The Digital Divide Based on Development and Availability: The Polish Perspective

Abstract

Contemporary Poland has achieved great strides in the process of digitization of healthcare with some impressive implementations in recent years. However, despite the marked process, the situation regarding broadly understood digitization, innovation in medicine, development of approaches and solutions based on Artificial Intelligence remains imperfect. Having said that, there are dynamic institutions that have registered both clinical and scientific achievements based on digitisation and whose mission, in part, is to influence decision-makers so that the development of modern technologies encounters as few obstacles as possible. Current challenges are the result of the Polish healthcare system structure, which was problematic even without the context of innovation and digitalization. Consequently, this is reflected in the approach to medical education as well as in the attitude of healthcare professionals towards solutions which are considered new and unknown. This chapter summarises the challenges specific to Poland and healthcare digitisation, and highlights the progress that has been achieved thus far.

1 Introduction: An Overview on the Use of Artificial Intelligence in Healthcare

Artificial Intelligence (AI) has become one of the crucial tools for improving healthcare worldwide. Its disruptive nature is evident in almost every healthcare sector across the globe, and its exponential growth led to the use of the term “AI revolution” in healthcare. In the EU and US there were only 21 certified AI algorithms for healthcare in 2015 and this number has increased to more than 450 algorithms after just 5 years.¹ AI growth is particularly visible in radiology, cardiology and oncology, among other fields.

Artificial intelligence has no strict definition, as it’s challenging to define AI in a way that won’t become outdated quickly. Typically, we use the term “AI” to describe algorithms that can perform tasks requiring human intelligence, such as visual perception, speech recognition, decision-making, and natural language processing. In other words, AI is a mathematical algorithm that learns from data we provide.

The benefits of AI in healthcare are numerous. One of the most significant advantages is enhanced accuracy, leading to more precise diagnoses and treatment plans. AI algorithms can improve efficiency by automating many routine tasks, relieving healthcare professionals to focus on more complex and personalised patient care. AI can, therefore, help healthcare providers save time and money.

2 Real Examples of AI Implementations in Medical Area in Poland

Poland has embraced the use of innovative technologies in healthcare. According to a report created in 2022 by the Polish e-Health Agency, 6.6% of hospitals, and 2.5% of all medical entities, use AI solutions. The vast majority of them use AI for CT and MRI analysis.² When it comes to telemedicine, it is used by 32.4% of hospitals and 25.5% of all medical entities.

Some new technologies in healthcare are reimbursed by the National Health Fund, however it’s not a common standard. One of the examples is the system of continuous glucose monitoring by scanning and measuring the concentration of glucose in the intra-tissue fluid using a sensor placed on the arm and sending the result to a smartphone application (or a dedicated reader). It tracks the level and variability of glycemia and alerts about dangerous hypoglycemics by displaying trend arrows.

Few AI projects are funded directly by the Ministry of Health. During the COVID-19 pandemic pilot project on electronic stethoscopes that uses AI algorithms for sound interpretation was implemented in Polish healthcare. There are also public grants for hospitals and technological companies that create and validate AI algorithms, provided by the National Centre for Research and Development and Medical Research Agency.

AI is also implemented and successfully used in medical establishments throughout the country on their individual initiatives. For example, the Radom Oncology Center is one the first such institutions in Poland that has reached for AI in imaging diagnostics.³ Next, the hospital in Ostróda is implementing a voicebot system which is supposed to facilitate the operation of the facility. “Odrodzenie” hospital in Zakopane uses AI solutions for histological specimen analysis and uses hospital budget as a funding source.⁴ Some projects are also funded by European Funds, as with project implemented in National Cardiology Institute on AI solutions for non-invasive myocardial infarction diagnose or Copernicus Hospital project that was focused on finding kidney tumours in CT scans by AI algorithm.⁴

The Polish healthcare startup sector is experiencing rapid growth. According to the report entitled Top Disruptors in Healthcare 2022, third edition (a document prepared annually by the AI in Health Coalition), 47% of startups are developing AI solutions for medicine.⁵ In following editions, more than 60% of startups were involved in AI for healthcare. Almost 30% of startups have solutions certified with a CE mark, and nearly 90% have their product at the stage of at least MVP (Minimum Viable Product—a product version that has enough functionality to meet the expectations of the first customers and provide feedback for further product development).⁵ This demonstrates that Poland is an attractive location for the creation of new technologies.

3 Regulations and Data Poverty Trap

Thanks to the activity of the already mentioned AI in Health Coalition, Poland is also one of the first countries to introduce guidelines on AI implementation in medical entities. According to the declarations provided at the official website⁶: “AI in Health Coalition aims to shape policy for the development of artificial intelligence in the Polish health care system.” Their goal is to create an environment that enables rapid and widespread use by the Polish health care system of the latest advances in AI, as they believe that AI solutions in health care should respect the central role of the medical professional in patient care and instil confidence in the patient. More details can be found in the AI in Health Coalition Manifesto. An important document, i.e. White Paper on AI in clinical practice was co-created by AI in Health Coalition and representatives of Ministry of Health, e-Health Agency, Medical Research Agency, Patients Ombudsman Office, Polish Hospital Federation, National Medical Chamber and other important healthcare organisations.⁷ The White Paper answers common questions about AI implementation in medicine, such as: May AI diagnose by itself? Should patients be always informed about AI diagnosis? Should we ask patients for consent for using AI? May AI decide about the admission of a patient? And who is responsible for AI diagnosis?

