AI and Big Data in oncology:
How ready are we?
The availability of larger volumes of data is already a reality in healthcare and represents an opportunity for clinicians to improve cancer care. However, technical and socio-cultural issues limit their use in practice. ESMO has launched a Big Data Taskforce aiming to address these issues.
In an era where data are being generated at a phenomenal pace—by 2020, it is expected that medical data will double every 73 days, and that the average person will generate more than 1 million gigabytes of health-related data in their lifetime—it is increasingly difficult for clinicians to process all the available information that could influence treatment decisions. Furthermore, traditional analytics and machine technology have limited capability to utilise large complex datasets such as the so-called ‘big data’, and a change of paradigm is needed today to make the most from these potential sources of information.
Artificial intelligence (AI), also increasingly called ‘cognitive computing’, represents an exciting potential solution to analysing and extracting meaningful information from big data.
This new computation paradigm is capable of learning and building knowledge from various information sources; it understands natural language and can enhance the cognitive process of professionals to help improve the quality and consistency of decision making.
In 2018, the European Commission called for €20 billion to increase investments on AI and innovation across Member States by 2020, and presented a series of measures to encourage cooperation, exchange best practices and define a plan for the future in the field. Early steps have resulted in the setup of a European Coordinated Plan on Artificial Intelligence, based on the European Strategy on Artificial Intelligence, and boosted research opportunities at international and national levels. The plan details the actions that need to take place between 2019-2027, and focuses on increasing investment in AI; making more data available through the creation of European data spaces; fostering talent by supporting advanced degrees in AI; and ensuring trust by developing ethical and trustworthy AI. Importantly, health has been identified as a strategic area for AI and innovation.
ESMO is committed to taking an active part in understanding and discussing the role and impact of big data and AI in the future of oncology, in order to inform its members.
The Society decided to create the Big Data Task force that aims to address the issues of artificial intelligence and big data, and their impact on the oncology community.
Champalimaud Clinical Center, Lisbon, Portugal; Chair ESMO Task Force for Big Data
How important do you think AI and big data are in shaping the future of oncology management?
Big data are already here, this is not a futuristic concept. Data are being generated and captured in many ways, and at a pace we can no longer process as humans. This includes highly controlled structured data from clinical trials, which currently forms the basis for most decision-making. However, most medical information is generated and captured in a less controlled way, including from registries and electronic patient files, or in unstructured forms such as social media (e.g. patient blogs and internet chat rooms). Unstructured data are much harder to process.
We need to be smarter in the way we collect and use information .Clinical trials will always be important; however, many questions cannot be answered by trials, for example the best sequence of treatments for each individual patient. Using real-world data could help answer these questions.
How do you see AI and big data being used in clinical practice?
An example of how AI is already being used is the collection of health data through smart watches; this is already a part of normal life, but we can make better use of this technology.
Cognitive computing machines in development have the capacity to read, understand and process natural language information. These tools have the potential to become a key part of clinical practice. Using predictive analytics (a system that can analyse data and devise the answer to a specific question), clinicians will be able to obtain evidence-based answers to complex diagnostic and management questions relevant to individual patients. Furthermore, it will allow integration of different types of information, such as clinical, genomic and other ‑omics, imaging and patient-reported outcomes, in a faster and easier way.
What are the challenges to integrating AI and big data into clinical practice?
A major challenge is determining how to extract valuable information from the enormous amount of data available, and research is ongoing to determine the best methodology to analyse data and reduce/eliminate unhelpful ‘noise’. Another area currently being investigated is how to store and transport these huge amounts of data, and importantly how to keep the data secure. There is also an ethical challenge relating to patient privacy. Many patients don’t want their health data in the public domain due to fears of prejudice, insurance issues, etc. New EU laws such as the General Data Protection Regulation mean ensuring that patients retain their right to keep their personal information private by refusing permission to include/use their data in cognitive computing tools.
What do clinicians and patients feel about using AI in cancer management?
Overall, people are excited but with some fears, especially around invasion of privacy. We need to take measures to avoid privacy issues and tackle other challenges pertaining to AI and big data use in oncology. That is why ESMO is committed to be a partner in discussions on big data and AI, to define the questions that need to be asked and to safeguard the rights of patients and ESMO members, with the ultimate goal of improving patient outcomes. AI integration to oncology is inevitable. Clinicians are realising this and want to make the most of it. We need to embrace AI with the necessary safeguards in place.
Netherlands Cancer Institute, Amsterdam, The Netherlands
How important do you think AI and big data are in shaping the future of oncology management?
I have not met anyone who isn’t convinced that big data and AI will reshape healthcare on multiple levels, including diagnosis and treatment. Most clinicians, including myself, are enthusiastic about having a ‘learning healthcare system’ whereby data (hospital data, clinical trial data, etc) available from all patients are utilised to improve outcomes. However, some cautiousness is needed as the use of AI tools in clinical practice may not happen if the challenges surrounding its use are not discussed and overcome. A concerted effort is needed to make sure this happens.
ESMO has tasked itself to understand how AI can be best used to benefit its members and their patients. For example, important discussions include how the use of AI may impact ESMO Clinical Practice Guidelines in the future and how to incorporate AI into the guidelines process.
How do you see AI and big data being used in future clinical trials and research in oncology?
At the Netherlands Cancer Institute, we are conducting research in the AI area and partnering/collaborating with other organisations and companies who are researching and developing AI for use in oncology. AI could be used to analyse and make sense of the vast and increasing amount of genomic and proteomic research data being generated in cell lines. AI could also be used to help generate new cancer treatment hypotheses that are then tested in cell lines. Indeed, AI is already being used to investigate cell signalling pathways to find actionable targets in cancers. In diagnostics, AI is being used to analyse radiological scans to facilitate diagnosis of cancer.
How do you think AI and big data will change the doctor–patient relationship?
Most clinicians and patients are cautiously enthusiastic about the concept of AI in healthcare. There is a realisation that it could allow more quality time to be spent in doctor–patient communications, providing advice and discussing the evidence-based management options that have been identified with the help of cognitive computing tools. Importantly, these tools will facilitate and not replace the decision-making process—it will remain essential for doctors to discuss and consider more intangible factors, such as patient preferences, when making the treatment decision.
What are the challenges to utilising AI and big data in clinical practice?
Integrating AI into healthcare is much more complex than integrating AI into something like a game of chess. In chess, there are a limited number of possibilities (moves) and basic rules; it is relatively easy to teach a machine how to play chess and the machine can learn to win. However, in oncology the situation is much more complex, with no solid rules, many possibilities and varying compounding factors. Teaching a machine to analyse patient data and come up with answers is much more complicated. In addition, social aspects and patient wishes must be considered in treatment decisions. The challenge is to find a way of processing all the variables that provides simple and useful answers. Another challenge is understanding whether a cognitive computing tool can adapt to geographical differences in attitudes to healthcare, availability of medicines, etc. Importantly, a concerted effort is needed from all stakeholders (healthcare professionals, programmers, AI vendors, etc) to discuss and agree their ideas, attitudes and goals for cognitive computing in oncology. This is vital to ensure mutual commitment to the development and integration of clinically useful tools, and achieving the best outcomes for patients.