pervez, Ikram A.
It is plausible that certain keywords, such as “low- and Burney
high-grade tumors,” “low- and high-risk tumors,” and Formal Analysis: Syed A. Raza, Momena Ahmed
“localized and metastatic diseases,” may not have been Methodology: Syed A. Raza
provided by the authors. Finally, content analysis focused Supervision: Syed A. Raza
on the top 50 cited articles published in the last 3 years Writing – original draft: All authors
only (2019–2021). Among the 20 most-cited articles, six Writing – review & editing: Syed A. Raza, Momena Ahmed
were published in the 3-year period, and another seven
were published in the preceding 2 years. Our scope was to Ethics approval and consent to participate
develop a deeper understanding of the current direction of Not applicable.
research, leading us to concentrate our content analysis on
the most recently published articles. Consent for publication
In conclusion, diagnostic accuracy poses a significant Not applicable.
challenge in the context of prostate cancer. Early
diagnostics is also vital in timely addressing the cancer, Availability of data
thereby contributing to an improved survival rate. Data used in this work is available from the corresponding
The utilization of AI/ML tools is increasingly gaining author upon reasonable request.
prominence as a means to improve the diagnostic accuracy
of prostate cancer. In addition, the advancement in References
imaging technologies has greatly contributed to both the 1. Sung H, Ferlay J, Siegel RL, et al., 2020, Global cancer
accuracy and early diagnostics of cancers. The systematic statistics: GLOBOCAN estimates of incidence and mortality
literature review, conducted using BA tools, focusing on worldwide for 36 cancers in 185 countries. CA Cancer J Clin,
the application of AI/ML in prostate cancer diagnosis and 71: 209–249.
treatment, elucidates an exponential growth in the number https://doi.org/0.3322/caac.21660
of published papers over the past 3 years. The two most
consistent themes identified are predictive modeling and 2. Siegel RL, Miller KD, Fuchs HE, et al., 2021, Cancer statistics.
the application areas of the AI/ML tools, especially in CA Cancer J Clin, 71: 7–33.
cancer grading and radiomics. Notably, a subset of papers https://doi.org/10.3322/caac.21654

Volume 1 Issue 1 (2024) 11 https://doi.org/10.36922/aih.1958