However, despite resolving the aforementioned uncertainties, we still face many regulatory challenges. AI algorithms used for diagnosis and treatment are considered “high-risk” AI algorithms, so we have to be sure their sensitivity and specificity are comparable to currently used diagnostic and therapeutic methods—particularly in regard to the effectiveness of physician’ decision. There is a wide range of literature that discusses issues related to AI bias, explainability, and other challenges, however one of them is directly related to transitional economies—that is access to medical data.

To create effective and safe AI algorithms in healthcare, we have to use vast amounts of good quality data. We should also train algorithms on data that comes from patients similar to those who will be treated and diagnosed by a particular algorithm. In other words, if we want to use AI algorithms for European patients, this algorithm should be trained mostly on European patients’ data. An algorithm trained on data, e.g., Asian patients could have significantly lower sensitivity and specificity when used on patients from Europe or the US. This means that countries that cannot provide access to medical data for AI algorithms creation could fall into a “data poverty” trap⁸ and therefore could not develop AI in medicine, even if they are able to buy existing solutions created in other parts of the world. It’s crucial to understand that this may be a gap that we cannot overcome with money, therefore we have to act on preventing the emergence of this gap.

On the other hand, there is a strong need to protect personal data, especially such sensitive data as medical data. Decisions of using data by third parties must be based on data subject consent. We have to secure access to medical data that will be fair, democratic, decentralised, patient-centric—and scalable.

In Poland medical data sharing model is being developed based on Data Governance Act.⁹ The same way patients donate their blood or marrow, they can consent for the usage of their data for research and development purposes. The Polish Donate Your Data Foundation is a legal controller of patients’ data, and based on patients’ consent it can collect, anonymize, and distribute data to the third parties.¹⁰ Thanks to this model access to data is democratised without compromising patients’ right to privacy.

In summary, several implementations of innovative solutions in healthcare were succeeded and already functioning within institutions that support the development of this area. Nevertheless, difficulties still arised. In the following paragraphs, some examples of difficulties because of digital divide remain a challenge for Poland is presented.

4 Basic Challenges in Data Generation

The Polish healthcare system generates an immense amount of data. For instance, in the in-hospital setting, from admission to discharge, the gathered data concerns all the necessary bureaucracy, i.e., patients’ written consents, but more importantly, medical data. These data constitute priceless material for all socioeconomic and scientific analyses. However, several general barriers interrupt its wider analytical use.

The most crucial obstacle is still widespread keeping medical records in paper form, making its analysis more time-consuming. These should be gradually improving, as some legislation that pushes electronic medical documentation (EDM) has been passed in Poland.¹¹ Furthermore, even if the EDM is implemented, there are various software providers for specific recipients. In other words, each hospital or outpatient clinic may use individual single software. Systems usually do not communicate with each other—the medical data cannot be easily transferred between the systems; thus, the integration is also difficult.

There were attempts to unify the data gathering—since December 2019, all health-providing institutions in Poland have been obliged to upload the information about medical events to the government-administered platform—P1.¹² Physicians cannot prescribe the drugs, and sick leaves omitting the P1. The platform is a promising step, but it is insufficient. The analytically most valuable, precise medical data, including biochemical parameters or clinical descriptions of the patients, is not gathered in P1. This is a significant shortcoming of the platform, considerably reducing the value of the data for scientific or analytical purposes.

Moreover, the uploaded information is not free from bias, e.g., the uploaded ICD-10 codes, which should describe the patient’s primary diagnosis, frequently do not reflect the actual medical condition. This is a common practice to slightly modify the ICD-10 code to increase the amount of received money from the public payer. Some ICD-10 diagnoses and ICD-9 procedures are being paid better than others, tempting the health institutions managers to upgrade the reported diagnosis.

Given all the problems mentioned above, telemedicine has not had a favourable environment for further development. All of the links in the healthcare chain function in isolation, with no access to the medical data generated by its different parts. The development of the structured, universal informatics system for medical data collection and analysis has merely been initiated.

The issue is highlighted when the available medical data is supposed to be prepared for artificial intelligence analyses. In our institution—Institute of Heart Disease, Wroclaw Medical University—we performed a series of experiments that included unsupervised machine learning techniques called clustering,¹³ to analyse the heterogeneity of the acute heart failure (AHF) population. Two AHF registries gathered in 2010–2012 and 2016–2017 had to be prepared to apply the algorithms. The collected data, although very similar content, was stored and described differently—the columns in the excel files had distinct names and formats. The first task which had to be performed was the unification of coding. Then the files were merged. Furthermore, the data included many outliers and missing data—it was manually put from patient documentation to the electronic records—therefore, it was prone to mistakes. These limitations were not sustainable for the machine learning analyses—the data had to be manually curated. Implementing more advanced database systems for medical data gathering, which, e.g., enable the validation of the input data, would make it more convenient in further stages of data analysis. Eventually, once the data was manually pre-processed and prepared for the ML algorithms, the calculations were relatively easy to perform, which resulted in 2 full-text scientific articles.¹⁴

