Ethical and Secure Evidence Generation from Regionwide Clinical Data through a Collaborative Environment for Advancing Predictive Care

Ethical and Secure Evidence Generation from Regionwide Clinical Data through a Collaborative Environment for Advancing Predictive Care | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Research Article Ethical and Secure Evidence Generation from Regionwide Clinical Data through a Collaborative Environment for Advancing Predictive Care Dolores Muñoyerro-Muñiz, Román Villegas, Víctor de la Oliva, Alberto Esteban-Medina, and 8 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-5389651/v1 This work is licensed under a CC BY 4.0 License Status: Posted Version 1 posted You are reading this latest preprint version Abstract Background : Data protection is often regarded as a major hurdle in clinical research, especially when handling sensitive patient information. However, by managing sensitive data within a secure environment, it becomes possible to mitigate these challenges and promote evidence generation through clinical data studies. This manuscript introduces a collaborative environment which facilitates the secure and ethical secondary use of clinical data for medical research. The environment combines a comprehensive health population database, integrating real-world data (RWD) from over 15 million patients, with a legal framework for patient privacy and data security and a secure processing environment (SPE) for ethically compliant clinical data research. Methods : The SPE of this collaborative environment is a computational infrastructure (IRWD), located within the healthcare corporative network, that enables large-scale studies essential for real-world evidence (RWE) generation. Within iRWD, diverse studies can be conducted, including treatment efficacy assessments, survival analyses, and the development of predictive models. These studies leverage RWD to perform robust analyses while maintaining compliance with stringent regulatory standards, such as the European Health Data Space (EHDS) and the General Data Protection Regulation (GDPR), which govern data security and patient privacy. Results : One of the key outcomes facilitated by this infrastructure is the systematic development of early endpoint predictors. These predictive algorithms can identify high-risk patients before symptoms emerge, enabling preventive interventions. The approach promotes a shift in healthcare from a reactive model to a preventive one, allowing for early, efficient, and cost-effective treatments that improve patient outcomes. Additionally, the model supports public-private partnerships, generating revenues that sustain this collaborative environment while reinforcing its capacity for continuous healthcare innovation. Conclusions : The integration of clinical big data within a legally compliant and secure framework provided by the SPE offers a sustainable and proactive model for healthcare improvement. The infrastructure supports the development of cost-effective predictive models that are particularly valuable for an aging population, ultimately transforming healthcare delivery into a proactive, data-driven system. By positioning the health system as a central player in knowledge generation and wealth creation, this collaborative initiative sets a new benchmark for data-driven healthcare models across Europe and beyond, highlighting its potential for global adaptation and impact. Real-World Data Real-World Evidence electronic health records predictive medicine secure data processing endpoint predictors European Health Data Space public-private partnerships machine learning data privacy. Figures Figure 1 Figure 2 Figure 3 Background The use of Real-World Data (RWD) in healthcare is revolutionizing the generation of medical knowledge, offering unprecedented opportunities for research, innovation, and improved patient outcomes [1]. RWD, which includes data derived from a variety of sources such as electronic health records (EHRs), minimum basic data sets (inpatients, outpatient mayor surgery, hospital emergencies and medical day hospital), mental health information systems, analytical and image tests, vaccines, renal patients, and pharmacy, reflects the real-time experiences of patients in routine clinical settings. Although the use of RWD in healthcare has its challenges, such as potential biases, incomplete information, and privacy concerns, this data provides a more comprehensive picture of patient populations and treatment effects than traditional clinical trials, which often have strict inclusion criteria that limit the diversity of participants [2]. As such, RWD is instrumental in generating Real-World Evidence (RWE) that can inform clinical decision-making, guide the development of new therapeutic approaches, and support regulatory decisions [3]. The ability to analyse vast datasets in real-world conditions enables researchers to better understand the effectiveness of interventions across diverse patient populations and healthcare environments, ultimately contributing to more personalized and effective medical care [4]. One of the key advantages of RWD is its potential to accelerate the pace of medical innovation. By utilizing data already being generated within the healthcare system, researchers can bypass the lengthy and expensive process of traditional randomized controlled trials (RCTs) in some cases, while still obtaining valuable insights into treatment efficacy and safety [5]. This is especially relevant in the context of rare diseases, where limited patient populations make traditional trials challenging, or in urgent public health crises like the COVID-19 pandemic, where timely insights are essential for guiding treatment strategies [6, 7]. Moreover, RWD is increasingly being used not only for post-marketing surveillance of drugs, allowing for the continuous monitoring of drug safety and efficacy after they have been approved, providing an additional layer of patient protection [8], but also for a variety of studies that range from conventional epidemiologic studies, including cost of treatment, etc. to more sophisticated endpoint predictors involving the use of artificial intelligence [9]. In the European context, the utilization of RWD is becoming increasingly structured and formalized. The European Health Data Space (EHDS) [10], proposed by the European Parliament, aims to create a unified framework for the use of health data across Europe . This legislation seeks to promote the secure and ethical use of health data for research and innovation while ensuring the protection of patient privacy and data security. The EHDS emphasizes the importance of creating trusted research environments where data can be analysed securely, ensuring compliance with the General Data Protection Regulation (GDPR) [11] and safeguarding the rights of individuals. Such a framework is essential for fostering public trust and enabling the responsible use of health data on a large scale. Here we present a leading example of the effective use of RWD within a secure processing environment, fully compliant with the EHDS, provided by the Andalusian Public Health System (SSPA, acronym for “Sistema Sanitario Público Andaluz” in Spanish). The SSPA has developed one of the most comprehensive and ambitious health information systems in Europe, known as the Andalusian Health Population Database (BPS, acronym for “Base Poblacional de Salud” in Spanish). The BPS includes medical records dating back to 2001 for more than 15 million patients, constituting a vast repository of clinical data covering almost the entire population of Andalusia [12]. This data includes diagnoses, laboratory tests, medical imaging, pharmaceutical records, and more, making it an invaluable resource for both clinical and epidemiological research. The integration of BPS with a Secure Processing Environments, such as the iRWD platform [13], offers an invaluable environment for researchers to analyse clinical data while maintaining compliance with GDPR and EHDS regulations, ensuring patient privacy. The potential of RWD to generate new medical knowledge is vast, but its effective use requires the right infrastructure, legal frameworks, and data governance practices. The combination of comprehensive data systems like BPS, secure environments for data analysis, and adherence to regulatory standards such as those proposed by the EHDS is essential for maximizing the benefits of RWD while safeguarding patient rights. The Andalusian Health System exemplifies how this can be achieved, offering a model, open to collaborations, for the integration of healthcare data into the research ecosystem in a way that benefits both patients and the broader medical community. Importantly, the Andalusian model demonstrates how a region-wide, integrated health system can leverage RWD to foster innovation and improve healthcare delivery. By utilizing BPS data within a secure, legally compliant framework, Andalusia has positioned itself at the forefront of the secondary use of healthcare data for research, setting a standard for other regions and countries to follow. Methods A collaborative environment for secure and ethics secondary use of medical data The collaborative infrastructure for secure and ethical secondary use of medical data is structured around three fundamental components that work together to facilitate advanced medical research within the medical data space of Andalusia. First, Andalusia houses an extensive repository of medical data, encompassing a wide range of patient information collected from the health system across the region. This vast pool of data constitutes a rich resource with the potential for conducting large-scale studies, supporting the development of predictive models, and advancing medical knowledge. Second, the use of data from Andalusian patients is governed by legal regulations designed to ensure that patient privacy is protected while enabling the responsible use of information for research purposes. These regulations provide a clear legal framework that balances the need for scientific advancement with ethical considerations surrounding data security and patient consent. Third, the availability of a secure, state-of-the-art secure processing environment where data can safely be analysed. This environment is designed to maintain the integrity of the data, safeguard patient confidentiality, and facilitate complex computational analyses. Together, these three elements—the vast medical data repository, the legal framework, and the secure processing environment infrastructure—create a robust ecosystem that fosters innovation in medical research while ensuring compliance with ethical standards. The Andalusian Health Population Database The SSPA provides service to a population of approximately 8.5 million users, and comprises a total of 55 hospitals and 34 primary care districts with a total of 1505 primary care centres. The SSPA activity encompasses more than 80 million primary and specialized visits, half a million hospital admissions, 5 million of hospital emergencies per year, and all the diagnosis, prescriptions and an ample variety of data derived from these interventions. The complete Health System is fully digitalized by means of the digital system Diraya, where all the data are indexed and referred to the patient. Diraya incrementally dumps all patient data in the Health Population Database (BPS, acronym of “Base Poblacional de Salud” in Spanish) on a monthly basis. The BPS is one of the most ambitious resources of the SSPA. BPS is a health information system that collects clinical data and data on the use of health resources of each person receiving health care in the SSPA, totalling over 15 million patients accumulated since 2001. From the data collected in BPS, it is possible to obtain estimates on health, the behaviour of users in relation to health services and to stratify the population in order to guide the provision of these services. The BPS also enables longitudinal studies to be carried out, the incidence of pathologies to be estimated and projections to be made on the state of health of the population and its resource needs. It also makes it possible to analyse efficiency in the use of resources by health care providers. Since its inception, BPS was conceived with a strong focus on research. In fact, it has been included in the Repository of Innovative Practices in Active and Healthy Aging of the European Commission [14]. This repository aims to collect the most innovative initiatives of European partners, to facilitate the dissemination of knowledge and mobilize resources to ensure the implementation of these initiatives at European level. The structure of BPS is defined in a reference publication [12], which describes the development of this information system that connects data from multiple Electronic Health Records (EHR) to improve assistance to patients, health services administration, management, evaluation, and inspection, as well as public health and research. BPS connects pseudonymized data from nearly any individual of the whole Andalusia population. In fact, BPS has an estimated coverage of 99% of the Andalusian population [12]. The data are sourced from different SSPA information systems including: the Andalusian Public Health System patient database, DIRAYA electronic medical records, the minimum basic data sets (inpatients, outpatient major surgery, hospital emergencies and medical day hospital), mental health information systems, analytical and image tests, vaccines, renal patients, and pharmacy, among others. In order to have the data as structured as possible, an automatic coder developed in-house for hospital emergency and primary care electronic medical records [15] is used to code clinical diagnoses into ICD10. Also 80 chronic pathologies were identified and coded. To illustrate the dimension of BPS it is worth mentioning some figures (see Table 1) about the data stored, such as the 874 million diagnoses, the 2,428 million analytical tests, or the 7,000 million medical images, to cite just a few examples. This volume makes BPS one of the largest medical RWD repositories in the world. Table 1. BPS in figures. Data Number (millions) Users (2001- September 2024) 15.8 Medical diagnoses (90% automatic coding) 874.1 Nursing diagnoses 51.3 Analytical test results 2,428 PACS images 7,000 Vaccination events 79 Prosthesis and Implant records 0.7 Temporary Incapacity (TI) Processes 11.2 Vital signs (weight, height, BMI, blood pressure) 83.5 Hospital stays 3.5 Functional evaluations 4.2 Cognitive assessments 2.7 Procedure for the access to medical data in Andalusia The most recent regulation for the use of medical data for research purposes was issued the 4 th December, 2021, in the Joint Resolution 1/2021 of the General Secretariat for Research, Development and Innovation in Health of the Regional Ministry of Health and Families and the Management Directorate of the Andalusian Health Service [16]. This innovative resolution anticipated most of the strategy and procedures for access to medical data which has further been described in the recent European Parliament legislative resolution of 24 April 2024 in the proposal for a Regulation of the European Parliament and of the Council on the EHDS [10]. The Joint Resolution 1/2021 defines a Health Data Access Body (Data Access Committee, DAC) responsible for granting access to electronic health data for secondary use. In order to evaluate the convenience of granting the access to the data requested, the DAC requires: i) the report of the