It’s Time for Trainee and Early-Career Oncologists to Welcome the Artificial Intelligence Era in Oncology

Jan 14, 2022

  • Dr. Erick Saldanha

By Erick Saldanha, MD

In 2020, artificial intelligence (AI) and its potential application in cancer diagnosis and monitoring were listed as one of the significant milestones in cancer as part of the Milestone Collection project by Nature.1 Other milestones in cancer within this project included liquid biopsies for noninvasive diagnosis and monitoring of patients, the first US Food and Drug Administration approval for an anti-PD-L1 inhibitor, inhibition of KRAS G12C, and a clinical trial of CAR-T cells to target BCMA in a patient with multiple myeloma, among other recent discoveries that have changed the understanding and current landscape of cancer. My guess is that of those recent developments, the use of AI in oncology is the one that trainees and early-career oncologists know the least about. In this article, I will walk you through essential definitions and the main reasons AI must be part of the medical student curriculum moving forward. 

What Is AI?

What is the definition of AI? First, I would like to call your attention to the fact that the first definitions of AI date back to the 1940s and ‘50s, when early computer scientists, such as Alan Turing and John McCarthy, first proposed formal definitions of machine intelligence.2 
 
AI can be broadly defined as a field of computer science capable of copying human characteristics, the capacity of learning, and the storage of knowledge.3 This form of “cognitive computing” executes human brain tasks in most fields in all aspects of our daily lives using big data applications.3,4 Machine learning (ML) refers to a subfield of AI in which mathematical and statistical approaches are applied to improve the performance of computers.5 ML is the fundamental technology required to meaningfully process data that exceed the human brain’s capacity to comprehend.6 

What Can AI Do?

The myriad of potential applications of AI and its branch ML in the oncology field is tremendous: enhancing the quality of and decreasing the time to diagnosis; potential frameworks to improve prognosis and toxicity prediction; optimizing cancer drug discovery, development, and administration; and reducing disparities by enhancing access to clinical trials. To date, the most concrete applications of AI in cancer are those focusing on using imaging to diagnose malignancies.7 As an example, a groundbreaking study published in 2017 by Esteva et al. used 129,450 clinical images of skin disease to train a deep convolutional neural network to classify skin lesions. The accuracy of the system in detecting malignant melanomas and carcinomas matched that of trained dermatologists.8 Using a multicenter data set of 28,953 mammograms, McKinney et al. developed a model that predicts the 2-year development of breast cancer with superior performance characteristics compared with trained radiologists using the Breast Imaging-Reporting and Data System criteria.9 
 
An exciting application of AI in oncology is the potential to make clinical trials more inclusive. Data-driven algorithms combined with real-world data can improve different aspects of clinical trials, including accessibility. Liu et al. evaluated the effect of varying eligibility criteria on cancer trial populations and outcomes with real-world data using the computational framework of Trial Pathfinder. Investigators used Trial Pathfinder to emulate completed trials of advanced non-small cell lung cancer (NSCLC) using data from a nationwide database of electronic health records comprising 61,094 patients with advanced NSCLC. When they used a data-driven approach to broaden restrictive criteria, the pool of eligible patients more than doubled on average. The hazard ratio of the overall survival decreased by an average of 0.05.10 This result suggests that many patients who were not eligible under the original trial criteria could benefit from the treatments. Another promising study conducted by Beck et al. utilized an AI-enabled system to match patients with breast cancer to clinical trials. The use of system assistance decreased the time required for research coordinators to assess trial eligibility and identified eligible patients with more than 90% sensitivity in three of four trials.11

What Are the Obstacles to Applying AI in Oncology Practice?