5 Barriers in the Field of AI-Based Medical Images Analysis

As mentioned above, the use of AI -based systems in medical imaging is perhaps one of the most promising achievements in healthcare. Nowadays, several AI-based models show the ability to generate meaningful information from complex images such as angiograms, computed tomography scans or magnetic resonance images. It is also known that such solutions can be approved by institutions such as Food and Drug Administration (see: an example with diabetic retinopathy detection¹⁵). This approach creates great opportunities in healthcare. However, there are also numerous barriers, especially in countries that are not leaders in this area.

One of the major barriers is that most of the advanced software requires dedicated, high-performance hardware. These AI-models need a significant amount of computing power for processing complex datasets. Several programs are created and tested on specific components such as graphic cards. In consequence, to use these programs only recommended hardware should be installed. It concerns especially models in the early stage of development, with a limited number of compatible components. Another limitation is combining AI technologies with existing imaging systems in hospitals, which usually cause a lot of challenges. For instance, real-time analysis of angiograms during diagnostic procedures requires a connection between modern software and an angiographic system. Frequently it is associated with the necessity to redevelop the whole existing diagnostic laboratories. It underlines the necessity to design new IT infrastructure suitable for the expandability of advanced tools.

In the Polish healthcare system implementation of AI models analysing medical images is commonly limited to offline analysis. This is due to a couple of obstacles. In the majority of centres, all medical imaging data are not collected in a systematic, structured way. AI analysis of medical records need standardised and correctly labelled images. Frequently, a group of enthusiasts manually search databases to find the appropriate projection of images, formulate it and send it further to external professionals. In the end, the process is time-consuming and exposed to bias during gathering data. It also reduces the possibility to analyse a large amount of data. In a wider perspective, it limits the chances to train algorithms on the major datasets. It creates the need to develop transparent methods for data collection, for instance, compatible with commonly utilised software managing medical images. Furthermore, the developed methodology should be focused on research purposes, selecting data based on particular features. From a technological point of view, the lack of automated, prepared datasets and the possibility of data transfer, lead to discouragement in several centres to take the first steps in AI implementation.

6 Limitations Resulting from Educational Reasons

As mentioned in documents published by the WHO^16,17 in recent years Digital Health has turned out to be an extremely important trend in the development of medicine. Partially this resulted from the pandemic, when the WHO supported the international mobilisation aimed at strengthening collaboration related to digital health interventions, addressing such as challenges as pandemic management. The G20 Digital Health Taskforce highlighted the value of sharing resources, utilities at the national and international levels for contemporary society. Further initiatives of WHO resulted in publishing a Global Strategy on Digital Health 2020–2025,⁴¹ which clearly refers to the fact that digital technologies constitute an integral element of daily life. Therefore, it is worth noting that with extensive exploitation of the potential of digitalization of medicine, we require readiness for the appropriate treatment of unimaginably large data sets. The challenges associated with it are not only related to safe and effective data sharing, but also developing new discoveries and drawing conclusions that could not be drawn before. Such opportunities are provided by modern data analysis technologies, such as artificial intelligence (AI).

Consequently, we can predict that modern analytical techniques, like AI will soon become an integral part of many diagnostic, treatment, research and scientific processes. The optimum use of the AI’s potential in the medical and pharmaceutical sectors involves raising the awareness and engagement of health professionals. Hence, it should definitely appear as a part of education conducted at medical universities.

However, in Poland, as in other European countries, the content of medical studies is prepared and announced at the level of the government regulations^18,19 due to the fact that the profession of a medical doctor is associated with particular rules. For example, many invasive procedures (e.g. blood collection) must be performed by a person with officially confirmed qualifications (for the doctor, for the nurse). Consequently, these rules also affect the areas of education. In other words, the primary objective for any Medical University is to provide educational contents which will be in line with the contents of the official documents.

Importantly, the regulations referring to the Polish MD education, until the end of 2022 (when this publication is created)⁴² did not contain any elements which were directly related to education within any innovative area of medicine (such as digital health, e-health or computational medicine, which may refer to AI). In other words, for example, when Wroclaw Medical University proposed an idea to educate young doctors in the field of artificial intelligence in medicine, it was necessary to propose an entirely new set of criteria, created “de novo” in order to propose the documents describing the course. Such a procedure is necessary while planning the content and characterise educational offer addressed to students of course (e.g., what such a course could give to a person, who attends this training as an element of his or her medical education) and to confirm that such a course can be officially included within the programme of medical education.

The draft of the amendments to official regulations⁴³ related to medical education in Poland, are currently being prepared and will probably be implemented in 2023. Characteristically, the new standards include issues related to broadly understood electronic services (e-services, e.g., e-prescription, e-documentation) including principles of operation of information and communication tools and services in health care (e-health), or issues related to health services using ICT systems. However, the regulations still lack topics related to computational medicine, algorithms or any issues related to, for example, supporting diagnostics or supporting medical decisions, by solutions based on artificial intelligence.

