The healthcare system, patients and businesses, can all benefit from using our pool of health data in new and unique ways. But the area lacks the legislation necessary to make it happen.
When a patient visits the hospital with a broken finger or a swollen ankle, their scan images are likely to be viewed and assessed with assistance from the artificial intelligence created by Danish startup Radiobotics.
“We have been quite successful in building algorithms that can detect diseases from just an X-ray scan. Our solution has received a CE mark, which means it has been approved for hospitals across Europe,” says Stine Mølgaard Sørensen, COO & Co-founder of Radiobotics.
Their algorithm is specialised in analysing musculoskeletal X-rays. It highlights relevant findings and can even compose a text report for doctors. Before the introduction of this new algorithm, these were routine tasks that needed to be completed frequently, yet weren’t particularly difficult. Now, in an area where there is already a lack of clinicians, this new system is considered to be a huge boon.
“The potential uses of health data are vast because the information they contain is the cornerstone of training these new algorithms. Today, there are huge shortages of radiologists, and we must do everything we can in order to support them in becoming better and faster at what they do,” Mølgaard states.
However, if we wish for more innovative ideas like this one to find their way to the market, a new system is needed: a system ensuring that health data is made more readily available and used to enable patients to get better treatment, while also not forgetting to take data security into account.
Retrospective data benefits future patients
In Denmark, we have health data that is uniquely well-structured, due to decades of digitisation. But there is no adequate legislation governing the so-called “secondary use” of health data, which is a problem, according to Henrik Krogen, Managing Director of GE Healthcare.
A new legislation was passed in Finland last year, describing the guidelines for the secondary uses of health data – i.e. how to use data from one patient in order to improve the situations of many other potential patients in the future.
For us, access is crucial, but if I want a new partnership with another hospital, there are no clear procedures as each institution can make their own interpretations of the current rules
Stine Mølgaard Sørensen
Rules like this must take patients’ data security into account at all times. Krogen believes there can be no compromise on the matter.
“We can’t just create new products without making data security a top priority. At the same time, we mustn’t let data security become an obstacle for creating better diagnostics and therapies. We can send a man to the moon, why, then, is handling data security and innovative use of said data at the same time so difficult?” he asks.
“Having enough data is the most important factor for the development of AI in the health care sector. But it appears we already have almost all the data necessary to improve the treatment of most major diseases. I believe that it is both our ethical and scientific duty to make the information we have available to people experimenting with new medical techniques so that the data of yesterday’s patients can benefit the ones of tomorrow,” Krogen points out.
Alongside legislation for data usage, he proposes to create a data pool, where companies and researchers can use the data in a structured way – which can also be a platform for testing these new sets of rules.
“In order to make sure that everyone uses data properly, there must be legal procedures and sets of guidelines built around its management. There should also be an AI institute under the Health Board to enable access to data for healthcare professionals, both from hospitals and from other practices in the field, but only after they have complied with full security requirements,” Krogen says.
No data – no business
In order to develop their algorithm, Radiobotics needed access to public health data, which they acquired through a collaboration with Bispebjerg Hospital and the head of their radiographic department. Since hospitals own the health data that they produce, Radiobotics has a data processing agreement with the hospital, and the hospital needs to have a license from the Danish Agency for Patient Safety.
“The data isn’t ours; we just have access to it. The data we get is pseudonymised, which means Bispebjerg knows the name of the patients, but we do not. For us, access is crucial, but if we want a new partnership with another hospital, there are no clear procedures, as each institution may have their own interpretations of the current rules,” Mølgaard explains.
Although the young company has been granted access to the data needed, its COO agree with Krogen that clear rules are lacking in this area.
“There isn’t enough of a clear framework on how to gain access to the information we wish to use . This makes it hard to plan for future products, as we still don’t fully understand what the rules are,” she says.
A challenge that requires an alliance
When the question of data access comes into play, both Radiobotics and GE Healthcare have experienced how much it can delay a project.
This is a challenge Michael Grunkin, founder and managing director of Visiopharm, also knows very well.
Although there is a strong motivation for hospital researchers and Visiopharm to collaborate and innovate, things tend to stall when they have to address regulatory data requirements.
“This is a problem both companies and regulatory committees share. We want a powerful new technology on the market because it benefits patients and lowers costs in the health care system. Legislators alone cannot balance these various considerations – we must achieve it together, as a society. This situation calls for an alliance between all interested parties,” Grunkin says.
In his opinion, private and public stakeholders must work together to maximise the utility of health data and he points to similar alliance initiatives in both Germany and the United States.
“One of the biggest challenges the health care sector is facing all over the world is that technology is evolving faster than legislators and government agencies can keep up. This is a fundamental structural problem. We believe that alliance initiatives – ones used to provide input and recommendations to legislators, regulators and the industry as a whole – are the most effective responses to these complex challenges,” he claims.
And if the problem persists?
“In Denmark, we have a strong entrepreneurial environment, with innovative AI-based companies working in health tech. If we are unable to provide effective solutions to these complex challenges, it will be very difficult for companies – new ones in particular – to succeed. If they cannot access data in a competitive and scalable way, they will not be able to survive. If this happens, only very large companies will have any success in the sector,” Grunkin concludes.
Competitors are not waiting
Radiobotics is one of the many companies to originate from the Danish AI scene. They are not discouraged by all the regulations and requirements they have to comply with: they merely see it as a part of the health tech industry.
“I don’t have any problems with following the rules. I have more of an issue with not always knowing in what way the current rules are being interpreted,” Mølgaard explains.
If there is no consistent, dependable way available to access the Danish databases, the best solution may be to go abroad. But she does not think this will be optimal for her company or for the world.
“To put things into perspective: we have a competitor in the US that has exclusive rights to all data in an entire hospital in New York. I think data access is something we can all benefit from both for research and industry purposes. Patients and our society as a whole will see the great benefits this new system could offer,” she says.