study for which the data is requested, ii) the permission of the Coordinating Committee on Biomedical Research Ethics of Andalucía (CCEIBA) [17], iii) the Impact Assessment on Data Protection (IADP) [18] document, subject to the GDPR and taking into account the Horizon 2020 Programme Guidance How to complete your ethics self-assessment [19] and, iv) signed commitment of the principal researcher in which he/she undertakes not to redistribute data, not to attempt to re-identify individuals in the dataset and removing the dataset once the study has finished. In essence, the data access regulation described is completely compliant with the EHDS regulations and would fit perfectly in the proposed future federated structure of the EHDS. The Infrastructure iRWD, a Secure Processing Environment The third key piece of this collaborative model that facilitates the secondary use of medical RWD for the generation of new medical knowledge is a Secure Processing Environment (SPE). This SPE consists of a computational infrastructure (iRWD), funded within the scope of the Andalusian Plan for Research, Development and Innovation [13]. This infrastructure is located within the SSPA corporate network and has specifically designed for the secure analysis of data protected by the GDPR, being compliant with the definition of SPE as described in Article 50 of the Resolution of the European Parliament of 24 April 2024 on the proposal for a Regulation of the European Parliament and of the Council on the EHDS [10]. Secure data management in iRWD This SPE follows a data management procedure that avoids any aspect that could compromise privacy, as described in the joint Resolution 1/2021 of the General Secretariat for Research, Development and Innovation in Health of the Regional Ministry of Health and Families and the Management Directorate of the Andalusian Health Service [16], also in accordance with the recent European Parliament legislative resolution of 24 April 2024 on the proposal for a Regulation of the European Parliament and of the Council on the European Health Data Space [10]. Since the iRWD is located within the SSPA corporate network and is operated by personnel of the Fundation Progress and Health, which belongs to the health system, data never leave the secure environment of the health system and are managed by trusted personnel from the health system. These two aspects are crucial for an IADP which minimizes the relative risk for the data used in the study [18]. The IADP is a document consisting of a description of the data life cycle, detailing the activities to be carried out, the specific data to be processed and the people and technologies involved, both for the data acquisition process and for its storage, processing, transfer to third parties and final destruction. The IADP also includes an analysis of the necessity and proportionality of the treatment, including the report on the legitimacy of the use of databases without informed consent in health research in certain circumstances, as well as the legitimacy of the use of the BPS as a research infrastructure. And, finally, the document identifies and assesses the probability and impact derived from the possibility of a risk materializing, with the objective of establishing preventive, corrective and mitigating actions to minimize risk exposure. The final section of conclusions specifies the potential hazards, their inherent and residual risk, and mitigation measures that could be implemented. By using the iRWD, an action plan is not required [18]. Summarizing, the data management procedure (described in the IADP as the life cycle of the data) is as follows: i) the iRWD work team request the data to BPS with the corresponding approval of the DAC, ii) the BPS team extract the data and pseudonymize them, iii) the BPS transfer the pseudonymized data to the iRWD, iv) the iRWD work team carries out the analysis of the data as described in the study report, v) when the study is done the data are removed from the iRWD infrastructure. The use of pseudonymized data is an important feature specific to our collaborative model approach (Figure 1 left part), as it allows patient re-identification when necessary, provided that the study has approval from the ethics committee. This approach contrasts with conventional data analysis frameworks (Figure 1, right side), which utilize anonymized data that do not permit any form of individual identification. As a result, studies requiring patient-specific insights for their benefit cannot be conducted under a fully anonymized data framework. Sample size An attractive aspect of RWD region-wide studies is that no sample size justification is necessary. In conventional studies using hospital databases, sample size estimation is required to define the minimal number of individuals which constitute a significative representation of the whole population and, consequently, can be used to respond to the objectives of the study with adequate power and precision so that the results can be generalized to the entire population. Region-wide studies include all patients who meet the selection criteria, i.e., the sample is directly the entire target population. Hardware and software resources For the sake of reproducibility and explainability, all the software used in the studies carried out in the iRWD SPE is open. For data processing, we rely on Python [20] and its extensive ecosystem of libraries, such as Pandas [21] and NumPy [22]. This allows us to efficiently clean, transform, and explore our data, setting the stage for subsequent analysis. When it comes to statistical analysis, we utilize mainly Python libraries like Numpy , Scipy [23] and statsmodels [24], whereas R [25] and Bioconductor [26] packages are used for some advanced analysis. AI-driven models, including machine learning and deep learning, are backed by the core libraries of the Python scientific distribution, like NumPy and SciPy , as well as specialized tools like scikit-learn [27], and TensorFlow [28]. To ensure computational reproducibility, we employ a multi-layered approach. Data versioning is controlled by the BPS while software versioning is achieved using locally managed gitlab servers. Conda environments, drawing from the conda-forge [29] and Bioconda [30] channels, provide project-specific software dependencies, minimizing conflicts and ensuring consistent results. Additionally, we encapsulate entire analysis environments within Docker containers [31], further enhancing reproducibility across different computing platforms. Finally, interactive analyses are facilitated through securely-served Jupyter notebooks [32], leveraging the aforementioned reproducibility infrastructure for computation. The hardware infrastructure supporting the SPE is specifically designed to meet the rigorous computational demands of modern clinical Big Data research. A general-purpose computing cluster provides significant computational capacity, featuring 1,832 cores, 13,568 GB of RAM, and 993 TB of storage, making it well-suited for data-intensive studies. For deep learning applications, a specialized GPU server cluster is available to handle the high-performance requirements of AI-driven research. This cluster includes 192 CPU cores, 4,864 GB of RAM, and 18 high-performance GPUs. Each server is equipped with 1 TB of local storage and access to a shared 150 TB network storage system, facilitating collaborative research and efficient data management. This infrastructure ensures that even the most demanding computational tasks can be executed efficiently. Results Activity of the iRWD infrastructure and analysis portfolio Multiple type of studies can be carried out in large RWD repositories. Figure 2 summarizes the most common studies already performed, ongoing or under consideration in the iRWD infrastructure. Clockwise from the top, the most common studies currently requested by pharma companies are epidemiological studies of prevalence or incidence of diseases (a total of 60%), followed by more detailed studies of cost of disease or interventions (approximately 20%). Survival studies [37] are fundamental in medical research as they provide crucial insights into patient prognosis, the efficacy of treatments, and the natural course of diseases. These studies allow for the estimation of survival probabilities over time, identifying factors that may influence patient outcomes, such as comorbidities, demographics, and treatment modalities. By analysing survival data, researchers can also identify potential risk factors and guide clinical decision-making. Since many patients will present concomitant drug treatments, it is relatively easy to extend the concept of survival studies to repurposing studies by modelling the effect of the concomitant drugs in the outcome of the patient. Similarly, specific treatments or interventions can be assessed, taking into consideration all the possible confusion variables, providing valuable information on the efficacy of these. Actually, in the case of post-marketing surveillance of drugs, they are Phase IV studies [8] of clinical trials. As previously mentioned, the secure data management environment used here involves the utilisation of pseudonymized data, which makes possible patient re-identification, if required by the study and authorised by the ethics committee. Studies aiming to detect undiagnosed patients of rare diseases affected by the well-known diagnostic odyssey [38] or to discover undiagnosed patients of an infectious condition, like Hepatitis C, HIV, etc., for eradication programs [39], or other similar ones, are ultimately oriented to the identification of individuals, which is possible with pseudonymized data but not with anonymized data. Probably one of the most interesting studies of Real-World Evidence (RWE) generation using retrospective cohorts are the development of early predictors of endpoints related to the evolution of the disease in distinct patient types. Several examples of successful application of Machine Learning (ML) techniques to clinical data repositories are: Deep Patient , which predicts the development of various diseases with 90% accuracy [9], Doctor AI , which makes preventive future diagnoses and recommends treatments [40] or Deepcare , which makes predictions of disease progression, along with intervention recommendations [41], just to cite a few cases. Apart from the obvious use of retrospective data in Phase IV studies mentioned above, other innovative applications are also possible in clinical trials. Recruitment and retention of control intervention arm patients, generally consisting of placebo, poses different ethical and logistic challenges, especially in oncology [42]. Thus a variety of synthetic control statistical methods can be used to evaluate the comparative effectiveness of an intervention using external control data, defined as cohorts of patients from external sources [43]. Another innovative application of Deep Learning (DL) methods to large datasets is the generation of synthetic data. They can be extremely useful if they meet two conditions: (1) high fidelity, meaning the generated data maintain utility for the intended task, such as yielding comparable performance when training a diagnostic model; and (2) compliance with privacy standards, ensuring that no real patient identities are disclosed in the synthetic dataset [44]. Under this approach, DL algorithms use real data to “learn” its structure and the relationships among its descriptor variables that are further used to generate high fidelity synthetic data with the characteristics of real data. In the case of simulated EHR, these are so realistic that could be used to perform new studies, given that the algorithm accurately captures and reproduces in the results the relationship among the variables. Since the data does not correspond to any real patient they are not subject to GDPR and could, in principle, be used without legal restrictions for clinical research purposes. Generative adversarial networks (GANs) have seen remarkable success, giving rise to diverse generative models for EHR synthesis, oriented to various clinical purposes [45-47]. The importance of having access to original RWD to generate high fidelity simulated data is clearly demonstrated by the fact that AI models tend to collapse when trained on recursively generated data [48] It is worth noting that, unlike in static databases, BPS is updated on a monthly basis, opening thus the possibility of prospective or ambispective study proposals. Some successful use cases Many retrospective studies using large databases of RWD focus on the efficacy of treatments or drugs. In a recent study carried out in this SPE the evidence of the association between increased use of direct oral anticoagulants and a reduction in the rate of atrial fibrillation-related stroke and major bleeding at the population level was demonstrated using a population of 95,085 patients [49]. Actually, the same methodology can be used for different types of interventions, and some original retrospective studies can be carried out. As an example, the Andalusian Genomic Surveillance System [50], specifically the COVID-19 circuit [51], made available a large number of SARS-CoV-2 viral genomes which, in combination with the clinical data of the patients, has been used to carry out an interesting study of the effect of the viral lineage and specific viral mutations on patient survival [52]. Another different application of the methodology is the study of the effect of other concomitant pharmacologic treatments in the outcome of the disease. Finding unexpectedly good prognostics associated with other drugs provided by other reasons to the patients can lead to drug repurposing proposals. Some examples of drug repurposing have been made in the iRWD during the recent pandemics. Thus, it has been demonstrated that several drugs, like vitamin D [53] or antipsychotic drugs like aripiprazole have a significant protective effect on COVID-19 patient survival [54]. This study was further generalized to discover a significant protective effect in a total of 21 drugs of common use in patients [7]. Another aspect of paramount interest is the identification of the population at risk of different diseases. In the recent COVID-19 pandemics, the identification of individuals at risk of severe infection was a priority for clinicians and health systems, and was successfully addressed using RWD from BPS [55]. Early endpoint predictors are also of paramount interest for the health system. Recently, a model was developed to identify individuals at high risk of ovarian cancer without the need of using specific tumour markers or prior stratification into risk groups. Utilizing clinical variables from BPS that include demographics, chronic diseases, symptoms, blood test results, and healthcare utilization patterns, a ML algorithm achieved a sensitivity of 0.65 and a specificity of 0.85 [56]. Recently, as an example of innovative clinical data utilization, approximately 1 million real electronic health records (EHRs) from diabetic patients were employed to train a Generative Adversarial Network (GAN), the medGAN [45], a specific type Deep Learning algorithm. The goal was to generate synthetic EHRs that closely mimic the characteristics of diabetic patients while ensuring that the data do not correspond to any actual individuals, thus preserving patient privacy. These data are available within the “Synthetic Clinical Health Records” challenge of the Critical Assessment on Massive Data Analysis conference [57]. Governance and