Paradoxically, the White Paper⁷ mentioned above remains the only available document created to prepare the medical community for the changes resulting from the development of artificial intelligence in medicine.

7 Limitations Resulting from Lack of Awareness

Lack of education in the field of innovation in medicine, implemented at the systemic level at the earliest possible stages of medical studies, may be both the cause and the effect of the lack of awareness in the medical community regarding the principles or the possibilities of using AI.

In 2022 we conducted an anonymous survey to assess the beliefs and knowledge of AI in the medical community. The survey was designed to measure the broadly understood beliefs, knowledge and awareness among medical staff (i.e., employees of medical universities, clinics and hospitals, medicine students, etc.) in the field of artificial intelligence in health.

The questions related to beliefs, the person was asked to express his/her opinion about a given belief using a scale starting from: completely disagree and finished by: completely agree. Sample questions in this part of the survey were: “AI-based solutions can support medical staff by allowing them to focus more on the patient”, “AI-based solutions will make patient treatment more effective” or “AI-based solutions will improve the quality of patient treatment”.

Questions about the level of knowledge were structured in order to ask the subject to judge whether a given sentence was correct or not, using a similar scale (“this sentence is wrong” vs “this sentence is correct”). Examples sounded as follows: “Data from hundreds or thousands of patients (subjects) need to be obtained to develop an AI-based algorithm” or “Solutions based on certified AI algorithms are safe and dependable”.

At the end of the questionnaire, there were questions about the surveyed person (age, sex, education, with only people practising a medical profession or studying at a medical university to choose from).

The survey was available online for approximately 8 months. The link to it was intensively promoted both on-line (the website of the project, the website of the Ai-in-Health Coalition and even at the and even on a subpage of the website of the Polish Chancellery of the Prime Minister, as part of the initiative of the health working group) and on leaflets.

The survey and its results will be the subject of an extensive scientific publication, therefore in this chapter we will only provide a general summary of the observations.

There were over 800 visits to the survey’s website. However, the most surprising and disturbing finding from our research was that only few respondents actually solved the entire survey. Despite full anonymity and the relatively neutral nature of the questions, the respondents did not want to demonstrate their knowledge (or lack of knowledge) or beliefs. We had only 130 surveys which were completed and 691 which were solved only partially.

Beliefs were assessed positively, most of the responses oscillated around full agreement or slightly below full agreement. The respondents appreciate the benefits of AI, such as accelerating healing processes or improving the quality of medical services.

The answers regarding the lack of legal regulations regarding the use of AI in the diagnostic process were definitely diverse—the respondents either rather agreed with it or marked the answer from the middle of the scale, which suggested that they did not know whether such regulations exist.

The vast majority of respondents, who were medical doctors or students from the medical faculty, did not agree that “AI-based solutions can replace physicians” however there were single answers which allowed such possibility.

In a question where we asked the respondents to rate their own level of knowledge in the area of AI using a 10-point scale (with 1 for have no knowledge and 10 for an expert) most people rated themselves at 5, i.e., half of the scale. However, further questions where it was verified whether the respondents knew basic terms in the field of AI, such as decision trees, neural networks clustering or supervised learning showed that these concepts are still not well identified, as the responses were very diverse.

To sum up, there is a need for intensive educational activities among doctors aimed at increasing the awareness and understanding of how AI can support medicine. Such education should start as early as possible, preferably during medical studies. Unfortunately, this task has not yet been fully implemented and it is difficult to assess how much time it will take to prepare for educating in this area at the system level. One thing is certain—without education, the use of AI in medicine will be definitely limited, which will be a big obstacle in keeping up with the dynamic development of these issues all around the modern world.

8 Challenges Related to Biobanking in the Era of Big Data and Digitalization

8.1 Types of Data Stored in Biobanks

Nowadays, biobanking has become an increasingly visible field, without which reliable research in the area of drug discovery and development or new medical treatment implementation would be impossible. That is why biobanks shall be recognized as professional tools for innovative research performance. Generally, biobanks are associated with places, where the responsibility for professional preparation of biological material for intended purpose is held. Nevertheless, it should be also emphasised that biobanks also act primarily on data, which are a critical element of all operations. Three categories of data can be identified in the biobanks: (1) data from pre-analytics, (2) data directly connected with participant/donor and (3) data resulting from the obligation to safeguard participant/donor/patent’s rights fulfilment.

The first group of data is mainly generated by biobank. Their quality and the amount strictly depend on the biobank, specific requirements from implemented standard(s) or particular agreement conditions between biobank and partner (i.e., Pharma/Biotech co-operation, where the specification is precisely detailed). Following technological related processes documentation data can be included here:

• collection (date/hour, staff traceability, type of equipment and materials used in the process, adverse event/incidental findings)

• transport (start/finish time, temperature registries, deviations)

• qualification and biological material reception (volume, sample coding, deviations such as lipaemia, haemolysis, etc.)

• processing (time, materials and reagents used in the process, process parameters, aliquoting and the amounts of aliquots)

• storage (traceability, location, storage conditions).

