The data revolution in life science and healthcare

Introduction

Life science is all about deciphering biological complexity. With the rapid development over recent decades of new technologies and novel methods of data analysis, we are now in a position where integrated biological systems can be studied holistically. Data has always been the foundation for research, and the number of available data sources and data points is currently increasing at a remarkable rate. In combination with powerful computational resources and advanced methods of analysis, data can today provide new insights into human biology and disease mechanisms at an unprecedented level. This has brought life science into the era of systems biology and enabled the development of precision medicine whereby optimal treatments can be selected with the help of precise diagnostic methods. 

There is consensus within the scientific community that development of systems biology and the implementation of precision medicine is hindered, or at least slowed down, by challenges of sharing and utilizing data. This report addresses important questions related to data in life science and healthcare. In the first section, examples are provided of state-of-the-art data utilization in research, with the ultimate goal of providing new and better drugs and treatments. In the second part, different types of data, their value, and challenges of sharing and taking full advantage of available data are presented. The final part provides a broader discussion on the importance of trust, openness and collaboration in order to accelerate scientific advances for the benefit of all. Ensuring transparency in systems in which new knowledge is generated, is vital for the function of societies around the world.

The report builds on presentations and discussions at the symposium “The future of life science – the data revolution in life science and healthcare” held at Engelsberg Ironworks, Sweden, in May 2022 (see the appendix for the program with speakers). The aim is not to cover all the topics discussed at the symposium, but rather to focus on key topics to highlight immediate challenges and, most importantly, possibilities for the future.

Part 1: Implications of the data revolution in life science and healthcare

Molecular mechanisms, systems biology and precision medicine

What is meant by the data revolution in life science? Even though data has always been fundamental to all research, there are several developments that have led to the rise of the term “data-driven life science” during the last few years. Firstly, the advances have been fueled by an explosion in the amount of data from different sources. As Janne Lehtiö, Professor of Medical Proteomics at Karolinska Institutet, pointed out during the symposium, there has been an exponential increase in the amount of life science data during the recent decade, and in computing power.

In combination with increasing interest in cross-disciplinary collaborations, the scope of many biological and medical studies has expanded. Biological processes, mechanisms of disease, and the action of different drugs can now be studied at both a detailed mechanistic level and a systemic level. The result is the emergence of systems biology. 

Systems biology is the study of biological systems whose behavior cannot be reduced to a linear sum of the functions of their parts, and therefore focuses on complex interactions within, for example, an organism, tissue or cell. It includes experimental approaches and mathematical modelling. Increased knowledge at the mechanistic and systemic levels has led to the development of precision medicine within healthcare, where the best-suited treatment can be given to a specific patient at the right time, based on individual factors including genes, environment and lifestyle.

What is the status of data-driven life science, systems biology and precision medicine today? It can be argued that the road leading to our current position was paved by the Human Genome Project and the first sequenced human genome 20 years ago. After that breakthrough, several large mapping projects including the mapping of all human proteins (Human Protein Atlas, led from Science for Life Laboratory in Stockholm) have expanded the foundation for understanding human biology. Crucially, research data from such projects can be readily accessed, and this has led to rapid advancements in many areas. Mathias Uhlén, Professor of Microbiology at KTH Royal Institute of Technology, Stockholm, told at the symposium that at least half of the data used for research carried out by his group is data already available in the open access resource as part of more than 15 million web pages.

Cancer treatment is probably the field in which precision medicine has advanced the furthest. Thirty years ago, a cancer type such as lung cancer was classified as one disease. Advancements in immunohistochemistry enabled division into subclasses about 25 years ago, and with the recent introduction of genomic tests, different mutations and genetic profiles are now used as a basis for diagnostics. Based on this knowledge, novel drugs and optimized combination treatments have been developed. As a result, new treatments which are more effective, often with fewer or less severe adverse effects, are now available, leading to increased survival expectance among several groups of cancer patients.

Recent and ongoing research studies are able to explore how multi-omics approaches (those using a combination of data from different methods) can be used to diagnose cancers and allow even better and more precise selection of treatments (see Fact Box 1 for an overview of different -omics methods). Janne Lehtiö’s research group is combining genomics, epigenomics, transcriptomics, proteomics and immunohistochemistry in order to better match each patient with different anti-cancer therapies based on tumor characteristics. Analysis of data from the different sources shows specific patterns that can be used to group the patients. This new segmentation of cancer types results in greater precision in the diagnoses, with rich molecular information on cancer drivers, which is now being used to find novel combinations of treatments, to achieve better outcomes. The next step is to find new protein types, caused by cancer coupled defects in the genome, among the different cancers that can be used as targets for the development of cancer vaccines.