opportunities for collaboration Currently, an internal committee within IRWD is responsible for prioritizing collaboration proposals based firstly on their alignment with the policies of the Andalusian Health System. In addition to this alignment, the committee evaluates proposals according to the available resources within the organization, as well as the resources and budget provided by the proposed project. This structured evaluation ensures that collaborations are both strategically relevant and feasible, optimizing the use of IRWD's capabilities for impactful research and innovation. Currently, more than 10 major pharmaceutical companies and three Contract Research Organizations (CROs) have established agreements with iRWD. The revenue generated from these partnerships not only sustains the operations of iRWD but also enables its infrastructure to be utilized by the health system and other stakeholders within the research ecosystem for various clinical studies. This collaborative model significantly enhances the generation of medical knowledge, fostering innovation and supporting a wide range of research initiatives that benefit public health. At the international level, IRWD has laid the foundation for participation in collaborative European and global federated projects, such as the European Medicines Agency's Darwin initiative [33] or the former European Health Data and Evidence Network (EHDEN) [34]. In such projects, a network of partners conduct large-scale RWE studies by federating analytical tools and adopting the OMOP common data standard [35]. The adoption of a common data standard allows the same code to be distributed and executed across sites, with the results collected and aggregated in a federated approach. This is part of the Line of actuation “Implementation and analysis of databases in precision medicine” of the “Biotechnology applied to health” area of the Complementary plans with the Autonomous Regions [36]. Discussion Prospects for preventive medicine A key component of precision preventive medicine is the use of endpoint predictors that can anticipate adverse health outcomes before they manifest [58]. By leveraging these predictors, healthcare providers can implement timely, personalized interventions that may reduce the risk of disease progression and improve overall patient outcomes. For instance, machine learning algorithms have been employed to predict cardiovascular events by analyzing patient data, allowing for more precise preventive strategies [59]. Similarly, the use of predictive analytics in cancer prevention has shown promise in identifying individuals at high risk of developing malignancies, guiding early interventions [60]. Actually, most of the current algorithms for the predictions of endpoints, prognostic, etc., are applied to individuals previously stratified as risk population, which means that some previous symptom has already been detected and most likely they are under a more detailed control. To unleash the potential of preventive medicine, ideal early predictors should be directly applicable to the general population, with no previous indication of risk, and using the data that the health system captures from them for other reasons. An example is the early predictor of ovarian cancer mentioned above [56], based on a series of features, such as analytics, other concomitant diagnosis and visits to the different services of the health system that can trigger an alert of risk of ovarian cancer with a reasonable accuracy. Actually, ovarian cancer serves as a compelling example for preventive medicine, as its high mortality rate (over 75%) contrasts sharply with the significantly higher cure rate (90%) achieved through early diagnosis, when the tumor is still confined to the ovaries (stage I) [61]. Preventive early warning of risk of diagnosis as a pre‑screening strategy based in a predictor clearly constitutes a cost-effective strategy [56]. It is important to highlight that developing an accurate predictor requires not only sufficient data, but also skilled teams, robust computational infrastructure, and time to develop it. However, once developed, these predictors can be massively applied to large datasets with relatively short runtimes. As illustrated in Figure 3, a strategy focused on the creation of carefully selected early endpoint predictors holds the potential to transform healthcare systems. This transformative approach positions the data repository, traditionally located at the end of the healthcare data production chain and primarily used for non-clinical purposes, as a critical frontline tool in preventive medicine by identifying at-risk patients even before symptoms emerge. As a result, the healthcare system will gradually shift from its traditional reactive role, which involves treating patients once their symptoms become severe enough to prompt a hospital visit, to a more preventive role. In this new approach, patients will be identified and contacted before symptoms arise, enabling interventions that are more efficient, less invasive, and likely less costly. Sustainability of the health system The vision of thin initiative is deeply rooted in the long-term sustainability of the healthcare system by transforming it into an essential component in the cycle of knowledge and wealth generation and public well-being creation. A key aspect of this vision is the gradual evolution of the healthcare system from its current reactive model to a more predictive one, where medical interventions can be deployed before conditions fully manifest. This transition will be achieved through the progressive incorporation of carefully selected endpoint predictors, chosen not only for their medical efficacy but also for their cost‑effectiveness. By doing so, the healthcare system will be able to provide more timely and targeted treatments, improving patient outcomes and ultimately enhancing life expectancy and quality of life and, at the same time, will become more sustainable. Moreover, the expanded use of clinical big data will play a crucial role in this transformation. Leveraging the vast amounts of patient data available, the system will facilitate the generation of new knowledge at an accelerated pace. This will drive innovation and transformation within the healthcare system on two key fronts. First, it will hasten the development and adoption of cutting-edge medical technologies and treatments, keeping the system at the forefront of global healthcare advancements. Second, the ability to conduct large-scale observational studies using RWD also contributes to the sustainability of the health system by reducing the time and the personnel required for data collection, thereby optimizing resource use and minimizing costs associated with traditional data gathering methods, ultimately fostering a more efficient healthcare model. And third, the revenues generated from public-private partnerships, fuelled by the exploitation of these data for research and innovation, will flow back into the public healthcare system. This reinvestment will not only bolster the financial sustainability of the system but also ensure that it remains fair and accessible to all. At the heart of this vision is a fundamental principle of justice [62]. The data collected within the Andalusian Public Health System belongs to the patients, and it is only fair that the benefits derived from the use of these data ultimately return to them. By using patient data to enhance care, advance medical knowledge, and drive the development of more effective treatments, the system ensures that the patient, who is the ultimate source of this data, reaps the rewards. This reinvestment in patient care, both directly through better health outcomes and indirectly through a more efficient and sustainable healthcare system, upholds the principle of justice, ensuring that the patient remains the primary beneficiary of their own data. Conclusions The collaborative model presented here constitutes a significant step toward transforming healthcare offering a collaborative environment with a large repository of clinical data that can be used within a robust legal framework, and a secure processing environment (iRWD) to foster medical research and innovation. By leveraging large-scale medical datasets and developing advanced predictive models, the system has the potential to shift the focus of healthcare from reactive to preventive, enabling early interventions that improve patient outcomes and enhance life expectancy. The strategic incorporation of cost-effective endpoint predictors, combined with an ethical and secure infrastructure, offers the promise of a more efficient and sustainable healthcare system. Moreover, the ability to use clinical big data not only accelerates innovation but also generates revenues through public-private partnerships, which are reinvested back into the healthcare system, making it more equitable and financially sustainable. Crucially, this model upholds the principle of justice by ensuring that the benefits derived from patient data ultimately return to the patients themselves, enhancing both care quality and healthcare equity. As a result, the Andalusian Health system is positioned as a leading example of how data-driven healthcare can contribute to a fairer, more efficient, and innovative future. Abbreviations BPS: Population Health Database (from “Base Poblacional de Salud” in Spanish) CCEIBA: Coordinating Committee on Biomedical Research Ethics of Andalucía CRO: Contract Research Organizations EHDS: European Health Data Space GDPR: General Data Protection Regulation IADP: Impact Assessment on Data Protection iRWD: Infrastructure for evidence generation from Real World Data RWD: Real-World Data RWE: Real World Evidence Declarations Ethics approval and consent to participate Not applicable Consent for publication Not applicable Availability of data and materials Not applicable Competing interests he authors declare that they have no competing interests Funding This work has been co-financed by the Spanish Ministry of Science and Innovation with funds from the European Union NextGenerationEU (PRTR-C17.I1) and the Regional Ministry of University, Research and Innovation of the Autonomous Community of Andalusia within the framework of the Biotechnology Plan applied to Health. It is also supported by the Spanish Ministry of Science and Innovation (PID2020-117979RB-I00), by the Instituto de Salud Carlos III (ISCIII), co-funded with European Regional Development Funds (IMP/00019), and by the Consejeria de Salud y Consumo, Junta de Andalucia (IE19_259 FPS). Acknowledgements Not applicable References Liu F, Panagiotakos D: Real-world data: a brief review of the methods, applications, challenges and opportunities . BMC Medical Research Methodology 2022, 22 (1):287. Ramagopalan SV, Simpson A, Sammon C: Can real-world data really replace randomised clinical trials? BMC medicine 2020, 18 (1):1-2. Use of real-world evidence to support regulatory decision-making for medical devices: guidance for industry and Food and Drug Administration staff [https://www.fda.gov/downloads/medicaldevices/deviceregulationandguidance/guidancedocuments/ucm513027.pdf] Makady A, de Boer A, Hillege H, Klungel O, Goettsch W: What is real-world data? A review of definitions based on literature and stakeholder interviews . Value in Health 2017, 20 (7):858-865. Corrigan-Curay J, Sacks L, Woodcock J: Real-World Evidence and Real-World Data for Evaluating Drug Safety and Effectiveness . JAMA : the journal of the American Medical Association 2018, 320 (9):867-868. Sherman RE, Anderson SA, Dal Pan GJ, Gray GW, Gross T, Hunter NL, LaVange L, Marinac-Dabic D, Marks PW, Robb MA: Real-world evidence—what is it and what can it tell us . N Engl J Med 2016, 375 (23):2293-2297. Loucera C, Carmona R, Esteban-Medina M, Bostelmann G, Muñoyerro-Muñiz D, Villegas R, Peña-Chilet M, Dopazo J: Real-world evidence with a retrospective cohort of 15,968 COVID-19 hospitalized patients suggests 21 new effective treatments . Virol J 2023, 20 (1):226. Suvarna V: Phase IV of drug development . Perspectives in clinical research 2010, 1 (2):57-60. Miotto R, Li L, Kidd BA, Dudley JT: Deep patient: an unsupervised representation to predict the future of patients from the electronic health records . Scientific reports 2016, 6 (1):1-10. Regulation of the European Parliament and of the Council on the European Health Data Space [https://eur-lex.europa.eu/legal-content/EN/TXT/?uri=CELEX:52022PC0197] Regulation (EU) 2018/1725 of the European Parliament and of the Council of 23 October 2018 on the protection of natural persons with regard to the processing of personal data by the Union institutions, bodies, offices and agencies and on the free movement of such data, and repealing Regulation [https://eur-lex.europa.eu/eli/reg/2018/1725/oj] Muñoyerro-Muñiz D, Goicoechea-Salazar J, García-León F, Laguna-Tellez A, Larrocha-Mata D, Cardero-Rivas M: Health record linkage: Andalusian health population database . Gaceta Sanitaria 2019, 34 (2):105-113. IRWD. Platform for the secure generation of biomedical knowledge from clinical big data of the population health database. [https://www.clinbioinfosspa.es/projects/iRWD/indexEsp.html] Active and Healthy Living in the Digital World [https://futurium.ec.europa.eu/en/active-and-healthy-living-digital-world] Salazar JAG, GARCÍA MAN, Téllez AL, Casasola VDC, Herrera JR, CABEZAS FM: Desarrollo de un sistema de codificación automática para recuperar y analizar textos diagnósticos de los registros de servicios de urgencias hospitalarios . Emergencias 2013, 25 (6):430-436. Joint Resolution 1/2021 of the General Secretariat for Health Research, Development and Innovation of the Regional Ministry of Health and Families and the Management Directorate of the Andalusian Health Service. [https://www.sspa.juntadeandalucia.es/servicioandaluzdesalud/sites/default/files/sincfiles/wsas-media-sas_normativa_mediafile/2021/resolucion_conjunta_acceso_a_datos_investigacion_def_20211201%28F%29.pdf] Coordinating Committee on Biomedical Research Ethics of Andalucía (CCEIBA) [http://si.easp.es/eticaysalud/content/comite-coordinador-etica-investigacion-biomedica-andalucia] García-León F, Villegas-Portero R, Goicoechea-Salazar J, Muñoyerro-Muñiz D, Dopazo J: Impact assessment on data protection in research projects . Gaceta sanitaria 2020, 34 (5):521-523. Guidance. How to complete your ethics self-assessment [https://ec.europa.eu/research/participants/data/ref/h2020/grants_manual/hi/ethics/h2020_hi_ethics-self-assess_en.pdf] Python [https://www.python.org/] McKinney W: Data structures for statistical computing in python . Proceedings of the 9th Python in Science Conference 2010, 445 (1):51-56. Harris CR, Millman KJ, van der Walt SJ, Gommers R, Virtanen P, Cournapeau D, Wieser E, Taylor J, Berg S, Smith NJ: Array programming with NumPy . Nature 2020, 585 (7825):357-362. Virtanen P, Gommers R, Oliphant TE, Haberland M, Reddy T, Cournapeau D, Burovski E, Peterson P, Weckesser W, Bright J: SciPy 1.0: fundamental algorithms for scientific computing in Python . Nature methods 2020, 17 (3):261-272. Seabold S, Perktold J: Statsmodels: Econometric and statistical modeling with python . Proceedings of the 9th Python in Science Conference 2010, 57 (61):10-25080. R: A Language and Environment for Statistical Computing [http://www.R-project.org] Gentleman RC, Carey VJ, Bates DM, Bolstad B, Dettling M, Dudoit S, Ellis B, Gautier L, Ge Y, Gentry J: Bioconductor: open software development for computational biology and bioinformatics . Genome biology 2004, 5 (10):1-16. Pedregosa F, Varoquaux G, Gramfort A, Michel V, Thirion B, Grisel O, Blondel M, Prettenhofer P, Weiss R, Dubourg V: Scikit-learn: Machine learning in Python . Journal of machine learning research 2011, 12 (Oct):2825-2830. Abadi M, Barham P, Chen J, Chen Z, Davis A, Dean J, Devin M, Ghemawat S, Irving G, Isard M: TensorFlow: A System for Large-Scale Machine Learning . In: OSDI: 2016 . 