The hype is real; AI has already proved its potential to revolutionize cancer research. However, critical challenges are posed towards its implementation into clinical practice: 
  • Lack of prospective randomized validation data. We wait for the results of current ongoing prospective trials. 
  • Skepticism in the health care provider community. It is fundamental to provide mentors and ML education to physicians. 
  • Data bias. To mitigate misleading recommendations and inaccurate predictions, we must guarantee representative sampling of traditionally underrepresented populations (e.g., women and ethnic minorities) within the data set. 
  • Implementation and data curation. It is essential to make better use of electronic health records as a source of AI data and properly define standardizations to improve data curation. 
  • Regulatory and transparency issues. Implementation of regulatory frameworks and better transparency is paramount to guarantee patient safety and privacy. 
Efforts to educate physicians about AI and to integrate the use of technology in our practices are paramount. The skepticism is real, but we’re already using AI in our lives every day. Whenever we use GPS software, digital assistants like Amazon’s Alexa or Apple’s Siri, and social media platforms, we benefit from ML algorithms and their capacity to generate accurate predictions. The idea of applying AI and its branches in oncology research is to be empowered, not underpowered. 
 
Embracing the AI era is a necessity. Trainees and early-career oncologists are uniquely poised to help AI live up to its promise in oncology: we have the potential to critically appraise AI validation trials, ensure patient safety and privacy, improve accessibility to clinical trials, and reduce health disparities.

How Can I Learn More?

For those at the starting line on the road of AI application in oncology research, I strongly recommend regularly reading JCO Clinical Cancer Informatics to keep up with the latest research in the field. I also recommend the following four cutting-edge articles which will provide a solid foundation for your future knowledge. 

Dr. Saldanha is a medical oncology fellow at AC Camargo Cancer Center, São Paulo, Brazil. He believes that using different platforms to share knowledge and experiences can improve patient care; he is also delighted with the opportunity to serve as a 2021-2022 member of the ASCO Trainee and Early Career Advisory Group. Follow him on Twitter @SaldanhaMd.

References

  1. Foronda M. The AI Revolution in Cancer. Nature Milestones: Cancer. 2020 Dec 10. Available at: https://www.nature.com/articles/d42859-020-00082-9. Accessed Nov 8, 2021.
  2. Nagy M, Radakovich N, Nazha A. Machine Learning in Oncology: What Should Clinicians Know? JCO Clin Cancer Inform. 2020;4:799-810.
  3. Yang YC, Islam SU, Noor A, et al. Influential Usage of Big Data and Artificial Intelligence in Healthcare. Comput Math Methods Med. 2021;2021:5812499. 
  4. Kantarjian H, Yu PP. Artificial Intelligence, Big Data, and Cancer. JAMA Oncol. 2015;1:573-4.
  5. Shimizu H, Nakayama KI. Artificial intelligence in oncology. Cancer Sci. 2020;111:1452-60.
  6. Rajkomar A, Dean J, Kohane I. Machine Learning in Medicine. N Engl J Med. 2019;380:1347-58. 
  7. Elemento O, Leslie C, Lundin J, et al. Artificial intelligence in cancer research, diagnosis and therapy. Nat Rev Cancer. Epub 2021 Sep 17. 
  8. Esteva A, Kuprel B, Novoa RA, et al. Dermatologist-level classification of skin cancer with deep neural networks. Nature. 2017;542:115-8.
  9. McKinney SM, Sieniek M, Godbole V, et al. International evaluation of an AI system for breast cancer screening. Nature. 2020;577:89-94.
  10. Liu R, Rizzo S, Whipple S, et al. Evaluating eligibility criteria of oncology trials using real-world data and AI. Nature. 2021;592:629-33.
  11. Beck JT, Rammage M, Jackson GP, et al. Artificial Intelligence Tool for Optimizing Eligibility Screening for Clinical Trials in a Large Community Cancer Center. JCO Clin Cancer Inform. 2020;4:50-9.

Recent Articles

How I Became a Community Medical Oncologist and Hematologist
Navigating the Board Certification Process for Your Oncology Subspecialty
How I Became a Genitourinary Oncologist and Clinical Researcher
(Y)Our Voice Matters: Seven Lessons for Effective Health Care Advocacy
Say Yes to Health Care Advocacy: Making an Impact for Our Patients and Our Profession

Advertisement
Back to Top