For all indicated processes only some of the examples have been shown. Despite data from technological processes also information from auxiliary procedures shall be collected such as.: infrastructure supervision, including validation, calibration and internal checking, environmental conditions, personnel responsibility for dedicated processes, qualifications and competences requirements, internal/external audits, data from service/materials providers. Quality control (QC) data also provides a relevant number of records. Regarding to the international ISO standard 20387:2018 dedicated for biobanks, QC data shall be derived from biological material, associated data and processes (clause 7.8: Quality control of biological material and associated data).²⁰

The second and third group of data which can be identified in biobanks is related to the participant/sample donor. Among all data related to participants taken part in the study, it is possible to perform advanced research and development work that meet the assumptions for Evidence-Based Medicine.

Among the data related to the study participant, it is possible to distinguish such data, thanks to which it is possible to carry out advanced research and development work, and thus to meet the assumptions for Evidence-Based Medicine.

The following data can be included such as diagnostics data, treatment information, disease severity status, pathology data, demographics, case history, any questionnaire data, recurrence/follow ups data. Image data (CT, MRI, PET-CT, X-ray, ultrasonography, histopathological results) are a significant part of the data. All data in biobanks are usually stored in two formats: (1) paper (informed consent, reports on retrieval, transport, qualification, processing, sometimes surveys) and electronic (e. g. surveys, test results, medical data). As for data storage systems, the most common are excel files, databases (on-premises or in the cloud), sometimes allocating access to resources in another location. However, it should be noted that any paper document can be converted to an electronic version, which is then backed up. Some biobanks use dedicated systems, which consist of many interconnected and interdependent modules. As a result, the biobank is able not only to store data, but also to manage it in a complete way at every stage of biobanking and in every area supporting the biobanking process (e. g. storage, disposal, reporting).

Biobanks use data primarily from medical procedures or scientific research involving humans. Due to the generation of a broad data repository during standard patient treatment, clinical trials or population cohort study, it is possible to conduct and apply an innovative approach to standardisation of methods supported by evidence published in systematic reviews, randomised meta-analyses or observational studies. Critical data that should always be present in biobanks are informed consent forms with exclusions and consent to data processing resulting from the GDPR.

8.2 Data Collection, Storage and their Usage. Trust in Biobanking in Terms of National Solutions to Overcome Digital Divide Based on Development and Availability-Regulations and Standards

The vast majority of data collected and processed in biobanks will be sensitive, personal/identifying data and will therefore be subject to the requirements of the GDPR for EU countries. This is particularly important to ensure maximum protection for the subject. The functioning of biobanks in some countries has its own regulations in the form of statutory regulations (Estonia-Human Genes Research Act,²¹ Hungary- National legislative act on the protection of human genetic data, on the human genetic studies, on research and on the operation of the biobanks,²² Iceland-Acts on Biobanks no.110²³). Unfortunately, Poland has not yet developed a biobanking law, despite the efforts of the BBMRI. pl Consortium, which was established to achieve the common objectives of the European BBMRI-ERIC infrastructure policy in the field of biobanking. Thus, biobanking is not regulated by law in Poland. Biobanks in the scope of their activities are guided by those prepared by the consortium BBMRI. pl as part of the project “Creating a Polish Biobanking Network based on BBMRI-ERIC infrastructure” or Code of Conduct of Processing of Personal Data for Scientific Purposes by Biobanks in Poland²⁴ and Quality Standards for Polish Biobanks (QSPB),²⁵ Manual of Biobank Quality Management (MBQM).²⁶ Nevertheless, it is important to underline that standards and codes are voluntary and there is no legal obligation to apply them outside the members of the Polish Biobanking Network. Furthermore, their usage allows them to function according to the guidelines adopted by the environment.²⁷

Standards refer to the safety and security procedures for biobank operations (Chap. 15 from QSBP^25,26). The requirements for basic methods of securing IT infrastructure including biobank system sample management, where digital data are stored were also underlined. It directly influences the donations willingness and donor’s trust, that his/her samples and associated data will be used properly and stored in a secured manner^20,21.

The QSPB Standards, which have been prepared and implemented within Polish biobanking activities, also pay special attention to ethical and legal aspects (Chap. 5 from QSBP^25,26) where principal recommendations for impartiality, confidentiality, responsibility and respect for autonomy are raised. The detailed information regarding rules for informed consent preparation, communication with participants and bioethics committee role together with sharing of biological material and data for non-commercial and commercial usage of biological samples/data are presented. Moreover, ownership issues and privacy protection are described. Biobanking that complies with relevant ethical and legal standards may increase the quality and public trust in science and research ^20,21.