A promising field in the diagnosis of cancer, and probably other diseases, is profiling proteins in blood samples. A project led by Mathias Uhlén has mapped protein and metabolite levels in human blood. The technology that made this possible was recently developed by the Swedish company Olink. Starting with comprehensive blood protein analysis of 200 healthy individuals, it was apparent that each had its own unique protein profile, providing a fingerprint that was stable during the two years of the study. In the subsequent study, 10 000 new blood samples from patients were analyzed, covering 100 diseases and almost 100 patients per disease. Brand new preliminary data showed that there are specific differences between patients with different cancer types. The next question to explore is whether it will be possible to detect cancer-specific changes in the protein fingerprints before any symptoms occur. If so, this type of blood testing could offer a highly effective screening method for the early detection of cancer, which would be instrumental in successful treatment of most cancer types. Other factors that have been important for improved cancer survival are not connected with understanding molecular mechanisms or more effective drugs but rather the processes of patient interaction and procedures for disease assessment and care. In these cases, the utilization of data is important. 

Screening programs can be very effective for early detection of cancer but mass screening including all individuals of a certain age range can be problematic. The technologies used for testing may be expensive, and if the test method produces false positives it will cause unnecessary worries and anxiety. By developing smart questionnaires, only individuals who are at risk of disease are called for testing and the screening can be more effective.

Epidemiology research increases our understanding of determinants of health

Epidemiology studies the distribution, patterns and determinants of health and disease conditions in populations. Access to a large amount of data is obviously a cornerstone for these studies. The field has gone through a revolutionary development, following the advance in molecular and systems biology and progression in causal inference theory. Epidemiology constantly provides new insights into how different factors affect our health, including the effect of specific interventions on the public health status. 

Cecilia Magnusson, Professor of Public Health Epidemiology at Karolinska Institutet, pointed out that studies have often shown the power of preventive measures. It is clear that preventive interventions have decreased the number of deaths for several health disorders, such as heart and coronary diseases, and thereby contributing to longer life expectancy. Coronary heart disease mortality has decreased by 70 % during the last 40 years. Studies have shown that prevention accounts for 48 % of this decrease, while the contribution from medical treatment is 42 %. Prevention in this case means both lifestyle changes such as less smoking, but also interventions to lower the blood pressure and to treat pre-diabetes throughout a population.

Even though preventive measures evidently are powerful, the lion’s share of investment is in research and development to provide new treatments. With the models for financing healthcare systems in most of the world today, prevention is often not commercially viable. One would think that the incentives for government to work more actively on prevention should be strong, since investment in prevention will pay back in lower healthcare costs. However, preventive measures often include actions that affect living conditions and lifestyle. The introduction of requests or regulations that may affect the individual freedoms will encounter resistance and may therefore be less attractive for politicians to implement.

The links between environmental factors affecting our health (as observed in epidemiological studies) and biological mechanisms at the molecular level are often still shrouded in mystery. One field in which the understanding of molecular mechanisms is still very limited is psychiatric and neurodevelopmental disorders. At the same time, there is a growing number of people diagnosed with such disorders worldwide, and more research in these fields is necessary.

PART 2: The value of health data and challenges of sharing data

The examples of tremendous advancements in the understanding of human biology and disease mechanisms given in this report have largely been fueled by technology development in combination with sharing of data. At the same time, it is often claimed that the implementation of novel technologies in healthcare is too slow, and that the data generated in healthcare is underutilized due to challenges in sharing data. Restrictions in sharing health data and other sensitive data are necessary to protect personal integrity. However, a common view is that the regulations are no longer fit for purpose and that an updated, and less restrictive, regulation is necessary for research in order to take full advantage of modern technologies aiming to provide better treatments and care.

Data collected within healthcare is primarily used to plan, monitor and evaluate the care of individual patients or for planning and evaluation of the way in which care is provided. The use of healthcare data for purposes such as research, product development or community planning, is called secondary use. Today, challenges usually arise when patient data or other types of sensitive data is used by other stakeholders such as university scientists, R&D-departments of pharma or technology companies, or even a different department within the same healthcare provider.

During the last ten years it has also become evident that data from sources other than the laboratory bench or the healthcare system is often valuable for understanding health-related issues. For example, during the Covid pandemic, data from mobile phones was used to analyze how people traveled and how that affected the spread of the infection.