265-283. Conda-Forge Community: The conda-forge project: Community-based software distribution built on the conda package format and ecosystem . Zenodo 2015. Grüning B, Dale R, Sjödin A, Chapman BA, Rowe J, Tomkins-Tinch CH, Valieris R, Köster J: Bioconda: sustainable and comprehensive software distribution for the life sciences . Nature methods 2018, 15 (7):475-476. Boettiger C: An introduction to Docker for reproducible research . ACM SIGOPS Operating Systems Review 2015, 49 (1):71-79. Shen H: Interactive notebooks: Sharing the code . Nature 2014, 515 (7525):152-152. Data Analysis and Real World Interrogation Network (DARWIN EU) [https://www.darwin-eu.org/] European Health Data & Evidence Network (EHDEN) [https://www.ehden.eu/] Bathelt F: The usage of OHDSI OMOP–a scoping review . Proceedings of the German Medical Data Sciences (GMDS) 2021:95-95. Complementary plans with the Autonomous Regions [https://www.ciencia.gob.es/en/Estrategias-y-Planes/Plan-de-Recuperacion-Transformacion-y-Resiliencia-PRTR/Planes-complementarios-con-CCAA.html] Emmert-Streib F, Dehmer M: Introduction to survival analysis in practice . Machine Learning and Knowledge Extraction 2019, 1 (3):1013-1038. Bauskis A, Strange C, Molster C, Fisher C: The diagnostic odyssey: insights from parents of children living with an undiagnosed condition . Orphanet journal of rare diseases 2022, 17 (1):233. Thomas DL: Global control of hepatitis C: where challenge meets opportunity . Nature medicine 2013, 19 (7):850-858. Choi E, Bahadori MT, Schuetz A, Stewart WF, Sun J: Doctor ai: Predicting clinical events via recurrent neural networks . In: Machine Learning for Healthcare Conference: 2016 . 301-318. Pham T, Tran T, Phung D, Venkatesh S: Deepcare: A deep dynamic memory model for predictive medicine . In: Pacific-Asia Conference on Knowledge Discovery and Data Mining: 2016 . Springer: 30-41. Chiodo GT, Tolle SW, Bevan L: Placebo-controlled trials: good science or medical neglect? Western Journal of Medicine 2000, 172 (4):271. Thorlund K, Dron L, Park JJ, Mills EJ: Synthetic and external controls in clinical trials–a primer for researchers . Clinical epidemiology 2020:457-467. Chen RJ, Lu MY, Chen TY, Williamson DF, Mahmood F: Synthetic data in machine learning for medicine and healthcare . Nature Biomedical Engineering 2021, 5 (6):493-497. Choi E, Biswal S, Malin B, Duke J, Stewart WF, Sun J: Generating multi-label discrete patient records using generative adversarial networks . In: Machine learning for healthcare conference: 2017 . PMLR: 286-305. Beaulieu-Jones BK, Wu ZS, Williams C, Lee R, Bhavnani SP, Byrd JB, Greene CS: Privacy-preserving generative deep neural networks support clinical data sharing . Circulation: Cardiovascular Quality and Outcomes 2019, 12 (7):e005122. Baowaly MK, Lin C-C, Liu C-L, Chen K-T: Synthesizing electronic health records using improved generative adversarial networks . Journal of the American Medical Informatics Association 2019, 26 (3):228-241. Shumailov I, Shumaylov Z, Zhao Y, Papernot N, Anderson R, Gal Y: AI models collapse when trained on recursively generated data . Nature 2024, 631 (8022):755-759. Loucera C, Carmona R, Bostelmann G, Muñoyerro-Muñiz D, Villegas R, Gonzalez-Manzanares R, Dopazo J, Anguita M: Evidence of the association between increased use of direct oral anticoagulants and a reduction in the rate of atrial fibrillation-related stroke and major bleeding at the population level (2012-2019) . Med Clin (Barc) 2024, 162 (5):220-227. Casimiro-Soriguer CS, Pérez-Florido J, Robles EA, Lara M, Aguado A, Rodríguez Iglesias MA, Lepe JA, García F, Pérez-Alegre M, Andújar E et al : The integrated genomic surveillance system of Andalusia (SIEGA) provides a One Health regional resource connected with the clinic . Scientific reports 2024, 14 (1):19200. Dopazo J, Maya-Miles D, García F, Lorusso N, Calleja M, Pareja MJ, López-Miranda J, Rodríguez-Baño J, Padillo J, Túnez I et al : Implementing Personalized Medicine in COVID-19 in Andalusia: An Opportunity to Transform the Healthcare System . J Pers Med 2021, 11 (6). Loucera C, Perez-Florido J, Casimiro-Soriguer CS, Ortuño FM, Carmona R, Bostelmann G, Martínez-González LJ, Muñoyerro-Muñiz D, Villegas R, Rodriguez-Baño J et al : Assessing the Impact of SARS-CoV-2 Lineages and Mutations on Patient Survival . Viruses 2022, 14 (9). Loucera C, Peña-Chilet M, Esteban-Medina M, Muñoyerro-Muñiz D, Villegas R, Lopez-Miranda J, Rodriguez-Baño J, Túnez I, Bouillon R, Dopazo J: Real world evidence of calcifediol or vitamin D prescription and mortality rate of COVID-19 in a retrospective cohort of hospitalized Andalusian patients . Scientific reports 2021, 11 (1):23380. Loucera-Muñecas C, Canal-Rivero M, Ruiz-Veguilla M, Carmona R, Bostelmann G, Garrido-Torres N, Dopazo J, Crespo-Facorro B: Aripiprazole as protector against COVID-19 mortality . Scientific reports 2024, 14 (1):12362. Carmona-Pirez J, Ioakeim-Skoufa I, Gimeno-Miguel A, Poblador-Plou B, González-Rubio F, Muñoyerro-Muñiz D, Rodriguez-Herrera J, Goicoechea-Salazar JA, Prados-Torres A, Villegas-Portero R: Multimorbidity profiles and infection severity in COVID-19 population using network analysis in the andalusian health population database . International Journal of Environmental Research and Public Health 2022, 19 (7):3808. de la Oliva V, Esteban-Medina A, Alejos L, Munoyerro-Muniz D, Villegas R, Dopazo J, Loucera C: Early prediction of ovarian cancer risk based on real world data . medRxiv 2024:2024.2007. 2026.24310994. The Synthetic Clinical Health Records Challenge [https://bipress.boku.ac.at/camda-play/the-camda-contest-challenges/#synthetic-health-records] Ginsburg GS, Phillips KA: Precision medicine: from science to value . Health affairs 2018, 37 (5):694-701. Ambale-Venkatesh B, Yang X, Wu CO, Liu K, Hundley WG, McClelland R, Gomes AS, Folsom AR, Shea S, Guallar E: Cardiovascular event prediction by machine learning: the multi-ethnic study of atherosclerosis . Circulation research 2017, 121 (9):1092-1101. Wang X, Zhang Y, Hao S, Zheng L, Liao J, Ye C, Xia M, Wang O, Liu M, Weng CH: Prediction of the 1-year risk of incident lung cancer: prospective study using electronic health records from the state of Maine . Journal of medical Internet research 2019, 21 (5):e13260. Ye B, Gagnon A, Mok SC: Recent technical strategies to identify diagnostic biomarkers for ovarian cancer . Expert Rev Proteomics 2007, 4 (1):121-131. Beauchamp TL, Childress JF: Principles of biomedical ethics : Edicoes Loyola; 1994. Additional Declarations No competing interests reported. Cite Share Download PDF Status: Posted Version 1 posted You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-5389651","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":373972009,"identity":"2f0684f3-23ec-4df3-9faa-493f5c7dd68e","order_by":0,"name":"Dolores Muñoyerro-Muñiz","email":"","orcid":"","institution":"Subdirección Técnica Asesora de Gestión de la Información. Servicio Andaluz de Salud","correspondingAuthor":false,"prefix":"","firstName":"Dolores","middleName":"","lastName":"Muñoyerro-Muñiz","suffix":""},{"id":373972011,"identity":"12db9de3-87e1-44cd-8522-1ac21f1fb3d4","order_by":1,"name":"Román Villegas","email":"","orcid":"","institution":"Subdirección Técnica Asesora de Gestión de la Información. Servicio Andaluz de Salud","correspondingAuthor":false,"prefix":"","firstName":"Román","middleName":"","lastName":"Villegas","suffix":""},{"id":373972013,"identity":"283fac0e-c599-4462-9429-0adda529598f","order_by":2,"name":"Víctor de la Oliva","email":"","orcid":"","institution":"Andalusian Public Foundation Progress and Health- FPS","correspondingAuthor":false,"prefix":"","firstName":"Víctor","middleName":"de la","lastName":"Oliva","suffix":""},{"id":373972015,"identity":"a4de1cde-831d-49a7-bb6a-75aa4462b238","order_by":3,"name":"Alberto Esteban-Medina","email":"","orcid":"","institution":"Andalusian Public Foundation Progress and Health- FPS","correspondingAuthor":false,"prefix":"","firstName":"Alberto","middleName":"","lastName":"Esteban-Medina","suffix":""},{"id":373972017,"identity":"1644bbb7-5eda-42c0-9694-d9bd2162439c","order_by":4,"name":"Patricia Fernández Valle","email":"","orcid":"","institution":"Andalusian Public Foundation Progress and Health- FPS","correspondingAuthor":false,"prefix":"","firstName":"Patricia","middleName":"Fernández","lastName":"Valle","suffix":""},{"id":373972018,"identity":"6fddb4bd-56e1-4241-8530-05533ebc6525","order_by":5,"name":"Ana Sánchez","email":"","orcid":"","institution":"Andalusian Public Foundation Progress and Health- FPS","correspondingAuthor":false,"prefix":"","firstName":"Ana","middleName":"","lastName":"Sánchez","suffix":""},{"id":373972019,"identity":"49fd99ef-8ff3-482f-b922-c1c43d5630a9","order_by":6,"name":"M. Belen Susin","email":"","orcid":"","institution":"Andalusian Public Foundation Progress and Health- FPS","correspondingAuthor":false,"prefix":"","firstName":"M.","middleName":"Belen","lastName":"Susin","suffix":""},{"id":373972020,"identity":"fb0bb7d3-a9e0-4ba8-bfb8-1f4c8bff773b","order_by":7,"name":"Isidoro Gutierrez-Alvarez","email":"","orcid":"","institution":"Andalusian Public Foundation Progress and Health- FPS","correspondingAuthor":false,"prefix":"","firstName":"Isidoro","middleName":"","lastName":"Gutierrez-Alvarez","suffix":""},{"id":373972021,"identity":"29d3edf3-af0c-4f20-8fd9-9b6ac80635b7","order_by":8,"name":"Marta Reboredo","email":"","orcid":"","institution":"Andalusian Public Foundation Progress and Health- FPS","correspondingAuthor":false,"prefix":"","firstName":"Marta","middleName":"","lastName":"Reboredo","suffix":""},{"id":373972022,"identity":"550da5d0-dbba-4e74-aedf-7c368abbc516","order_by":9,"name":"Laura Alejos","email":"","orcid":"","institution":"Subdirección Técnica Asesora de Gestión de la Información. Servicio Andaluz de Salud","correspondingAuthor":false,"prefix":"","firstName":"Laura","middleName":"","lastName":"Alejos","suffix":""},{"id":373972023,"identity":"d349292c-2ff1-45ea-bf8c-f231af605579","order_by":10,"name":"Carlos Loucera","email":"","orcid":"","institution":"Andalusian Public Foundation Progress and Health- FPS","correspondingAuthor":false,"prefix":"","firstName":"Carlos","middleName":"","lastName":"Loucera","suffix":""},{"id":373972024,"identity":"80c1cf79-6686-4959-a89f-6af5a04ef0f2","order_by":11,"name":"Joaquin Dopazo","email":"data:image/png;base64,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","orcid":"","institution":"Andalusian Public Foundation Progress and Health- FPS","correspondingAuthor":true,"prefix":"","firstName":"Joaquin","middleName":"","lastName":"Dopazo","suffix":""}],"badges":[],"createdAt":"2024-11-04 16:30:31","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-5389651/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-5389651/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":68354982,"identity":"9b4dae94-ef6d-4322-866b-e3b34e2e0a99","added_by":"auto","created_at":"2024-11-06 11:12:12","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":121365,"visible":true,"origin":"","legend":"\u003cp\u003eSchematic representation of the collaborative environment for secure and ethics data analysis. It is composed of: the data authorization process (central column), the Health Population Database (BPS, bottom centre) and the Secure Processing Environment (IRWD, bottom left). The data journey in a conventional study (right) and the data journey within the Secure Processing Environment (left) are also depicted.\u003c/p\u003e","description":"","filename":"image1.png","url":"https://assets-eu.researchsquare.com/files/rs-5389651/v1/8f9c25dec6dad4a79b2cba53.png"},{"id":68354983,"identity":"8ed26c5e-5520-4a3d-85d0-056b4f301001","added_by":"auto","created_at":"2024-11-06 11:12:12","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":311307,"visible":true,"origin":"","legend":"\u003cp\u003eA selection of the main types of studies that can be performed in the iRWD.\u003c/p\u003e","description":"","filename":"image2.png","url":"https://assets-eu.researchsquare.com/files/rs-5389651/v1/6fffee3b0d8cf29e943c06a8.png"},{"id":68354984,"identity":"1cb29b4d-c364-4216-81ec-444cf7290e6a","added_by":"auto","created_at":"2024-11-06 11:12:12","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":307865,"visible":true,"origin":"","legend":"\u003cp\u003eSummarized architecture of the Andalusian Health System with a double layer of digitalization: the first layer, Diraya, for primary use of clinical data for the management of the patient, and a second layer, BPS, for permanent data storage for administrative and research purposes. The SPE is connected to the BPS in order to promote secondary use of clinical data for research purposes. In particular, predictors developed using BPS data can be run directly over BPS. By doing this, the data repository at the end of the data production chain of the health system becomes the first line in the application of preventive medicine.\u003c/p\u003e","description":"","filename":"image3.png","url":"https://assets-eu.researchsquare.com/files/rs-5389651/v1/385997be87c796145954a787.png"},{"id":75334025,"identity":"8eb94f38-8429-4766-919b-872e96747eda","added_by":"auto","created_at":"2025-02-03 13:01:20","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":3527308,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-5389651/v1/128de8ff-d7c9-4239-b497-37633e3a6601.pdf"}],"financialInterests":"No competing interests reported.","formattedTitle":"Ethical and Secure Evidence Generation from Regionwide Clinical Data through a Collaborative Environment for Advancing Predictive Care","fulltext":[{"header":"Background","content":"\u003cp\u003eThe use of Real-World Data (RWD) in healthcare is revolutionizing the generation of medical knowledge, offering unprecedented opportunities for research, innovation, and improved patient outcomes\u0026nbsp;[1]. RWD, which includes data derived from a variety of sources such as electronic health records (EHRs), minimum basic data sets (inpatients, outpatient mayor surgery, hospital emergencies and medical day hospital), mental health information systems, analytical and image tests, vaccines, renal patients, and pharmacy, reflects the real-time experiences of patients in routine clinical settings. Although the use of RWD in healthcare has its challenges, such as potential biases, incomplete information, and privacy concerns, this data provides a more comprehensive picture of patient populations and treatment effects than traditional clinical trials, which often have strict inclusion criteria that limit the diversity of participants\u0026nbsp;[2]. As such, RWD is instrumental in generating Real-World Evidence (RWE) that can inform clinical decision-making, guide the development of new therapeutic approaches, and support regulatory decisions\u0026nbsp;[3]. The ability to analyse vast datasets in real-world conditions enables researchers to better understand the effectiveness of interventions across diverse patient populations and healthcare environments, ultimately contributing to more personalized and effective medical care\u0026nbsp;[4].