Abstract Recent advances in computational science mean that the capacity of artificial intelligence (AI) to predict disease risk, diagnose illness, and guide therapeutic decision making is now a realistic prospect in the foreseeable future. The use of AI has already shown promise in a wide variety of fields in medicine including cardiology, renal medicine, critical care, and mental health. For a variety of reasons, cancer is an area where AI is expected to have a significant impact in the near term, and cancer is in many ways an obvious choice for enhancing the “reputation” of AI-based approaches in healthcare. First, cancer is common and imposes a considerable burden in terms of physical disability, emotional trauma, and economic cost for patients and society as a whole. Thus, almost without exception, across all cancer subtypes, there is a critical need to improve cancer-related outcomes. Second, at the individual level, there is huge patient demand for enhanced cancer diagnostics, prognostication, and therapy; cancers represent a major and indiscriminate cause of morbidity and mortality, and they evoke fear in a way that few other conditions do, because of the perception of cancer as a fatal diagnosis. From the cancer clinician’s perspective, there is an equally urgent drive since tumor heterogeneity and the exponential growth in onco-therapeutic options present major challenges in terms of decision making, not least due to sheer data burden. Rather alarmingly, it has been estimated that the modern-day cancer physician would need to spend in excess of 20 hours per day reading, in order to stay up to date with developments in the scientific literature. This challenge is well exemplified by current attempts to unravel the cancer genome—millions of molecular disparities are now coming to light that could, alone or in combination, influence cancer cell survival and progression, but harnessing these discoveries in a meaningful way is impossible without state-of-the-art computing. The impact of AI approaches is also expected to be significant for academia, where cancer research initiatives are likely to be more precisely steered by AI-based data-mining discoveries. This in turn is likely to mean that certain avenues of research in terms of cancer diagnostics and therapy are “shelved” either permanently or temporarily until new developments emerge that suggest otherwise. Furthermore, it is anticipated that for the pharmaceuticals industry, anticancer drug development and the development of companion diagnostics will benefit from an additional layer of premarket delivery validation, through AI-based analyses of actual and simulated clinical outcomes. One can thus appreciate that all stakeholders within the cancer health space stand to make potentially significant gains with the application of AI solutions.
Paper
Full text
Cancer diagnostics and treatment decisions using artificial intelligence
Semantic Scholar · Medicine · 2020
Abstract
Abstract Recent advances in computational science mean that the capacity of artificial intelligence (AI) to predict disease risk, diagnose illness, and guide therapeutic decision making is now a realistic prospect in the foreseeable future. The use of AI has already shown promise in a wide variety of fields in medicine including cardiology, renal medicine, critical care, and mental health. For a variety of reasons, cancer is an area where AI is expected to have a significant impact in the near term, and cancer is in many ways an obvious choice for enhancing the “reputation” of AI-based approaches in healthcare. First, cancer is common and imposes a considerable burden in terms of physical disability, emotional trauma, and economic cost for patients and society as a whole. Thus, almost without exception, across all cancer subtypes, there is a critical need to improve cancer-related outcomes. Second, at the individual level, there is huge patient demand for enhanced cancer diagnostics, prognostication, and therapy; cancers represent a major and indiscriminate cause of morbidity and mortality, and they evoke fear in a way that few other conditions do, because of the perception of cancer as a fatal diagnosis. From the cancer clinician’s perspective, there is an equally urgent drive since tumor heterogeneity and the exponential growth in onco-therapeutic options present major challenges in terms of decision making, not least due to sheer data burden. Rather alarmingly, it has been estimated that the modern-day cancer physician would need to spend in excess of 20 hours per day reading, in order to stay up to date with developments in the scientific literature. This challenge is well exemplified by current attempts to unravel the cancer genome—millions of molecular disparities are now coming to light that could, alone or in combination, influence cancer cell survival and progression, but harnessing these discoveries in a meaningful way is impossible without state-of-the-art computing. The impact of AI approaches is also expected to be significant for academia, where cancer research initiatives are likely to be more precisely steered by AI-based data-mining discoveries. This in turn is likely to mean that certain avenues of research in terms of cancer diagnostics and therapy are “shelved” either permanently or temporarily until new developments emerge that suggest otherwise. Furthermore, it is anticipated that for the pharmaceuticals industry, anticancer drug development and the development of companion diagnostics will benefit from an additional layer of premarket delivery validation, through AI-based analyses of actual and simulated clinical outcomes. One can thus appreciate that all stakeholders within the cancer health space stand to make potentially significant gains with the application of AI solutions.