James Wilson, Professor of Philosophy at University College London, highlighted what he thought was the most important question to address: How can health data be made more valuable for the purposes of clinical care, health research, and commercial companies, while maintaining high levels of public trust? To start with, he presented a model in which the data sources were divided into three groups:

1. Data collected by health professionals in the course of providing healthcare, or by health researchers (such as hospital records, clinical trial data and omics data).

2. Data self-collected by individuals for the purposes of monitoring or enhancing their health or wellbeing (such as data from health and wellbeing apps and smartwatches).

3. Data collected for reasons other than health that allows inferences about personal health or wellbeing to be drawn.

Professor Wilson further pointed out that governments can exercise most regulatory control over the first group. Hence, this data type is most intensely discussed by governments, even though it may not be most important overall. But failure to maintain high levels of trust in the last two data types – which may be hard to control by governments – may also affect public trust in the first type of data, and hence the health care system and public bodies. Therefore, both public organizations and commercial companies are responsible for maintaining high levels of trust in the whole system.

When discussing the value of data, it is important to remember that this will be different for different stakeholders and that the whole value may not be adequately captured in financial terms. Often, there will be win-win situations. This applies, for example, when data is given with consent from patients to researchers who use the data to develop new and more effective treatments, which benefit the patients, or when hospitals curate their administrative data and provide it to companies that develop better planning tools. However, values will sometimes be competing or conflicting. The value the patient places on their medical confidentiality differ from how researchers or innovators value a large dataset. If citizens’ choices are valued by giving them control over secondary uses of “their” health data, the value of the data for research or health system planning may be degraded. Professor Wilson argues that the conflict between competing values, as in the example of citizen choice and completeness of datasets, becomes more extreme if public trust diminishes.

PART 3: The matter of trust 

James Wilson argued that trust among stakeholders is crucial in order to share and to fully utilize the value of health data. Sharing health data, and probably other types of personal data, can be viewed in different ways. Firstly, health data can be considered as an ordinary market commodity that can be bought and sold, refined and transformed. Health data may also be seen as a gift, given without charge, as an act of generosity that will benefit others, and sometimes also the donor in the long run. Health data may also be shared as an act of solidarity where the data is used for the good of society or a community. 

All of these types of “transactions” have their ethical considerations and risks stakeholders’ mistrust. On a commercial market there is always a risk that the seller or the buyer feel that they have been misled. Is the price fair and what do the paragraphs in the contract mean? On the other hand, if the data is given as a gift, it is most probably important for the donor that it is clear that the data is to be used for something that is beneficial for individuals or society, otherwise the willingness to donate data will be most likely to decrease over time. If data is given as an act of solidarity, there is a risk of perceived exploitation if the distribution of risks and benefits is seen to be inequitable.

How can trust be created and maintained so that people are willing to share their data for research and development? The question of trust is presumably much broader than just an issue of trust linked to the sharing of personal data. The trust for governments, healthcare systems, academic research organizations or commercial companies obviously varies between individuals and citizens of different countries, as well as subgroups within society. In Sweden, trust in science and researchers has traditionally been high. In a public survey carried out in 2018 by the organization Forska! Sverige, 95 % of respondents were positive about sharing their health data for research and purposes aiming to promote health. 

Public level of trust for research and new knowledge is a matter of debate. Arne Jarrick, Professor Emeritus of History at Stockholm University, said at the symposium: “Some friends of knowledge say that we no longer live in a knowledge society.” However, he pointed out that such a statement would be hard to prove since historical data is not available and therefore any changes in the level, and distribution, of knowledge in society are impossible to identify. What history can tell us is that scientists are usually the pioneers when it comes to embracing, or at least accepting, new knowledge, and that the public follow with some delay.

Knowledge is instrumental for a democracy to work. Åsa Wikforss, Professor of Theoretical Philosophy at Stockholm University, even claimed that “democracies require that citizens accept expertise and research” and that today we live in a very unreliable information environment. Information technology platforms such as social media allow both true facts and false claims to be spread worldwide in an instant. So, even though the amount of accessible information is greater than ever, it is probably also more difficult than ever to distinguish true from false information.

Fake ‘facts’ can be spread for political or commercial purposes, and it is clear that false news often spread faster, longer and recieves wider attention on social media platforms than true facts. Sometimes it is even difficult to discern what science to believe. Adam Rutherford, Honorary Senior Research Associate at University College London, pointed out that historically there are several examples of how science has been used for political purposes, such as to legitimize racism in general and more specifically to justify eugenics programs during the first half of the 20th century. Even today and during the last two decades, alleged linkage between nationality, ethnicity and IQ received much attention, even though it has been shown by several scientists that the data on which these hypothesis rely, are incomplete and the analysis is flawed. 