\u003c/p\u003e\n\u003cp\u003eOne of the key advantages of RWD is its potential to accelerate the pace of medical innovation. By utilizing data already being generated within the healthcare system, researchers can bypass the lengthy and expensive process of traditional randomized controlled trials (RCTs) in some cases, while still obtaining valuable insights into treatment efficacy and safety\u0026nbsp;[5]. This is especially relevant in the context of rare diseases, where limited patient populations make traditional trials challenging, or in urgent public health crises like the COVID-19 pandemic, where timely insights are essential for guiding treatment strategies\u0026nbsp;[6, 7]. Moreover, RWD is increasingly being used not only for post-marketing surveillance of drugs, allowing for the continuous monitoring of drug safety and efficacy after they have been approved, providing an additional layer of patient protection\u0026nbsp;[8], but also for a variety of studies that range from conventional epidemiologic studies, including cost of treatment, etc. to more sophisticated endpoint predictors involving the use of artificial intelligence\u0026nbsp;[9].\u003c/p\u003e\n\u003cp\u003eIn the European context, the utilization of RWD is becoming increasingly structured and formalized. The European Health Data Space (EHDS)\u0026nbsp;[10], proposed by the European Parliament, aims to create a unified framework for the use of health data across Europe . This legislation seeks to promote the secure and ethical use of health data for research and innovation while ensuring the protection of patient privacy and data security. The EHDS emphasizes the importance of creating trusted research environments where data can be analysed securely, ensuring compliance with the General Data Protection Regulation (GDPR)\u0026nbsp;[11]\u0026nbsp;and safeguarding the rights of individuals. Such a framework is essential for fostering public trust and enabling the responsible use of health data on a large scale.\u003c/p\u003e\n\u003cp\u003eHere we present a leading example of the effective use of RWD within a secure processing environment, fully compliant with the EHDS, provided by the Andalusian Public Health System (SSPA, acronym for \u0026ldquo;Sistema Sanitario P\u0026uacute;blico Andaluz\u0026rdquo; in Spanish). The SSPA has developed one of the most comprehensive and ambitious health information systems in Europe, known as the Andalusian Health Population Database (BPS, acronym for \u0026ldquo;Base Poblacional de Salud\u0026rdquo; in Spanish). The BPS includes medical records dating back to 2001 for more than 15 million patients, constituting a vast repository of clinical data covering almost the entire population of Andalusia\u0026nbsp;[12]. This data includes diagnoses, laboratory tests, medical imaging, pharmaceutical records, and more, making it an invaluable resource for both clinical and epidemiological research. The integration of BPS with a Secure Processing Environments, such as the iRWD platform\u0026nbsp;[13], offers an invaluable environment for researchers to analyse clinical data while maintaining compliance with GDPR and EHDS regulations, ensuring patient privacy.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eThe potential of RWD to generate new medical knowledge is vast, but its effective use requires the right infrastructure, legal frameworks, and data governance practices. The combination of comprehensive data systems like BPS, secure environments for data analysis, and adherence to regulatory standards such as those proposed by the EHDS is essential for maximizing the benefits of RWD while safeguarding patient rights. The Andalusian Health System exemplifies how this can be achieved, offering a model, open to collaborations, for the integration of healthcare data into the research ecosystem in a way that benefits both patients and the broader medical community. Importantly, the Andalusian model demonstrates how a region-wide, integrated health system can leverage RWD to foster innovation and improve healthcare delivery. By utilizing BPS data within a secure, legally compliant framework, Andalusia has positioned itself at the forefront of the secondary use of healthcare data for research, setting a standard for other regions and countries to follow.\u003c/p\u003e"},{"header":"Methods","content":"\u003ch2\u003eA collaborative environment for secure and ethics secondary use of medical data\u003c/h2\u003e\n\u003cp\u003eThe collaborative infrastructure for secure and ethical secondary use of medical data is structured around three fundamental components that work together to facilitate advanced medical research within the medical data space of Andalusia. First, Andalusia houses an extensive repository of medical data, encompassing a wide range of patient information collected from the health system across the region. This vast pool of data constitutes a rich resource with the potential for conducting large-scale studies, supporting the development of predictive models, and advancing medical knowledge. Second, the use of data from Andalusian patients is governed by legal regulations designed to ensure that patient privacy is protected while enabling the responsible use of information for research purposes. These regulations provide a clear legal framework that balances the need for scientific advancement with ethical considerations surrounding data security and patient consent. Third, the availability of a secure, state-of-the-art secure processing environment where data can safely be analysed. This environment is designed to maintain the integrity of the data, safeguard patient confidentiality, and facilitate complex computational analyses. Together, these three elements\u0026mdash;the vast medical data repository, the legal framework, and the secure processing environment infrastructure\u0026mdash;create a robust ecosystem that fosters innovation in medical research while ensuring compliance with ethical standards.\u003c/p\u003e\n\u003ch2\u003eThe Andalusian Health Population Database\u003c/h2\u003e\n\u003cp\u003eThe SSPA provides service to a population of approximately 8.5 million users, and comprises a total of 55 hospitals and 34 primary care districts with a total of 1505 primary care centres. The SSPA activity encompasses more than 80 million primary and specialized visits, half a million hospital admissions, 5 million of hospital emergencies per year, and all the diagnosis, prescriptions and an ample variety of data derived from these interventions. The complete Health System is fully digitalized by means of the digital system Diraya, where all the data are indexed and referred to the patient. Diraya incrementally dumps all patient data in the Health Population Database (BPS, acronym of \u0026ldquo;Base Poblacional de Salud\u0026rdquo; in Spanish) on a monthly basis. The BPS is one of the most ambitious resources of the SSPA. BPS is a health information system that collects clinical data and data on the use of health resources of each person receiving health care in the SSPA, totalling over 15 million patients accumulated since 2001.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eFrom the data collected in BPS, it is possible to obtain estimates on health, the behaviour of users in relation to health services and to stratify the population in order to guide the provision of these services. The BPS also enables longitudinal studies to be carried out, the incidence of pathologies to be estimated and projections to be made on the state of health of the population and its resource needs. It also makes it possible to analyse efficiency in the use of resources by health care providers. Since its inception, BPS was conceived with a strong focus on research. In fact, it has been included in the Repository of Innovative Practices in Active and Healthy Aging of the European Commission\u0026nbsp;[14]. This repository aims to collect the most innovative initiatives of European partners, to facilitate the dissemination of knowledge and mobilize resources to ensure the implementation of these initiatives at European level.\u003c/p\u003e\n\u003cp\u003eThe structure of BPS is defined in a reference publication [12], which describes the development of this information system that connects data from multiple Electronic Health Records (EHR) to improve assistance to patients, health services administration, management, evaluation, and inspection, as well as public health and research. BPS connects pseudonymized data from nearly any individual of the whole Andalusia population. In fact, BPS has an estimated coverage of 99% of the Andalusian population [12]. \u0026nbsp;The data are sourced from different SSPA information systems including: the Andalusian Public Health System patient database, DIRAYA electronic medical records, the minimum basic data sets (inpatients, outpatient major surgery, hospital emergencies and medical day hospital), mental health information systems, analytical and image tests, vaccines, renal patients, and pharmacy, among others. In order to have the data as structured as possible, an automatic coder developed in-house for hospital emergency and primary care electronic medical records [15] is used to code clinical diagnoses into ICD10. Also 80 chronic pathologies were identified and coded.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eTo illustrate the dimension of BPS it is worth mentioning some figures (see Table 1) about the data stored, such as the 874 million diagnoses, the 2,428 million analytical tests, or the 7,000 million medical images, to cite just a few examples. This volume makes BPS one of the largest medical RWD repositories in the world.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eTable 1. BPS in figures.\u003c/p\u003e\n\u003ctable border=\"0\" cellspacing=\"0\" cellpadding=\"0\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 76.6784%;\"\u003e\n \u003cp\u003e\u003cstrong\u003eData\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 23.3216%;\"\u003e\n \u003cp\u003e\u003cstrong\u003eNumber (millions)\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 76.6784%;\"\u003e\n \u003cp\u003eUsers (2001- September 2024)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 23.3216%;\"\u003e\n \u003cp\u003e15.8\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 76.6784%;\"\u003e\n \u003cp\u003eMedical diagnoses (90% automatic coding)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 23.3216%;\"\u003e\n \u003cp\u003e874.1\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 76.6784%;\"\u003e\n \u003cp\u003eNursing diagnoses\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 23.3216%;\"\u003e\n \u003cp\u003e51.3\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 76.6784%;\"\u003e\n \u003cp\u003eAnalytical test results\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 23.3216%;\"\u003e\n \u003cp\u003e2,428\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 76.6784%;\"\u003e\n \u003cp\u003ePACS images\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 23.3216%;\"\u003e\n \u003cp\u003e7,000\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 76.6784%;\"\u003e\n \u003cp\u003eVaccination events\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 23.3216%;\"\u003e\n \u003cp\u003e79\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 76.6784%;\"\u003e\n \u003cp\u003eProsthesis and Implant records\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 23.3216%;\"\u003e\n \u003cp\u003e0.7\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 76.6784%;\"\u003e\n \u003cp\u003eTemporary Incapacity (TI) Processes\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 23.3216%;\"\u003e\n \u003cp\u003e11.2\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 76.6784%;\"\u003e\n \u003cp\u003eVital signs (weight, height, BMI, blood pressure)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 23.3216%;\"\u003e\n \u003cp\u003e83.5\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 76.6784%;\"\u003e\n \u003cp\u003eHospital stays\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 23.3216%;\"\u003e\n \u003cp\u003e3.5\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 76.6784%;\"\u003e\n \u003cp\u003eFunctional evaluations\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 23.3216%;\"\u003e\n \u003cp\u003e4.2\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 76.6784%;\"\u003e\n \u003cp\u003eCognitive assessments\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 23.3216%;\"\u003e\n \u003cp\u003e2.7\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e\n\u003cp\u003e\u0026nbsp;\u003c/p\u003e\n\u003ch2\u003eProcedure for the access to medical data in Andalusia\u003c/h2\u003e\n\u003cp\u003eThe most recent regulation for the use of medical data for research purposes was issued the 4\u003csup\u003eth\u003c/sup\u003e December, 2021, in the Joint Resolution 1/2021 of the General Secretariat for Research, Development and Innovation in Health of the Regional Ministry of Health and Families and the Management Directorate of the Andalusian Health Service [16]. This innovative resolution anticipated most of the strategy and procedures for access to medical data which has further been described in the recent European Parliament legislative resolution of 24 April 2024 in the proposal for a Regulation of the European Parliament and of the Council on the EHDS [10].\u003c/p\u003e\n\u003cp\u003eThe Joint Resolution 1/2021 defines a Health Data Access Body (Data Access Committee, DAC) responsible for granting access to electronic health data for secondary use. \u0026nbsp; In order to evaluate the convenience of granting the access to the data requested, the DAC requires: i) the report of the study for which the data is requested, ii) the permission of the Coordinating Committee on Biomedical Research Ethics of Andaluc\u0026iacute;a (CCEIBA) [17], iii) the Impact Assessment on Data Protection (IADP) [18] document, subject to the GDPR and taking into account the Horizon 2020 Programme Guidance How to complete your ethics self-assessment [19] and, iv) signed commitment of the principal researcher in which he/she undertakes not to redistribute data, not to attempt to re-identify individuals in the dataset and removing the dataset once the study has finished.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eIn essence, the data access regulation described is completely compliant with the EHDS regulations and would fit perfectly in the proposed future federated structure of the EHDS.\u0026nbsp;\u003c/p\u003e\n\u003ch2\u003eThe Infrastructure iRWD, a Secure Processing Environment\u003c/h2\u003e\n\u003cp\u003eThe third key piece of this collaborative model that facilitates the secondary use of medical RWD for the generation of new medical knowledge is a Secure Processing Environment (SPE). \u0026nbsp;This SPE consists of a computational infrastructure (iRWD), funded within the scope of the Andalusian Plan for Research, Development and Innovation [13]. This infrastructure is located within the SSPA corporate network and has specifically designed for the secure analysis of data protected by the GDPR, being compliant with the definition of SPE as described in Article 50 of the Resolution of the European Parliament of 24 April 2024 on the proposal for a Regulation of the European Parliament and of the Council on the EHDS [10].\u003c/p\u003e\n\u003ch2\u003eSecure data management in iRWD\u003c/h2\u003e\n\u003cp\u003eThis SPE follows a data management procedure that avoids any aspect that could compromise privacy, as described in the joint Resolution 1/2021 of the General Secretariat for Research, Development and Innovation in Health of the Regional Ministry of Health and Families and the Management Directorate of the Andalusian Health Service [16], also in accordance with the recent European Parliament legislative resolution of 24 April 2024 on the proposal for a Regulation of the European Parliament and of the Council on the European Health Data Space [10].