The Digital Transformation of Cancer Care

Wim Oyen, Co-Chair of the European Cancer Organisation’s Digital Health Network, opened the next session with fellow Co-Chair Regina Beets-Tan.

As well as being a Co-Chair of the European Cancer Organisation’s Digital Health Network, Regina Beets-Tan has been an active member of the EU Cancer Mission Board since 2019. She highlighted Recommendation 8 from the Mission,[25] which calls for the creation of a European Cancer Patient Digital Centre, where cancer patients and survivors can deposit and share their data for personalised care. Patients, she said, should be in the driving seat, with a health passport containing guideline- led information on their cancer, optimal treatment and lifestyle recommendations, and that includes them in shared decision-making.

A roadmap on patient-driven governance is needed, in which patients feel safe and do not have to beg for their own data, and their needs are taken into account. The data should be used to build artificial intelligence-led predictors of outcomes and make a real difference to cancer treatment.

digital 04955

Row 1 - Regina Beets-Tan (European Cancer Organisation’s Digital Health Network); Gilliosa Spurrier-Bernard (European Cancer Organisation’s Patient Advisory Committee); Anastassia Negrouk (MyData-TRUST). Row 2 - Wiro Niessen, (Erasmus MC/TU Delft); Wim Oyen (European Cancer Organisation’s Digital Health Network); Sara Cerdas MEP (European Parliament’s Special Committee on Beating Cancer). Row 3 - Fred Prior (University of Arkansas); Christian Stoeckigt (Hologic).

An Avalanche of Increasingly Complex Data

Fred Prior, Professor and Chair of the Department of Biomedical Informatics, University of Arkansas, said biomedicine has been inundated with an avalanche of more and more complex data that is too much for humans to process. Smartphones gathering data for tracking health and disease only adds to the deluge.

But with computing and mathematical modelling, it can be used to improve care in rural and underserved communities, and to improve therapeutics, precision health, care delivery and prevention, allowing clinicians to become proactive in disease management.

The huge pools of data from gene and population analyses can be examined, and predictive data modelling can turn data into actionable knowledge. The question is how to improve the delivery of healthcare for all. Personal medical histories can be leveraged to improve data quality and reduce errors, and perhaps the most valuable tool is artificial intelligence.

The tools used by healthcare professionals are becoming smarter, which is taking the labour out of diagnostics and leaving the decision-making to the expert. However, that presents challenges, the biggest being analytical models need huge amounts of data to represent the true variance of the human population. Diseases themselves and how they interact are also highly variable, and it is important to understand the way cancers adapt to the tumour microenvironment and alter following therapeutic challenges.

Figuring all of this out will require data pooled from millions of people, which presents a huge risk in terms of patient privacy. It is crucial it is protected and prevented from becoming a course of revenue, without putting roadblocks up in the way of research.


Combining Multiple Data Sources

Wiro Niessen, Professor in Biomedical Image Analysis at Erasmus MC/TU Delft, said these are interesting times in the field of medical imaging. Since 2012, artificial intelligence has been state-of-the-art in recognising images, which, despite concerns, has the potential to improve cancer care.

It is important to combine large datasets and use open data, which is possible with medical imaging. This means the images can not only be objectively analysed and quantified but also combined with genetic and environmental data and related to clinical outcomes to determine the optimal therapy for a particular patient. But beyond that, artificial intelligence needs to be able to stage a disease. For example, in low-grade glioma, an artificial intelligence image classifier could mean a patient could avoid having a biopsy. Certain tasks still require human intelligence, however, and it is important to understand where the line is drawn.

The future of data-driven health is to make all data available to deliver on the promise of personalised medicine. This will require investment in infrastructure, novel algorithms and validated techniques to create an ecosystem that optimises data use and implementation.


The Drawbacks of the General Data Protection Regulation

Anastassia Negrouk, Chief Operating Officer at MyData-TRUST, said the EU GDPR presents challenges in cancer care as it thinks data flow can be controlled and considers borders. This is an old-fashioned view of data and privacy, as data flow cannot be controlled with high precision to the end of its journey.

GDPR has been identified with negative impacts on life sciences with detractors suggesting its implementation is burdensome, that it confers heavy constraints with little proof of benefit, is a source of uncertainty and is a barrier to international cooperation. However, on the other hand, it has required more organisations to define workflows, created greater clarification about stakeholders, stimulated debate and enshrined risk-based regulation.

The major pitfalls are the lack of harmonisation in implementation, an insufficient code of conduct, suboptimal support, a lack of appropriate instruments for implementation and no standard contractual clauses for data transfer.

While the European data strategy,[32] and the European Cloud Initiative,[33] are both welcome, data knows no borders and the region must not lock itself out from the rest of the world. Data needs to be harmonised, which requires the right instruments for compliance.


Flexible and Progressive Data Consent

Gilliosa Spurrier-Bernard, Member of the European Cancer Organisation’s Patient Advisory Committee, Chair of WECAN, and Member of the Melanoma Patients Network Europe, said a European data centre is potentially very interesting and could provide some solutions. However, action is required, as data on even the best treatment are limited. Informed decision making would be improved, for example, by information on cellular function, lifestyle factors and expected response to treatment.

Patients need easy and compulsory access to their complete records, which should use standardised reporting and terminology. Data also needs to be shared safely, and processes put in place to ensure it does not expire or degrade. Consent should be flexible and progressive, so data is not left floundering for years.

For patients, this is a life or death issue. Accessible data would facilitate patient engagement and finally make research patient-centric, which is more valuable than a vague promise of scientific progress, and could give patients agency to initiate their own research.


AI Could Clear the COVID-19 Backlog

Christian Stoeckigt, Head of Scientific Affairs and Medical Education at Hologic, reminded the audience that breast cancer screening was stopped in many countries due to the COVID-19 pandemic, creating a huge backlog.

Artificial intelligence-based image guidance to assist clinicians is greatly needed to overcome this backlog and speed up diagnosis. There has been a great deal of progress since the introduction of analogue mammography and there is a huge chest of artificial intelligence tools now available.

However, the backlog will not be cleared soon due to data protection barriers. It is important to demonstrate patients are empowered by owning their own data. There are some issues to be addressed but they must be in control of their healthcare, and industry is onboard.


A Data Space to Boost Cancer Prevention

Sara Cerdas MEP, Vice-Chair of the European Parliament’s Special Committee on Beating Cancer, Shadow Rapporteur for the EU4Health programme, and Co-Chair of the ENVI Committee Health Working Group, said data has an important role to play in the fight against cancer, as it is multiple diseases, with multiple causes and multiple interventions.

She has called for the creation of a European data space that includes with a patient registry. This could help primary and secondary cancer prevention by providing better data not only on risk factors but also indicating whether individuals have been vaccinated and helping to ensure their screening information is up to date.

The data should be provided to researchers for observational and cohort studies, which could offer answers for many of the outstanding questions. While that requires better connectivity and the enshrining of patients’ right to a second opinion, the safety of data also needs to be guaranteed. Above all, we need to move on from talking and act on the data space, so that digitisation can fully contribute to the fight against cancer.

However, Prior emphasised in the following discussion that data needs to be carefully curated, and accurate from the beginning, as the further away you are from the entering the data the harder it is to clean up.

Negrouk added that the concept of ownership remains to be clarified, as one cannot sell human tissues but one can sell data. She nevertheless believes that private institutions, including drug and device developers can use data for good ends, which calls for a fair interaction based on a transparent relationship, framed by ethical values.