Should responsibility for distinguishing true from fake information be assigned to individuals? Will the problems be solved if people learn to think more critically? Professor Wikforss claims that this would be helpful, but that it is not enough. Studies on what is shaping our thinking and how our beliefs are formed, show that we tend to believe in ‘evidence’ that supports what we already believe in and that our thinking is very much dependent on our social environment. This means that it is most likely that we will believe in the same things tomorrow as we do today and that we are likely to believe what our friends or role models tell us. When we are surrounded by a lot of disinformation (false information that is deliberately spread to deceive people) our critical thinking will be undermined. Hence, Åsa Wikforss suggests that stronger legislation is necessary, which means that providers of technology platforms such as social media are to a larger extent responsible for the content. In addition to what Åsa Wikforss presented, Arne Jarrick put forward two important factors leading people to dismiss new knowledge even though it is supported by strong evidence: the new knowledge may be counterintuitive, and the new knowledge may challenge your usual behavior.

People generating and communicating new knowledge have an important mission today. Studies on how new knowledge has been accepted by the public throughout history may provide a few hints on what is likely to work. Arne Jarrick has identified four different strategies that scientists have used to achieve acceptance of their ideas:

• The low-key strategy. Downplay the novelty of the idea and try not to provoke.
• Be over-confidential and emphasize the fantastic features of the idea.
• Convince a specific target group. Identify a relevant group of people and focus on getting acceptance from them.
• Be honest. Be transparent about what you do and what you don’t know. Also show the limitations of the idea and if there are any circumstances where the idea is difficult to prove.

There is no historical evidence to show that any strategy is better than another, but it is a safe bet to suggest that you should understand your target group before choosing your strategy. The Covid pandemic has often been called an infodemic. Many felt overwhelmed by all the available information. There was, and still is, a large amount of false information in circulation regarding how the virus is spread, the impact and adverse effects of the vaccines, and political ulterior motives associated with different strategies. A strong polarization was observed on how best to fight the pandemic, and in some cases that tipped over into very strong emotional expressions and even direct threats. Sadly, there are examples of scientists who moved away from the public debate, not daring to share their knowledge and opinions.

Fortunately, there were also outcomes that were much more positive. Firstly, it is clear that open sharing of research data on the virus and its spread resulted in an extremely rapid increase in new knowledge that led to better understanding on how to best treat the patients and the development of new vaccines at unsurpassed speed. Secondly, news coverage of the pandemic has given many people outside the scientific community a better understanding of how research works – that research is an iterative process where ideas supported by data is presented and then tested against new data. Some hypotheses are dismissed, and others are modified, strengthened and refined until there is enough evidence to call it a fact. No single data point can tell us the truth, but when the route to new knowledge is transparent, it will be more easily accepted.

Moving forward

Shifting the focus back to the use of data in life science and healthcare, it is obvious that data-driven life science is already providing new means to treat severe diseases such as cancer – and hence saves lives. To continue and accelerate this development, effective sharing and utilization of data is key.

Data sharing and utilization is a question of technology, legislations and trust. In several countries investments are ongoing to build the technological infrastructure needed for data-driven life science and precision medicine. Efforts are also made to modernize legislation in order to facilitate development while at the same time protecting personal integrity. In the EU the first steps on the road to the so-called European Health Data Space (EHDS) have already been taken. The idea is to build an EU common infrastructure for health data, including rules, common standards and practices, and a governance framework. The aims are to provide citizens with increased access to and control over health data, and to provide “a consistent, trustworthy and efficient set-up for the use of health data for research, innovation, policy-making and regulatory activities”.

In May 2022 the Swedish government decided to set up a review of opportunities for secondary use of health data and how legislation may be updated to directly or indirectly strengthen health care.

New legislation will presumably include measures for maintaining trust among different stakeholders. Clear regulations will diminish uncertainties and the risk of conflicts. However, addressing the question of trust over and above the legislation is important. Ensuring high levels of trust in our research institutions, healthcare systems, public organizations and companies will most likely increase citizens’ willingness to share health data, especially if it is shown to be beneficial for society, for other people and for themselves.

People working in, or close to, research have a responsibility to safeguard and improve levels of trust. Explaining how new knowledge is generated, being open about what you do and don´t know, as well as why you have gained the knowledge is probably a good start. Hopefully threats against some scientists during the Covid pandemic will not scare other experts away from the public debate. Instead, it is vital that they ensure that their voices are heard.

Please note that all sources are listed in the PDF version.