\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eSince the iRWD is located within the SSPA corporate network and is operated by personnel of the Fundation Progress and Health, which belongs to the health system, data never leave the secure environment of the health system and are managed by trusted personnel from the health system. These two aspects are crucial for an IADP which minimizes the relative risk for the data used in the study [18]. The IADP is a document consisting of a description of the data life cycle, detailing the activities to be carried out, the specific data to be processed and the people and technologies involved, both for the data acquisition process and for its storage, processing, transfer to third parties and final destruction. The IADP also includes an analysis of the necessity and proportionality of the treatment, including the report on the legitimacy of the use of databases without informed consent in health research in certain circumstances, as well as the legitimacy of the use of the BPS as a research infrastructure. And, finally, the document identifies and assesses the probability and impact derived from the possibility of a risk materializing, with the objective of establishing preventive, corrective and mitigating actions to minimize risk exposure. The final section of conclusions specifies the potential hazards, their inherent and residual risk, and mitigation measures that could be implemented. By using the iRWD, an action plan is not required [18].\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eSummarizing, the data management procedure (described in the IADP as the life cycle of the data) is as follows: i) the iRWD work team request the data to BPS with the corresponding approval of the DAC, ii) the BPS team extract the data and pseudonymize them, iii) the BPS transfer the pseudonymized data to the iRWD, iv) the iRWD work team carries out the analysis of the data as described in the study report, v) when the study is done the data are removed from the iRWD infrastructure.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eThe use of pseudonymized data is an important feature specific to our collaborative model approach (Figure 1 left part), as it allows patient re-identification when necessary, provided that the study has approval from the ethics committee. This approach contrasts with conventional data analysis frameworks (Figure 1, right side), which utilize anonymized data that do not permit any form of individual identification. As a result, studies requiring patient-specific insights for their benefit cannot be conducted under a fully anonymized data framework.\u003c/p\u003e\n\u003ch2\u003eSample size\u003c/h2\u003e\n\u003cp\u003eAn attractive aspect of RWD region-wide studies is that no sample size justification is necessary. In conventional studies using hospital databases, sample size estimation is required to define the minimal number of individuals which constitute a significative representation of the whole population and, consequently, can be used to respond to the objectives of the study with adequate power and precision so that the results can be generalized to the entire population. Region-wide studies include all patients who meet the selection criteria, i.e., the sample is directly the entire target population.\u003c/p\u003e\n\u003ch2\u003eHardware and software resources\u003c/h2\u003e\n\u003cp\u003eFor the sake of reproducibility and explainability, all the software used in the studies carried out in the iRWD SPE is open. For data processing, we rely on \u003cem\u003ePython\u003c/em\u003e [20] and its extensive ecosystem of libraries, such as \u003cem\u003ePandas\u003c/em\u003e [21] and \u003cem\u003eNumPy\u003c/em\u003e [22]. This allows us to efficiently clean, transform, and explore our data, setting the stage for subsequent analysis. When it comes to statistical analysis, we utilize mainly \u003cem\u003ePython\u003c/em\u003e libraries like \u003cem\u003eNumpy\u003c/em\u003e, \u003cem\u003eScipy\u003c/em\u003e [23] and \u003cem\u003estatsmodels\u003c/em\u003e [24], whereas \u003cem\u003eR\u003c/em\u003e [25] and \u003cem\u003eBioconductor\u003c/em\u003e [26] packages are used for some advanced analysis. AI-driven models, including machine learning and deep learning, are backed by the core libraries of the Python scientific distribution, like \u003cem\u003eNumPy\u003c/em\u003e and \u003cem\u003eSciPy\u003c/em\u003e, as well as specialized tools like \u003cem\u003escikit-learn\u003c/em\u003e [27], and \u003cem\u003eTensorFlow\u003c/em\u003e [28].\u003c/p\u003e\n\u003cp\u003eTo ensure computational reproducibility, we employ a multi-layered approach. Data versioning is controlled by the BPS while software versioning is achieved using locally managed gitlab servers. Conda environments, drawing from the \u003cem\u003econda-forge\u003c/em\u003e [29] and \u003cem\u003eBioconda\u003c/em\u003e [30] channels, provide project-specific software dependencies, minimizing conflicts and ensuring consistent results. Additionally, we encapsulate entire analysis environments within Docker containers [31], further enhancing reproducibility across different computing platforms. Finally, interactive analyses are facilitated through securely-served Jupyter notebooks [32], leveraging the aforementioned reproducibility infrastructure for computation.\u003c/p\u003e\n\u003cp\u003eThe hardware infrastructure supporting the SPE is specifically designed to meet the rigorous computational demands of modern clinical Big Data research. A general-purpose computing cluster provides significant computational capacity, featuring 1,832 cores, 13,568 GB of RAM, and 993 TB of storage, making it well-suited for data-intensive studies. For deep learning applications, a specialized GPU server cluster is available to handle the high-performance requirements of AI-driven research. This cluster includes 192 CPU cores, 4,864 GB of RAM, and 18 high-performance GPUs. Each server is equipped with 1 TB of local storage and access to a shared 150 TB network storage system, facilitating collaborative research and efficient data management. This infrastructure ensures that even the most demanding computational tasks can be executed efficiently.\u003c/p\u003e"},{"header":"Results","content":"\u003ch2\u003eActivity of the iRWD infrastructure and analysis portfolio\u0026nbsp;\u003c/h2\u003e\n\u003cp\u003eMultiple type of studies can be carried out in large RWD repositories. Figure 2 summarizes the most common studies already performed, ongoing or under consideration in the iRWD infrastructure. Clockwise from the top, the most common studies currently requested by pharma companies are epidemiological studies of prevalence or incidence of diseases (a total of 60%), followed by more detailed studies of cost of disease or interventions (approximately 20%).\u003c/p\u003e\n\u003cp\u003eSurvival studies [37] are fundamental in medical research as they provide crucial insights into patient prognosis, the efficacy of treatments, and the natural course of diseases. These studies allow for the estimation of survival probabilities over time, identifying factors that may influence patient outcomes, such as comorbidities, demographics, and treatment modalities. By analysing survival data, researchers can also identify potential risk factors and guide clinical decision-making.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eSince many patients will present concomitant drug treatments, it is relatively easy to extend the concept of survival studies to repurposing studies by modelling the effect of the concomitant drugs in the outcome of the patient. Similarly, specific treatments or interventions can be assessed, taking into consideration all the possible confusion variables, providing valuable information on the efficacy of these. Actually, in the case of post-marketing surveillance of drugs, they are Phase IV studies [8] of clinical trials.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eAs previously mentioned, the secure data management environment used here involves the utilisation of pseudonymized data, which makes possible patient re-identification, if required by the study and authorised by the ethics committee. Studies aiming to detect undiagnosed patients of rare diseases affected by the well-known diagnostic odyssey [38] or \u0026nbsp; to discover undiagnosed patients of an infectious condition, like Hepatitis C, HIV, etc., for eradication programs [39], \u0026nbsp; or other similar ones, are ultimately oriented to the identification of individuals, which is possible with pseudonymized data but not with anonymized data.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eProbably one of the most interesting studies of\u0026nbsp;Real-World Evidence (RWE) generation using\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003eretrospective cohorts\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003eare the development of early predictors of endpoints related to the evolution of the disease in distinct patient types. Several examples of successful application of Machine Learning (ML) techniques to clinical data repositories are: \u003cem\u003eDeep Patient\u003c/em\u003e, which predicts the development of various diseases with 90% accuracy [9], \u003cem\u003eDoctor AI\u003c/em\u003e, which makes preventive future diagnoses and recommends treatments [40] or \u003cem\u003eDeepcare\u003c/em\u003e, which makes predictions of disease progression, along with intervention recommendations\u0026nbsp;[41], just to cite a few cases.\u003c/p\u003e\n\u003cp\u003eApart from the obvious use of retrospective data in Phase IV studies mentioned above, other innovative applications are also possible in clinical trials. Recruitment and retention of control intervention arm patients, generally consisting of placebo, poses different ethical and logistic challenges, especially in oncology\u0026nbsp;[42]. Thus a variety of synthetic control statistical methods can be used to evaluate the comparative effectiveness of an intervention using external control data, defined as cohorts of patients from external sources\u0026nbsp;[43].\u003c/p\u003e\n\u003cp\u003eAnother innovative application of Deep Learning (DL) methods to large datasets is the generation of synthetic data. They can be extremely useful if they meet two conditions: (1) high fidelity, meaning the generated data maintain utility for the intended task, such as yielding comparable performance when training a diagnostic model; and (2) compliance with privacy standards, ensuring that no real patient identities are disclosed in the synthetic dataset\u0026nbsp;[44].\u003c/p\u003e\n\u003cp\u003eUnder this approach, DL algorithms use real data to \u0026ldquo;learn\u0026rdquo; its structure and the relationships among its descriptor variables that are further used to generate high fidelity synthetic data with the characteristics of real data. In the case of simulated EHR, these are so realistic that could be used to perform new studies, given that the algorithm accurately captures and reproduces in the results the relationship among the variables. Since the data does not correspond to any real patient they are not subject to GDPR and could, in principle, be used without legal restrictions for clinical research purposes.\u0026nbsp;Generative adversarial networks (GANs) have seen remarkable success, giving rise to diverse generative models for EHR synthesis, oriented to various clinical purposes\u0026nbsp;[45-47].\u003c/p\u003e\n\u003cp\u003eThe importance of having access to original RWD to generate high fidelity simulated data is clearly demonstrated by the fact that AI models tend to collapse when trained on recursively generated data\u0026nbsp;[48]\u003c/p\u003e\n\u003cp\u003eIt is worth noting that, unlike in static databases, BPS is updated on a monthly basis, opening thus the possibility of prospective or ambispective study proposals.\u0026nbsp;\u003c/p\u003e\n\u003ch2\u003eSome successful use cases\u003c/h2\u003e\n\u003cp\u003eMany retrospective studies using large databases of RWD focus on the efficacy of treatments or drugs. In a recent study carried out in this SPE the evidence of the association between increased use of direct oral anticoagulants and a reduction in the rate of atrial fibrillation-related stroke and major bleeding at the population level was demonstrated using a population of 95,085 patients [49]. Actually, the same methodology can be used for different types of interventions, and some original retrospective studies can be carried out. As an example, the Andalusian Genomic Surveillance System [50], specifically the COVID-19 circuit [51], \u0026nbsp; made available a large number of SARS-CoV-2 viral genomes which, in combination with the clinical data of the patients, has been used to carry out an interesting study of the effect of the viral lineage and specific viral mutations on patient survival [52]. Another different application of the methodology is the study of the effect of other concomitant pharmacologic treatments in the outcome of the disease. Finding unexpectedly good prognostics associated with other drugs provided by other reasons to the patients can lead to drug repurposing proposals. Some examples of drug repurposing have been made in the iRWD during the recent pandemics. Thus, it has been demonstrated that several drugs, like vitamin D [53] or antipsychotic drugs like aripiprazole have a significant protective effect on COVID-19 patient survival [54]. This study was further generalized to discover a significant protective effect in a total of 21 drugs of common use in patients [7].\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eAnother aspect of paramount interest is the identification of the population at risk of different diseases. In the recent COVID-19 pandemics, the identification of individuals at risk of severe infection was a priority for clinicians and health systems, and was successfully addressed using RWD from BPS [55].\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eEarly endpoint predictors are also of paramount interest for the health system. Recently, a model was developed to identify individuals at high risk of ovarian cancer without the need of using specific tumour markers or prior stratification into risk groups. Utilizing clinical variables from BPS that include demographics, chronic diseases, symptoms, blood test results, and healthcare utilization patterns, a ML algorithm achieved a sensitivity of 0.65 and a specificity of 0.85\u0026nbsp;[56].\u003c/p\u003e\n\u003cp\u003eRecently, as an example of innovative clinical data utilization, approximately 1 million real electronic health records (EHRs) from diabetic patients were employed to train a Generative Adversarial Network (GAN), the medGAN [45], a specific type Deep Learning algorithm. The goal was to generate synthetic EHRs that closely mimic the characteristics of diabetic patients while ensuring that the data do not correspond to any actual individuals, thus preserving patient privacy. These data are available within the \u0026ldquo;Synthetic Clinical Health Records\u0026rdquo; challenge of the Critical Assessment on Massive Data Analysis conference [57].