According to the requirements which have been described in QSPB, in 2017–2020 an audit process has been carefully organised and performed in Polish Biobanking Network.^28,29 The results clearly showed that 85% from Polish Biobanking Network entities have fulfilled the requirements dedicated to ELSI aspects, while only 40% of requirements were achieved in the Safety/Security area. The excellent results (100%) have been accomplished by BBMRI.pl Consortium biobanks. It was also interesting that the QMS implementation was extremely significant if the comparison to the biobanks without any system was done, only in the ELSI area the impact has not been noticed.²⁴

Trust is one of the factors influencing the willingness of donors to donate biological material to biobanks. Therefore, BBMRI.pl went ahead of expectations. And an information security audit program was launched. Each biobank associated in the Polish Biobanking Network (both members and observers) which went through the program (voluntary participation) received report, recommendation and was offered support. What is more important the summary of anonymized reports revealed most frequent shortcomings, and areas that need more attention. These conclusions were reflected in Chap. 15 from QSBP^25,26 of the second version of The QSPB^25,26 and covered physical security, backup, cloud services and review of basic security mechanisms. Also, in terms of ICT technology, the biobank environment is strongly differentiated from very mature entities that can afford proper IT support, either within their own resources or the institutions they are subordinated to. Mature biobanks have implemented LIMS/BIMS systems, established, not necessarily formalised, methodology for dealing with research data. Less advanced or newly established ones are working on implementing appropriate mechanisms both in the areas of security and data processing.

At the network level, there were also prepared dedicated tools to facilitate sample information management—a BIMS class system as well as solutions for samples or data discovery.

It is important to note that all research regarding biological material and associated data are based on the same principles as any trial where human is engaged.^30,31 For biobanking trust and donor protection three step system in contemporary ethical assessment shall be kept, where independent ethical committee, international/national codes of ethics and finally regulations are present.³² Polish Bioethics Committees are dedicated to giving opinions for medical experiments and every project, where biological material and associated data are planned to be used. Presented regulations are contained in updated law act on doctor and dentist. It is required that each project of any experiment involving humans or biological material/data to be submitted to an independent bioethical committee for approval. It is also forbidden to perform any medical experiment before the positive opinion. The law is far too narrow and excludes all experts in the field of research where biological material/data are used. From 2021 in Poland only doctors and dentists can be accepted as head/manager of specific projects. This discriminates all researchers from other fields and disciplines (i.e., biologists, biotechnologists, bioinformaticians, data scientists) who are qualified to lead projects involving biological material/data. Such investigators often have a much higher level of competence in the laboratory techniques, IT and biostatistical methods used in scientific analysis and advanced research. This law unfortunately effectively has closed the way for excellent researchers to lead such projects.

Data sharing in national biobanking is essentially based on the patient’s informed consent, in which he/she determines what may be made available and on what terms. Provisions that do not explicitly refer to biobanks themselves, but may support them in terms of their activities are, for example, the Act on Patients’ Rights and the Patient’s Ombudsman (Dz.U.2022.1876). In Chapter 7 art. 26 p. 4 the legislator indicates the method of making medical records available: Medical records may also be made available to a university or research institute for use for scientific purposes, without disclosing the name and other data allowing identification of the person to whom the records relate.³³ This record makes it possible to conduct scientific activities based on anonymized medical data. The Act also specifies the method of making electronic medical records available for control purposes (legal act from 27 Aug. 2004 on publicly financed health care services Dz.U.2022.2561³⁴), referring to the data format records specified in separate regulations. These, in turn, refer to another law and Art. 18. Legal act from 17 Feb. 2005: Computerisation of the activities of entities performing public tasks Dz.U.2023.57.³⁵ The Act points out that at the request of the Minister responsible for computerisation, minimum requirements for ICT systems and public registers will be defined by means of a regulation. These actions shall ensure the consistency of the operation of ICT systems used for the performance of public tasks by specifying at least the specifications of the data formats, communication and encryption protocols to be used in the interface software, while maintaining the possibility of using those specifications free of charge; the efficient and secure exchange of information in electronic form between public bodies and also between public bodies and authorities of other States or international organisations. In addition, it is intended to ensure that the National Interoperability Framework meets the requirements regarding semantic, organisational and technological interoperability, taking into account the principle of equal treatment of different IT solutions, Polish Standards and other standardisation documents approved by the national standardisation body.

As a result of implementation of DIRECTIVE (EU) 2019/1024 OF THE EUROPEAN PARLIAMENT AND OF THE COUNCIL of 20 June 2019 on open data and the re-use of public sector information ³⁶ the Member States were obligated to enact national implementations in this field. Consequently, in 2021 the law “on open data and re-use of public sector information”³⁷ was announced. The introduced law does not change much in the field of sharing medical data. However, it introduces some changes to the reuse of scientific data. Firstly, by introducing the definition of research data and the obligation to share them if they were generated with public funds. At the same time, the Minister of Education and Science was indicated as the body responsible for the implementation by preparation of detailed guidelines in the form of policy in this area—at the time of writing this text, work is still in progress. Another important issue arising from the new law is the appointment of the minister competent for computerization, which is developing the “Data Opening Program”—a continuation of the previous responsibility of the Ministry of Digitization, which has been dissolved in the meantime. One of the six pillars of the Program is: “Stimulating the market for re-use of cultural resources and scientific data”. As part of this document, it was recommended that scientific data should be made available through thematic repositories, which have been intensively developed at universities in recent years—as examples Polish Platform of Medical Research.³⁸ On the other hand, in the document there was also identified the factors negatively affecting the use of data—the multitude of available registers. This may lead to a situation where scientific data repositories, which have been intensively developed in recent years, apply individual standards, which in turn will hinder interoperability and data exchange between individual systems. It will ultimately translate into difficulties in searching for data—the need to search each of the repositories separately. Among others further obstacles were indicated legal—lack of sufficient legislation on data sharing, competency—the lack of properly qualified personnel. The document also sanctions the participation of Poland in the European Open Science Cloud (“EOSC”). The report on the implementation of the program for 2021 indicates the creation of an association of six universities that will represent Poland in the EOSC and become responsible for the implementation in the country.

