The latest meeting of the My City-Lab Talk Series “AI in the Fight Against Cancer” took place on February 25th, 2020. The event was organised by Health First Europe, as a partner of My City-Lab project, to discuss the role of Artificial Intelligence (AI) in the fight against cancer, from screening to treatment.
Why is AI so important for cancer care? Cancer is one of the areas where AI is revolutionizing the health care sector. Thanks to AI, the software can analyse medical images, spotting irregularities and supporting health care professionals in deciding whether or not a second physician needs to visit the patient. Using AI, doctors will be able to select treatment options that work better with specific tumours or patients; machine learning algorithms can pull information from doctors’ and radiologists’ notes in electronic health records in order to identify how particular patients’ cancer progressed. A good illustration of AI application to cancer diagnostics can be found in thyroid cancer care: radiologists and artificial intelligence specialists partnered to develop an algorithm that can determine if lumps on a thyroid should be biopsied.
As pointed out by the moderator Damien Gruson, in the clinical laboratory, chemistry and haematology departments have been the first ones to adopt new technologies and algorithms into their workflow. Rapid changes in health care coupled with advances in technology have stimulated the evolution of new approaches for laboratory automation. In particular, the emergence of AI applied to laboratory robotic systems offers a great promise for streamlining the clinical laboratory.
Yiannos Tolias, from the European Commission’s Directorate-General for Health and Food Safety (DG SANTE), highlighted the commitment of the EU institutions in speeding up AI application for more effective and patient-centred healthcare while coping with AI challenges. The European Commission is currently working in two main areas: data collection and sharing for public interest and regulation for a trustworthy AI. With respect to health data sharing, the Commission announced an ambitious project: the European Health Data Space, to promote data exchange and support research to ultimately advance treatment on complicated conditions, such as cancer. Yet, some legal challenges arise, whereas there is consensus among member states on the primary use of data, not all the stakeholders seem to agree on the access to and rules over the secondary use of data. Moreover, interoperability and quality of data remain challenging. The White Paper on AI, launched by the Commission on 19 February, seeks to give direction with regards to AI regulation in order to achieve an ‘ecosystem of excellence’ and an ‘ecosystem of trust’. The document further acknowledges how AI can pose some important issues when it comes to ethics and the effective functioning of the liability regime.
The European Commission’s Directorate-General for Health and Food Safety (DG SANTE) and the Directorate-General for Communications Networks, Content and Technology (DG CNECT) are closely collaborating to find suitable solutions to tackle the challenges in the application of AI in health care. Margherita Fanos, from DG CNECT, explained that the Commission is working to boost national and regional funding for promoting the use of AI solutions in the public sector and to enhance digital skills amongst health care workers. With respect to cancer care and health data, Ms Fanos emphasised two main initiatives the European Commission launched:
- the EU Beating Cancer Plan, which aims to reduce the burden of cancer across Europe, by promoting research and innovative technologies. The use of artificial intelligence is seen as an essential tool to significantly improve the precision of early diagnosis, as it has been already demonstrated in breast cancer care.
- The “1+ Million Genomes” initiative, which aims at accessing 1 million sequenced genomes in the EU by 2022, setting up a collaboration mechanism with the potential to improve disease prevention, allow for more personalised treatments and provide a sufficient scale for new clinically impactful research.
Insights on the patients’ perspective came from Antonella Cardone, representing the European Cancer Patient Coalition (ECPC). She stressed how AI has been proven to be incredibly beneficial for cancer screening, notably in:
- Skin cancer screening, where AI can prove the accuracy and allow surgical removal before cancer spreads
- Colonoscopy, developing a real-time endoscopic image diagnosis support system
- Lung cancer screening, where AI system uses 3D volumetric deep learning to analyse the full anatomy on chest scans
Additionally, participants stressed the potential of AI in clinical trials. AI can predict cancer treatment toxicity and side effects of polypharmacology and link patient data to clinical data trials.
However, all these innovative solutions seem to be isolated best practices, not fully shared and applied in all EU countries. Why is that so? There is a lack of data infrastructures and interoperability systems and rules ultimately impacting on best practice sharing and the heterogeneity and quality of data. Each health care setting collects data on its own way and when shared might not follow the same parameters to be used or compared with other medical departments.
Beyond technical challenges, some ethical concerns arise, especially linked to the so-called “black box medicine” i.e. the use of opaque computational models to make health care decisions. Big data-based AI can suggest a certain cancer treatment without exposing the rationality behind its decision. Data science and cancer care still belong to different fields of professionals, where oncologists do not understand algorithms and programmers do not understand cancer care.
Also, to a certain extent, our legislation still needs to catch up to digital services and technologies. Jelena Malinina, representing the European Consumers Association (BEUC), pointed out that our legal systems were human and not machine focused, which arise questions over liability. AI in health sector challenges the existing legal liability rules, where clinicians are currently liable for software malfunctions that contribute to an incorrect diagnosis. Should the uptake of machine learning suggest more accountability from manufacturers and sellers of AI software? How we deal with the question of legal liability will definitely affect the spread of technology in the health sector.
The main take away of the discussion is that fighting cancer remains a substantial and complex challenge. Every person living with cancer has a unique journey, and AI and new technologies can support health care providers in addressing patients’ unique needs. The concerns on AI application to cancer care discussed during the meeting, from ethics to liability, are shared by all the EU member states, and it is worth working together to find common solutions and rules to build a trustworthy AI ecosystem to ultimately save lives.
Please click here to see the pictures of the event.
My City-Lab project – financed by the European Regional Development Fund (ERDF) – aims to create two new primary care structures, called My City-Lab, integrating the innovation of laboratory medicine and mobile health. The scope of the project is to facilitate access to laboratory tests as part of an integrated and collaborative approach to ambulatory care of a chronically ill individual, as well as to contribute to the dynamic monitoring of patients with chronic diseases.