\u0026nbsp;\u003c/p\u003e\n\u003ch2\u003eGovernance and opportunities for collaboration\u003c/h2\u003e\n\u003cp\u003eCurrently, an internal committee within IRWD is responsible for prioritizing collaboration proposals based firstly on their alignment with the policies of the Andalusian Health System. In addition to this alignment, the committee evaluates proposals according to the available resources within the organization, as well as the resources and budget provided by the proposed project. This structured evaluation ensures that collaborations are both strategically relevant and feasible, optimizing the use of IRWD\u0026apos;s capabilities for impactful research and innovation.\u003c/p\u003e\n\u003cp\u003eCurrently, more than 10 major pharmaceutical companies and three Contract Research Organizations (CROs) have established agreements with iRWD. The revenue generated from these partnerships not only sustains the operations of iRWD but also enables its infrastructure to be utilized by the health system and other stakeholders within the research ecosystem for various clinical studies. This collaborative model significantly enhances the generation of medical knowledge, fostering innovation and supporting a wide range of research initiatives that benefit public health.\u003c/p\u003e\n\u003cp\u003eAt the international level, IRWD has laid the foundation for participation in collaborative European and global federated projects, such as the European Medicines Agency\u0026apos;s Darwin initiative [33] or the former European Health Data and Evidence Network (EHDEN) [34]. In such projects, a network of partners conduct large-scale RWE studies by federating analytical tools and adopting the OMOP common data standard [35]. The adoption of a common data standard allows the same code to be distributed and executed across sites, with the results collected and aggregated in a federated approach. This is part of the Line of actuation \u0026ldquo;Implementation and analysis of databases in precision medicine\u0026rdquo; of the \u0026ldquo;Biotechnology applied to health\u0026rdquo; area of the Complementary plans with the Autonomous Regions [36].\u0026nbsp;\u003c/p\u003e"},{"header":"Discussion","content":"\u003ch2\u003eProspects for preventive medicine\u0026nbsp;\u003c/h2\u003e\n\u003cp\u003eA key component of precision preventive medicine is the use of endpoint predictors that can anticipate adverse health outcomes before they manifest\u0026nbsp;[58]. By leveraging these predictors, healthcare providers can implement timely, personalized interventions that may reduce the risk of disease progression and improve overall patient outcomes. For instance, machine learning algorithms have been employed to predict cardiovascular events by analyzing patient data, allowing for more precise preventive strategies\u0026nbsp;[59]. Similarly, the use of predictive analytics in cancer prevention has shown promise in identifying individuals at high risk of developing malignancies, guiding early interventions\u0026nbsp;[60].\u003c/p\u003e\n\u003cp\u003eActually, most of the current algorithms for the predictions of endpoints, prognostic, etc., are applied to individuals previously stratified as risk population, which means that some previous symptom has already been detected and most likely they are under a more detailed control. To unleash the potential of preventive medicine, ideal early predictors should be directly applicable to the general population, with no previous indication of risk, and using the data that the health system captures from them for other reasons. An example is the early predictor of ovarian cancer mentioned above\u0026nbsp;[56], based on a series of features, such as analytics, other concomitant diagnosis and visits to the different services of the health system that can trigger an alert of risk of ovarian cancer with a reasonable accuracy. Actually, ovarian cancer serves as a compelling example for preventive medicine, as its high mortality rate (over 75%) contrasts sharply with the significantly higher cure rate (90%) achieved through early diagnosis, when the tumor is still confined to the ovaries (stage I)\u0026nbsp;[61]. Preventive early warning of risk of diagnosis as a pre‑screening strategy based in a predictor clearly constitutes a cost-effective strategy\u0026nbsp;[56].\u003c/p\u003e\n\u003cp\u003eIt is important to highlight that developing an accurate predictor requires not only sufficient data, but also skilled teams, robust computational infrastructure, and time to develop it. However, once developed, these predictors can be massively applied to large datasets with relatively short runtimes. As illustrated in Figure 3, a strategy focused on the creation of carefully selected early endpoint predictors holds the potential to transform healthcare systems. This transformative approach positions the data repository, traditionally located at the end of the healthcare data production chain and primarily used for non-clinical purposes, as a critical frontline tool in preventive medicine by identifying at-risk patients even before symptoms emerge. As a result, the healthcare system will gradually shift from its traditional reactive role, which involves treating patients once their symptoms become severe enough to prompt a hospital visit, to a more preventive role. In this new approach, patients will be identified and contacted before symptoms arise, enabling interventions that are more efficient, less invasive, and likely less costly.\u003c/p\u003e\n\u003ch2\u003eSustainability of the health system\u003c/h2\u003e\n\u003cp\u003eThe vision of thin initiative is deeply rooted in the long-term sustainability of the healthcare system by transforming it into an essential component in the cycle of knowledge and wealth generation and public well-being creation. A key aspect of this vision is the gradual evolution of the healthcare system from its current reactive model to a more predictive one, where medical interventions can be deployed before conditions fully manifest. This transition will be achieved through the progressive incorporation of carefully selected endpoint predictors, chosen not only for their medical efficacy but also for their cost‑effectiveness. By doing so, the healthcare system will be able to provide more timely and targeted treatments, improving patient outcomes and ultimately enhancing life expectancy and quality of life and, at the same time, will become more sustainable.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eMoreover, the expanded use of clinical big data will play a crucial role in this transformation. Leveraging the vast amounts of patient data available, the system will facilitate the generation of new knowledge at an accelerated pace. This will drive innovation and transformation within the healthcare system on two key fronts. First, it will hasten the development and adoption of cutting-edge medical technologies and treatments, keeping the system at the forefront of global healthcare advancements. Second, the ability to conduct large-scale observational studies using RWD also contributes to the sustainability of the health system by reducing the time and the personnel required for data collection, thereby optimizing resource use and minimizing costs associated with traditional data gathering methods, ultimately fostering a more efficient healthcare model. And third, the revenues generated from public-private partnerships, fuelled by the exploitation of these data for research and innovation, will flow back into the public healthcare system. This reinvestment will not only bolster the financial sustainability of the system but also ensure that it remains fair and accessible to all.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eAt the heart of this vision is a fundamental principle of justice [62]. The data collected within the Andalusian Public Health System belongs to the patients, and it is only fair that the benefits derived from the use of these data ultimately return to them. By using patient data to enhance care, advance medical knowledge, and drive the development of more effective treatments, the system ensures that the patient, who is the ultimate source of this data, reaps the rewards. This reinvestment in patient care, both directly through better health outcomes and indirectly through a more efficient and sustainable healthcare system, upholds the principle of justice, ensuring that the patient remains the primary beneficiary of their own data.\u003c/p\u003e"},{"header":"Conclusions","content":"\u003cp\u003eThe collaborative model presented here constitutes a significant step toward transforming healthcare offering a collaborative environment with a large repository of clinical data that can be used within a robust legal framework, and a secure processing environment (iRWD) to foster medical research and innovation. By leveraging large-scale medical datasets and developing advanced predictive models, the system has the potential to shift the focus of healthcare from reactive to preventive, enabling early interventions that improve patient outcomes and enhance life expectancy. The strategic incorporation of cost-effective endpoint predictors, combined with an ethical and secure infrastructure, offers the promise of a more efficient and sustainable healthcare system. Moreover, the ability to use clinical big data not only accelerates innovation but also generates revenues through public-private partnerships, which are reinvested back into the healthcare system, making it more equitable and financially sustainable. Crucially, this model upholds the principle of justice by ensuring that the benefits derived from patient data ultimately return to the patients themselves, enhancing both care quality and healthcare equity. As a result, the Andalusian Health system is positioned as a leading example of how data-driven healthcare can contribute to a fairer, more efficient, and innovative future.\u003c/p\u003e\n\u003cp\u003e\u003cbr\u003e\u003c/p\u003e"},{"header":"Abbreviations","content":"\u003cp\u003eBPS: Population Health Database (from \u0026ldquo;Base Poblacional de Salud\u0026rdquo; in Spanish)\u003c/p\u003e\n\u003cp\u003eCCEIBA: Coordinating Committee on Biomedical Research Ethics of Andaluc\u0026iacute;a\u003c/p\u003e\n\u003cp\u003eCRO: Contract Research Organizations\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eEHDS: European Health Data Space\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eGDPR: General Data Protection Regulation\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eIADP: Impact Assessment on Data Protection\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eiRWD: Infrastructure for evidence generation from Real World Data\u003c/p\u003e\n\u003cp\u003eRWD: Real-World Data\u003c/p\u003e\n\u003cp\u003eRWE: Real World Evidence\u003c/p\u003e"},{"header":"Declarations","content":"\u003ch2\u003eEthics approval and consent to participate\u003c/h2\u003e\n\u003cp\u003eNot applicable\u003c/p\u003e\n\u003ch2\u003eConsent for publication\u003c/h2\u003e\n\u003cp\u003eNot applicable\u003c/p\u003e\n\u003ch2\u003eAvailability of data and materials\u003c/h2\u003e\n\u003cp\u003eNot applicable\u003c/p\u003e\n\u003ch2\u003eCompeting interests\u003c/h2\u003e\n\u003cp\u003ehe authors declare that they have no competing interests\u003c/p\u003e\n\u003ch2\u003eFunding\u003c/h2\u003e\n\u003cp\u003eThis work has been co-financed by the Spanish Ministry of Science and Innovation with funds from the European Union NextGenerationEU (PRTR-C17.I1) and the Regional Ministry of University, Research and Innovation of the Autonomous Community of Andalusia within the framework of the Biotechnology Plan applied to Health.\u0026nbsp;It is also supported by the Spanish Ministry of Science and Innovation (PID2020-117979RB-I00), by the Instituto de Salud Carlos III (ISCIII), co-funded with European Regional Development Funds (IMP/00019), and by the Consejeria de Salud y Consumo, Junta de Andalucia (IE19_259 FPS).\u003c/p\u003e\n\u003ch2\u003eAcknowledgements\u003c/h2\u003e\n\u003cp\u003eNot applicable\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n\u003cli\u003eLiu F, Panagiotakos D: \u003cstrong\u003eReal-world data: a brief review of the methods, applications, challenges and opportunities\u003c/strong\u003e. \u003cem\u003eBMC Medical Research Methodology \u003c/em\u003e2022, \u003cstrong\u003e22\u003c/strong\u003e(1):287.\u003c/li\u003e\n\u003cli\u003eRamagopalan SV, Simpson A, Sammon C: \u003cstrong\u003eCan real-world data really replace randomised clinical trials?\u003c/strong\u003e \u003cem\u003eBMC medicine \u003c/em\u003e2020, \u003cstrong\u003e18\u003c/strong\u003e(1):1-2.\u003c/li\u003e\n\u003cli\u003e\u003cstrong\u003eUse of real-world evidence to support regulatory decision-making for medical devices: guidance for industry and Food and Drug Administration staff \u003c/strong\u003e[https://www.fda.gov/downloads/medicaldevices/deviceregulationandguidance/guidancedocuments/ucm513027.pdf]\u003c/li\u003e\n\u003cli\u003eMakady A, de Boer A, Hillege H, Klungel O, Goettsch W: \u003cstrong\u003eWhat is real-world data? A review of definitions based on literature and stakeholder interviews\u003c/strong\u003e. \u003cem\u003eValue in Health \u003c/em\u003e2017, \u003cstrong\u003e20\u003c/strong\u003e(7):858-865.\u003c/li\u003e\n\u003cli\u003eCorrigan-Curay J, Sacks L, Woodcock J: \u003cstrong\u003eReal-World Evidence and Real-World Data for Evaluating Drug Safety and Effectiveness\u003c/strong\u003e. \u003cem\u003eJAMA : the journal of the American Medical Association \u003c/em\u003e2018, \u003cstrong\u003e320\u003c/strong\u003e(9):867-868.\u003c/li\u003e\n\u003cli\u003eSherman RE, Anderson SA, Dal Pan GJ, Gray GW, Gross T, Hunter NL, LaVange L, Marinac-Dabic D, Marks PW, Robb MA: \u003cstrong\u003eReal-world evidence\u0026mdash;what is it and what can it tell us\u003c/strong\u003e. \u003cem\u003eN Engl J Med \u003c/em\u003e2016, \u003cstrong\u003e375\u003c/strong\u003e(23):2293-2297.\u003c/li\u003e\n\u003cli\u003eLoucera C, Carmona R, Esteban-Medina M, Bostelmann G, Mu\u0026ntilde;oyerro-Mu\u0026ntilde;iz D, Villegas R, Pe\u0026ntilde;a-Chilet M, Dopazo J: \u003cstrong\u003eReal-world evidence with a retrospective cohort of 15,968 COVID-19 hospitalized patients suggests 21 new effective treatments\u003c/strong\u003e. \u003cem\u003eVirol J \u003c/em\u003e2023, \u003cstrong\u003e20\u003c/strong\u003e(1):226.\u003c/li\u003e\n\u003cli\u003eSuvarna V: \u003cstrong\u003ePhase IV of drug development\u003c/strong\u003e. \u003cem\u003ePerspectives in clinical research \u003c/em\u003e2010, \u003cstrong\u003e1\u003c/strong\u003e(2):57-60.\u003c/li\u003e\n\u003cli\u003eMiotto R, Li L, Kidd BA, Dudley JT: \u003cstrong\u003eDeep patient: an unsupervised representation to predict the future of patients from the electronic health records\u003c/strong\u003e. \u003cem\u003eScientific reports \u003c/em\u003e2016, \u003cstrong\u003e6\u003c/strong\u003e(1):1-10.\u003c/li\u003e\n\u003cli\u003e\u003cstrong\u003eRegulation of the European Parliament and of the Council on the European Health Data Space \u003c/strong\u003e[https://eur-lex.europa.eu/legal-content/EN/TXT/?uri=CELEX:52022PC0197]\u003c/li\u003e\n\u003cli\u003e\u003cstrong\u003eRegulation (EU) 2018/1725 of the European Parliament and of the Council of 23 October 2018 on the protection of natural persons with regard to the processing of personal data by the Union institutions, bodies, offices and agencies and on the free movement of such data, and repealing Regulation \u003c/strong\u003e[https://eur-lex.europa.eu/eli/reg/2018/1725/oj]\u003c/li\u003e\n\u003cli\u003eMu\u0026ntilde;oyerro-Mu\u0026ntilde;iz D, Goicoechea-Salazar J, Garc\u0026iacute;a-Le\u0026oacute;n F, Laguna-Tellez A, Larrocha-Mata D, Cardero-Rivas M: \u003cstrong\u003eHealth record linkage: Andalusian health population database\u003c/strong\u003e. \u003cem\u003eGaceta Sanitaria \u003c/em\u003e2019, \u003cstrong\u003e34\u003c/strong\u003e(2):105-113.