Another road map is "HEALTHY FUTURE STRATEGIC FRAMEWORK FOR THE DEVELOPMENT OF THE HEALTH CARE SYSTEM FOR 2021-2027, WITH A PERSPECTIVE UNTIL 2030." ³⁹ issued by the Ministry of Health. In terms of Digital Health, the document focuses mainly on e-services provided to patients, doctors or medical entities, technical equipment of facilities with adequate IT infrastructure. Building medical knowledge, developing methods of collecting and sharing data is not a priority. Based on the strategic documents presented earlier, a silo landscape emerges at the ministerial level—documents and guidelines are created in separate departments, probably without any attempt to coordinate work. According to the conclusions of the report “Digital Health Implementation approach to Pandemic Management” [Digital Health Implementation approach to Pandemic Management G20], this is not an approach that may herald failure in implementing an effective digital health system. Which requires coherent coordination of work at the national level. The WHO issued similar recommendations [Global strategy on digital health 2020–2025, WHO].

It is also important to note that in 2019 the Polish Medical Research Agency (MRA) will be established. The main scope of the Agency is to increase the potential of non-commercial clinical research especially for new treatment methods in oncology, cardiology and rare diseases areas. The MRA has been created by law (Act of 21/02/2019 about Medical Research Agency, Journal of Laws No. 447⁴⁰). Noteworthy is the fact that from 2022 The MRA has implemented the biobanking requirement activity for all clinical trials which are sponsored by the agency, where the blood sample is taken from a patient. Moreover, the biobanking must be performed according to the Quality Standards for Polish Biobanks (QSPB).^20,21 Additionally, in 2023 the MRA has announced the call for Digital Medicine Centres, where Biobanks as necessary units are also included and will be financed. However, it is strictly defined that the biobanking process must be carried out within ISO 20387:2018 and QSBP.^20,21 Thus, the Polish Biobanking Network membership based on unified and harmonised standards sustainability can assure that the fulfilment of the highest regulation proposed by national or governmental authorities will be achieved.

9 Digitization of Genomics Data in Poland

Sharing of data was the primary domain of Biobanks in Poland. Although biobanks have just begun to emerge in Poland in the beginning of the twenty-first century. However, from the origination they started to play a significant role in the sample and data sharing ecosystem of Polish science. In 2017 the BBMRI-ERIC Polish Node was established as well as BBMRI.pl consortium.⁴¹ The initiative joined 7 entities involved in biobanking. One of the project goals was to implement IT System for samples discovery and data sharing, integration with BBMRI-ERIC directory,⁴² creation of unified BIMS and integration with HIS, setting up a gate for information exchange between biobanks and national registries. Due to insufficient legal framework the last task was not even started. Integration of BIMS and HIS system was performed as a Proof of Concept and paused at this stage. First three tasks were completed. The unified BIMS is designed to support data/information exchange with API interface; there was also developed Central Platform dedicated to performing queries for data collected in central repository as well as in federated ecosystem—created with software installed in biobanks and fed with data directly from local BIMS. The consortium did not only focus on technical aspects and took challenges in changing the ELSI landscape, among others there was proposed the code of conduct based on GDPR art. 40—“Polish code of personal data processing by biobanks in Poland”.⁴³ Currently works of the consortium is held up due to a pause in funding.

Another initiative for creating infrastructure for data sharing is “Polish Genomic Map in open access—digitization of biomolecular resources of the Biobank Lab University of Lodz” project.⁴⁴ The idea is to set up Polish node of Federated European Genome-Phenome Archive (EGA)^45,.46 This infrastructure will be compliant with EU guidelines “as open as possible as closed as necessary”. It provides Public Key Infrastructure (PKI) encryption of datasets submitted to the repository. Access to the data is available under approval by the Data Access Committee (DAC). DAC can be set up by the Principal Investigator or the Institution. The Polish instance will be connected to Central EGA discovery service—providing an interface for researchers looking for data.

Thanks to EU funding, the Digital Poland funding scheme was provided to support ventures in the area of digitization. Beneath there are mentioned some projects that were supported and shared medical data and resources.

Digital Brain—main goal of the project is to digitise and share in an open access collection of brains stored by the biobank of Institute of Psychiatry and Neurology.

OpenCardio—main goal of the project is to disseminate the digitised results of pulmonary embolism diagnostics by creating a specialised digital platform openCARDIO and digitization of science resources on venous thromboembolism.

Medical Data Bank—in this case the project is performed by Lodz University of Technology and Institute of the Polish Mother’s Memorial Hospital in Lodz the aim is to digitise one million histopathology samples and share them with connected medical records.

