Behind the paper - “Artificial intelligence for precision oncology: beyond patient stratification”
The paper introduces AI and its applications in precision oncology. It discusses major advances and challenges beyond pattern recognition and classification tasks. It argues in favor of wider innovative uses of AI for bringing benefits to patients. In this post, the author talks about the motivation behind the paper, important topics not covered in it and the excitement of doing research in this field.
The full paper is freely available.
What motivated this paper?
I wanted to offer a brief introduction to AI and its applications in precision oncology. Apart from providing my perspective on advances and challenges in the field, I was motivated by the need to demystify AI in biomedical research, while at the same time showing that progress can be accomplished well beyond traditional pattern detection and classification. We need to move away from both hype and pessimism.
What was the biggest challenge in writing the paper?
First, there was the challenge of condensing fundamental concepts, advances and gaps in the area into a relatively short paper. Another challenge was to offer an article that should be both understandable and interesting to a wider readership. In this effort I acknowledge the helpful feedback received from the editor and reviewers.
Which other important research problems were not covered in the paper?
For the sake of conciseness and wider interest, the paper focused on a selection of research needs and their potential solutions. Other crucial research topics that deserve deeper discussions are the applications of AI to enable the prevention and early detection of cancers. These are areas that will greatly benefit from AI in combination with emerging technologies for the non- or less-invasive monitoring of different biological information readouts, for example those obtained in the blood.
For making the most of scientific and technological progress, there is also a need for discussing issues directly connected to the implementation of AI systems in the clinic, such as privacy and fairness.
This is not only about the potential legal and ethical implications of these advances. It is also about the inception and implementation of the underlying computational models. Researchers should consider such requirements, including their potential solutions and implications, as early as possible in their designs and investigations.
What currently most excites you about AI in cancer research?
I am excited about the possibilities of AI for bringing benefits to patients at different levels, as discussed in the article. Also I am excited about the advances and opportunities for helping researchers, clinicians and other health care professionals to make their work more efficient and meaningful, so that they can focus on the most demanding questions and problems of their work.
The development of systems that can learn by observing and interacting with human experts, and that can use such information to improve decision making, are thrilling research undertakings as well.
It is all about developing AI, data science and related domains as positive forces for humanity.