Although the Digital Poland funding stream is the great source for supporting digitization initiatives there is at least one serious oversight. Program became very popular and plenty of repositories raised a great variety of data. Unfortunately, there weren’t any general standards introduced. So, at the end there is a lot of data available, but the problem is lack of standard communication protocol or API introduced. Therefore, in future there might be problems with data interoperability and/or harmonisation. Also, a single search entry point might be problematic to implement– each repository needs to be queried individually.

The mentioned ventures should be considered as bottom-up initiatives performed by the scientific community. At the time of writing this text there is no systematic approach on a national level that would promote data sharing. There are even not defined any regulations that would oblige scientists or academia/institutions performing scientific research funded from public money for publishing data in open repositories.

One of the main sponsors of Polish science is National Science Center, during 10 years of its existence, it is estimated to have spent about 1,000,000,000 PLN (250,000,000 USD) for granting projects in which Human Biological Material was used—on average 93 projects per year (based on analysis of available projects abstracts on NSC web page—years 2014–2020). At the end of March 2023, in scientific repositories there were deposited 30 data sets—by Polish institutions or regarding Polish population DbGap—22 datasets, EGA—8 datasets. Which indicates a significant gap between the genomic data generated and data available.

On the other hand, there are some top-bottom initiatives that focus on accessing HIS and health registries to combine data and use them in scientific ecosystems or decision supporting tools. These approaches are supported with law regulations or policies established on a national level. At this point these initiatives do not focus on genomic data.

National Cancer Registry set up by Polish Ministry of Health⁴⁷— In the form of a web-based platform that aggregates information about cancer diagnosis, it can be fed with data directly from HIS system or manually by doctors. Data from the registry are then aggregated in a data warehouse and are available through a web interface to scientists or medical personnel.

National registry of Rare Diseases⁴⁸ is the next initiative that is planned to be set up in the near future to provide a transparent system for collecting information from HIS and provide aggregated data for policy creation.

The situation is slowly changing, in 2023 the Medical Research Agency (MRA) has announced the call for Digital Medicine Centers. The idea is to combine Hospitals, Biobanks Academia, entities running non-commercial clinical research, facilities providing genome sequencing services and facilities performing analysis (with proven experience in AI) in one regional node. The idea is to generate and collect for further re-use as much data connected with sample and donor as possible. This also includes genomics data. In the first phase there is planned creation of 10 such regional nodes. In the call there are no strict requirements how the cooperation needs to be performed nor which standards to be used. Actually, there are some recommendations, for instance suggestions to use HL7 FHIR standard for data interchange, qualification of personnel that need to engage into the project realisation. General rule is to set up connectivity between HIS and interface that would allow sharing or discovery samples and related data. There is also a need to provide IT infrastructure that would be used to perform federated analysis. The regional centre was left free to use any technology they found relevant and compatible with already existing infrastructure. In the next phase there is planned setting up a central hub which will coordinate data flows between regional centres and provide discovery services for scientists.

10 Summary

In contemporary Poland, the situation regarding broadly understood digitization, innovation in medicine, development of approaches and solutions based on AI is not particularly imperfect and/or neglected. We can be proud of some impressive implementations. We have both clinical and scientific achievements. There are dynamically developing institutions whose mission is to influence decision-makers so that the development of modern technologies encounters as few obstacles as possible. Despite this, we still struggle with various problems, due to which we remain less developed than the giants of digitization, such as Asian countries, Israel or even Estonia, which is the closest to us geographically, culturally and historically. The barriers result from the condition of the Polish healthcare system, which was problematic even without the context of innovation and digitalization. Consequently, this is reflected in the approach to medical education as well as in the attitude of healthcare professionals towards solutions which are considered new and unknown.

The trend of a significant data increase and collection in biobanks is also becoming more pronounced. Starting from clinical data, anthropometric, diagnostic data, it is worth pointing out that international standards determine the critical amount of data only from pre-analytics/processes. This settled the necessity of professional development of the IT infrastructure in each biobank and its proper supervision and protection by competent personnel. These data are increasingly noticed by the regulator and research funding communities. However, it is important that they are made available and used in a controlled way while maintaining the full range of data quality and according to FAIR principles. The availability of a variety of best practice documents, recommendations, policies and procedures supports the availability and use of biological material and related data, giving a major impetus to the global use of biological resources for scientific research and the exploitation of their potential in the innovative research sector. However, it is equally critical to have a regulatory package that will clearly define the possibility of using data, and how to transfer it securely for R&D purposes, including using it in an efficient and secure AI or Machine Learning research. There is also a need for broad social education in this area in order to objectively present the benefits and risks of the increasing volume of data being generated. Biobanks therefore appear as a key element of research infrastructure not only as institutions managing biological material but also as operators of data repositories. Data from both the health sector and science. In this respect, biobanks can use their skills to build relationships of trust with donors/patients/participants.

All the information presented in this chapter, referring both to the case description itself and presented issues which shall be improved, are summarised as digital divide determined challenges for future actions in the context of policy makers and decision makers and stakeholders (Table 1).

Table 1 Summary of factors related to digital divide