\u003c/li\u003e\n\u003cli\u003e\u003cstrong\u003eIRWD. Platform for the secure generation of biomedical knowledge from clinical big data of the population health database. \u003c/strong\u003e[https://www.clinbioinfosspa.es/projects/iRWD/indexEsp.html]\u003c/li\u003e\n\u003cli\u003e\u003cstrong\u003eActive and Healthy Living in the Digital World \u003c/strong\u003e[https://futurium.ec.europa.eu/en/active-and-healthy-living-digital-world]\u003c/li\u003e\n\u003cli\u003eSalazar JAG, GARC\u0026Iacute;A MAN, T\u0026eacute;llez AL, Casasola VDC, Herrera JR, CABEZAS FM: \u003cstrong\u003eDesarrollo de un sistema de codificaci\u0026oacute;n autom\u0026aacute;tica para recuperar y analizar textos diagn\u0026oacute;sticos de los registros de servicios de urgencias hospitalarios\u003c/strong\u003e. \u003cem\u003eEmergencias \u003c/em\u003e2013, \u003cstrong\u003e25\u003c/strong\u003e(6):430-436.\u003c/li\u003e\n\u003cli\u003e\u003cstrong\u003eJoint Resolution 1/2021 of the General Secretariat for Health Research, Development and Innovation of the Regional Ministry of Health and Families and the Management Directorate of the Andalusian Health Service. \u003c/strong\u003e[https://www.sspa.juntadeandalucia.es/servicioandaluzdesalud/sites/default/files/sincfiles/wsas-media-sas_normativa_mediafile/2021/resolucion_conjunta_acceso_a_datos_investigacion_def_20211201%28F%29.pdf]\u003c/li\u003e\n\u003cli\u003e\u003cstrong\u003eCoordinating Committee on Biomedical Research Ethics of Andaluc\u0026iacute;a (CCEIBA) \u003c/strong\u003e[http://si.easp.es/eticaysalud/content/comite-coordinador-etica-investigacion-biomedica-andalucia]\u003c/li\u003e\n\u003cli\u003eGarc\u0026iacute;a-Le\u0026oacute;n F, Villegas-Portero R, Goicoechea-Salazar J, Mu\u0026ntilde;oyerro-Mu\u0026ntilde;iz D, Dopazo J: \u003cstrong\u003eImpact assessment on data protection in research projects\u003c/strong\u003e. \u003cem\u003eGaceta sanitaria \u003c/em\u003e2020, \u003cstrong\u003e34\u003c/strong\u003e(5):521-523.\u003c/li\u003e\n\u003cli\u003e\u003cstrong\u003eGuidance. How to complete your ethics self-assessment \u003c/strong\u003e[https://ec.europa.eu/research/participants/data/ref/h2020/grants_manual/hi/ethics/h2020_hi_ethics-self-assess_en.pdf]\u003c/li\u003e\n\u003cli\u003e\u003cstrong\u003ePython \u003c/strong\u003e[https://www.python.org/]\u003c/li\u003e\n\u003cli\u003eMcKinney W: \u003cstrong\u003eData structures for statistical computing in python\u003c/strong\u003e. \u003cem\u003eProceedings of the 9th Python in Science Conference \u003c/em\u003e2010, \u003cstrong\u003e445\u003c/strong\u003e(1):51-56.\u003c/li\u003e\n\u003cli\u003eHarris CR, Millman KJ, van der Walt SJ, Gommers R, Virtanen P, Cournapeau D, Wieser E, Taylor J, Berg S, Smith NJ: \u003cstrong\u003eArray programming with NumPy\u003c/strong\u003e. \u003cem\u003eNature \u003c/em\u003e2020, \u003cstrong\u003e585\u003c/strong\u003e(7825):357-362.\u003c/li\u003e\n\u003cli\u003eVirtanen P, Gommers R, Oliphant TE, Haberland M, Reddy T, Cournapeau D, Burovski E, Peterson P, Weckesser W, Bright J: \u003cstrong\u003eSciPy 1.0: fundamental algorithms for scientific computing in Python\u003c/strong\u003e. \u003cem\u003eNature methods \u003c/em\u003e2020, \u003cstrong\u003e17\u003c/strong\u003e(3):261-272.\u003c/li\u003e\n\u003cli\u003eSeabold S, Perktold J: \u003cstrong\u003eStatsmodels: Econometric and statistical modeling with python\u003c/strong\u003e. \u003cem\u003eProceedings of the 9th Python in Science Conference \u003c/em\u003e2010, \u003cstrong\u003e57\u003c/strong\u003e(61):10-25080.\u003c/li\u003e\n\u003cli\u003e\u003cstrong\u003eR: A Language and Environment for Statistical Computing \u003c/strong\u003e[http://www.R-project.org]\u003c/li\u003e\n\u003cli\u003eGentleman RC, Carey VJ, Bates DM, Bolstad B, Dettling M, Dudoit S, Ellis B, Gautier L, Ge Y, Gentry J: \u003cstrong\u003eBioconductor: open software development for computational biology and bioinformatics\u003c/strong\u003e. \u003cem\u003eGenome biology \u003c/em\u003e2004, \u003cstrong\u003e5\u003c/strong\u003e(10):1-16.\u003c/li\u003e\n\u003cli\u003ePedregosa F, Varoquaux G, Gramfort A, Michel V, Thirion B, Grisel O, Blondel M, Prettenhofer P, Weiss R, Dubourg V: \u003cstrong\u003eScikit-learn: Machine learning in Python\u003c/strong\u003e. \u003cem\u003eJournal of machine learning research \u003c/em\u003e2011, \u003cstrong\u003e12\u003c/strong\u003e(Oct):2825-2830.\u003c/li\u003e\n\u003cli\u003eAbadi M, Barham P, Chen J, Chen Z, Davis A, Dean J, Devin M, Ghemawat S, Irving G, Isard M: \u003cstrong\u003eTensorFlow: A System for Large-Scale Machine Learning\u003c/strong\u003e. In: \u003cem\u003eOSDI: 2016\u003c/em\u003e. 265-283.\u003c/li\u003e\n\u003cli\u003eConda-Forge Community: \u003cstrong\u003eThe conda-forge project: Community-based software distribution built on the conda package format and ecosystem\u003c/strong\u003e. \u003cem\u003eZenodo \u003c/em\u003e2015.\u003c/li\u003e\n\u003cli\u003eGr\u0026uuml;ning B, Dale R, Sj\u0026ouml;din A, Chapman BA, Rowe J, Tomkins-Tinch CH, Valieris R, K\u0026ouml;ster J: \u003cstrong\u003eBioconda: sustainable and comprehensive software distribution for the life sciences\u003c/strong\u003e. \u003cem\u003eNature methods \u003c/em\u003e2018, \u003cstrong\u003e15\u003c/strong\u003e(7):475-476.\u003c/li\u003e\n\u003cli\u003eBoettiger C: \u003cstrong\u003eAn introduction to Docker for reproducible research\u003c/strong\u003e. \u003cem\u003eACM SIGOPS Operating Systems Review \u003c/em\u003e2015, \u003cstrong\u003e49\u003c/strong\u003e(1):71-79.\u003c/li\u003e\n\u003cli\u003eShen H: \u003cstrong\u003eInteractive notebooks: Sharing the code\u003c/strong\u003e. \u003cem\u003eNature \u003c/em\u003e2014, \u003cstrong\u003e515\u003c/strong\u003e(7525):152-152.\u003c/li\u003e\n\u003cli\u003e\u003cstrong\u003eData Analysis and Real World Interrogation Network (DARWIN EU) \u003c/strong\u003e[https://www.darwin-eu.org/]\u003c/li\u003e\n\u003cli\u003e\u003cstrong\u003eEuropean Health Data \u0026amp; Evidence Network (EHDEN) \u003c/strong\u003e[https://www.ehden.eu/]\u003c/li\u003e\n\u003cli\u003eBathelt F: \u003cstrong\u003eThe usage of OHDSI OMOP\u0026ndash;a scoping review\u003c/strong\u003e. \u003cem\u003eProceedings of the German Medical Data Sciences (GMDS) \u003c/em\u003e2021:95-95.\u003c/li\u003e\n\u003cli\u003e\u003cstrong\u003eComplementary plans with the Autonomous Regions \u003c/strong\u003e[https://www.ciencia.gob.es/en/Estrategias-y-Planes/Plan-de-Recuperacion-Transformacion-y-Resiliencia-PRTR/Planes-complementarios-con-CCAA.html]\u003c/li\u003e\n\u003cli\u003eEmmert-Streib F, Dehmer M: \u003cstrong\u003eIntroduction to survival analysis in practice\u003c/strong\u003e. \u003cem\u003eMachine Learning and Knowledge Extraction \u003c/em\u003e2019, \u003cstrong\u003e1\u003c/strong\u003e(3):1013-1038.\u003c/li\u003e\n\u003cli\u003eBauskis A, Strange C, Molster C, Fisher C: \u003cstrong\u003eThe diagnostic odyssey: insights from parents of children living with an undiagnosed condition\u003c/strong\u003e. \u003cem\u003eOrphanet journal of rare diseases \u003c/em\u003e2022, \u003cstrong\u003e17\u003c/strong\u003e(1):233.\u003c/li\u003e\n\u003cli\u003eThomas DL: \u003cstrong\u003eGlobal control of hepatitis C: where challenge meets opportunity\u003c/strong\u003e. \u003cem\u003eNature medicine \u003c/em\u003e2013, \u003cstrong\u003e19\u003c/strong\u003e(7):850-858.\u003c/li\u003e\n\u003cli\u003eChoi E, Bahadori MT, Schuetz A, Stewart WF, Sun J: \u003cstrong\u003eDoctor ai: Predicting clinical events via recurrent neural networks\u003c/strong\u003e. In: \u003cem\u003eMachine Learning for Healthcare Conference: 2016\u003c/em\u003e. 301-318.\u003c/li\u003e\n\u003cli\u003ePham T, Tran T, Phung D, Venkatesh S: \u003cstrong\u003eDeepcare: A deep dynamic memory model for predictive medicine\u003c/strong\u003e. In: \u003cem\u003ePacific-Asia Conference on Knowledge Discovery and Data Mining: 2016\u003c/em\u003e. Springer: 30-41.\u003c/li\u003e\n\u003cli\u003eChiodo GT, Tolle SW, Bevan L: \u003cstrong\u003ePlacebo-controlled trials: good science or medical neglect?\u003c/strong\u003e \u003cem\u003eWestern Journal of Medicine \u003c/em\u003e2000, \u003cstrong\u003e172\u003c/strong\u003e(4):271.\u003c/li\u003e\n\u003cli\u003eThorlund K, Dron L, Park JJ, Mills EJ: \u003cstrong\u003eSynthetic and external controls in clinical trials\u0026ndash;a primer for researchers\u003c/strong\u003e. \u003cem\u003eClinical epidemiology \u003c/em\u003e2020:457-467.\u003c/li\u003e\n\u003cli\u003eChen RJ, Lu MY, Chen TY, Williamson DF, Mahmood F: \u003cstrong\u003eSynthetic data in machine learning for medicine and healthcare\u003c/strong\u003e. \u003cem\u003eNature Biomedical Engineering \u003c/em\u003e2021, \u003cstrong\u003e5\u003c/strong\u003e(6):493-497.\u003c/li\u003e\n\u003cli\u003eChoi E, Biswal S, Malin B, Duke J, Stewart WF, Sun J: \u003cstrong\u003eGenerating multi-label discrete patient records using generative adversarial networks\u003c/strong\u003e. In: \u003cem\u003eMachine learning for healthcare conference: 2017\u003c/em\u003e. PMLR: 286-305.\u003c/li\u003e\n\u003cli\u003eBeaulieu-Jones BK, Wu ZS, Williams C, Lee R, Bhavnani SP, Byrd JB, Greene CS: \u003cstrong\u003ePrivacy-preserving generative deep neural networks support clinical data sharing\u003c/strong\u003e. \u003cem\u003eCirculation: Cardiovascular Quality and Outcomes \u003c/em\u003e2019, \u003cstrong\u003e12\u003c/strong\u003e(7):e005122.\u003c/li\u003e\n\u003cli\u003eBaowaly MK, Lin C-C, Liu C-L, Chen K-T: \u003cstrong\u003eSynthesizing electronic health records using improved generative adversarial networks\u003c/strong\u003e. \u003cem\u003eJournal of the American Medical Informatics Association \u003c/em\u003e2019, \u003cstrong\u003e26\u003c/strong\u003e(3):228-241.\u003c/li\u003e\n\u003cli\u003eShumailov I, Shumaylov Z, Zhao Y, Papernot N, Anderson R, Gal Y: \u003cstrong\u003eAI models collapse when trained on recursively generated data\u003c/strong\u003e. \u003cem\u003eNature \u003c/em\u003e2024, \u003cstrong\u003e631\u003c/strong\u003e(8022):755-759.\u003c/li\u003e\n\u003cli\u003eLoucera C, Carmona R, Bostelmann G, Mu\u0026ntilde;oyerro-Mu\u0026ntilde;iz D, Villegas R, Gonzalez-Manzanares R, Dopazo J, Anguita M: \u003cstrong\u003eEvidence of the association between increased use of direct oral anticoagulants and a reduction in the rate of atrial fibrillation-related stroke and major bleeding at the population level (2012-2019)\u003c/strong\u003e. \u003cem\u003eMed Clin (Barc) \u003c/em\u003e2024, \u003cstrong\u003e162\u003c/strong\u003e(5):220-227.\u003c/li\u003e\n\u003cli\u003eCasimiro-Soriguer CS, P\u0026eacute;rez-Florido J, Robles EA, Lara M, Aguado A, Rodr\u0026iacute;guez Iglesias MA, Lepe JA, Garc\u0026iacute;a F, P\u0026eacute;rez-Alegre M, And\u0026uacute;jar E\u003cem\u003e et al\u003c/em\u003e: \u003cstrong\u003eThe integrated genomic surveillance system of Andalusia (SIEGA) provides a One Health regional resource connected with the clinic\u003c/strong\u003e. \u003cem\u003eScientific reports \u003c/em\u003e2024, \u003cstrong\u003e14\u003c/strong\u003e(1):19200.\u003c/li\u003e\n\u003cli\u003eDopazo J, Maya-Miles D, Garc\u0026iacute;a F, Lorusso N, Calleja M, Pareja MJ, L\u0026oacute;pez-Miranda J, Rodr\u0026iacute;guez-Ba\u0026ntilde;o J, Padillo J, T\u0026uacute;nez I\u003cem\u003e et al\u003c/em\u003e: \u003cstrong\u003eImplementing Personalized Medicine in COVID-19 in Andalusia: An Opportunity to Transform the Healthcare System\u003c/strong\u003e. \u003cem\u003eJ Pers Med \u003c/em\u003e2021, \u003cstrong\u003e11\u003c/strong\u003e(6).\u003c/li\u003e\n\u003cli\u003eLoucera C, Perez-Florido J, Casimiro-Soriguer CS, Ortu\u0026ntilde;o FM, Carmona R, Bostelmann G, Mart\u0026iacute;nez-Gonz\u0026aacute;lez LJ, Mu\u0026ntilde;oyerro-Mu\u0026ntilde;iz D, Villegas R, Rodriguez-Ba\u0026ntilde;o J\u003cem\u003e et al\u003c/em\u003e: \u003cstrong\u003eAssessing the Impact of SARS-CoV-2 Lineages and Mutations on Patient Survival\u003c/strong\u003e. \u003cem\u003eViruses \u003c/em\u003e2022, \u003cstrong\u003e14\u003c/strong\u003e(9).\u003c/li\u003e\n\u003cli\u003eLoucera C, Pe\u0026ntilde;a-Chilet M, Esteban-Medina M, Mu\u0026ntilde;oyerro-Mu\u0026ntilde;iz D, Villegas R, Lopez-Miranda J, Rodriguez-Ba\u0026ntilde;o J, T\u0026uacute;nez I, Bouillon R, Dopazo J: \u003cstrong\u003eReal world evidence of calcifediol or vitamin D prescription and mortality rate of COVID-19 in a retrospective cohort of hospitalized Andalusian patients\u003c/strong\u003e. \u003cem\u003eScientific reports \u003c/em\u003e2021, \u003cstrong\u003e11\u003c/strong\u003e(1):23380.\u003c/li\u003e\n\u003cli\u003eLoucera-Mu\u0026ntilde;ecas C, Canal-Rivero M, Ruiz-Veguilla M, Carmona R, Bostelmann G, Garrido-Torres N, Dopazo J, Crespo-Facorro B: \u003cstrong\u003eAripiprazole as protector against COVID-19 mortality\u003c/strong\u003e. \u003cem\u003eScientific reports \u003c/em\u003e2024, \u003cstrong\u003e14\u003c/strong\u003e(1):12362.\u003c/li\u003e\n\u003cli\u003eCarmona-Pirez J, Ioakeim-Skoufa I, Gimeno-Miguel A, Poblador-Plou B, Gonz\u0026aacute;lez-Rubio F, Mu\u0026ntilde;oyerro-Mu\u0026ntilde;iz D, Rodriguez-Herrera J, Goicoechea-Salazar JA, Prados-Torres A, Villegas-Portero R: \u003cstrong\u003eMultimorbidity profiles and infection severity in COVID-19 population using network analysis in the andalusian health population database\u003c/strong\u003e. \u003cem\u003eInternational Journal of Environmental Research and Public Health \u003c/em\u003e2022, \u003cstrong\u003e19\u003c/strong\u003e(7):3808.\u003c/li\u003e\n\u003cli\u003ede la Oliva V, Esteban-Medina A, Alejos L, Munoyerro-Muniz D, Villegas R, Dopazo J, Loucera C: \u003cstrong\u003eEarly prediction of ovarian cancer risk based on real world data\u003c/strong\u003e. \u003cem\u003emedRxiv \u003c/em\u003e2024:2024.2007. 2026.24310994.\u003c/li\u003e\n\u003cli\u003e\u003cstrong\u003eThe Synthetic Clinical Health Records Challenge \u003c/strong\u003e[https://bipress.boku.ac.at/camda-play/the-camda-contest-challenges/#synthetic-health-records]\u003c/li\u003e\n\u003cli\u003eGinsburg GS, Phillips KA: \u003cstrong\u003ePrecision medicine: from science to value\u003c/strong\u003e. \u003cem\u003eHealth affairs \u003c/em\u003e2018, \u003cstrong\u003e37\u003c/strong\u003e(5):694-701.\u003c/li\u003e\n\u003cli\u003eAmbale-Venkatesh B, Yang X, Wu CO, Liu K, Hundley WG, McClelland R, Gomes AS, Folsom AR, Shea S, Guallar E: \u003cstrong\u003eCardiovascular event prediction by machine learning: the multi-ethnic study of atherosclerosis\u003c/strong\u003e. \u003cem\u003eCirculation research \u003c/em\u003e2017, \u003cstrong\u003e121\u003c/strong\u003e(9):1092-1101.\u003c/li\u003e\n\u003cli\u003eWang X, Zhang Y, Hao S, Zheng L, Liao J, Ye C, Xia M, Wang O, Liu M, Weng CH: \u003cstrong\u003ePrediction of the 1-year risk of incident lung cancer: prospective study using electronic health records from the state of Maine\u003c/strong\u003e. \u003cem\u003eJournal of medical Internet research \u003c/em\u003e2019, \u003cstrong\u003e21\u003c/strong\u003e(5):e13260.\u003c/li\u003e\n\u003cli\u003eYe B, Gagnon A, Mok SC: \u003cstrong\u003eRecent technical strategies to identify diagnostic biomarkers for ovarian cancer\u003c/strong\u003e. \u003cem\u003eExpert Rev Proteomics \u003c/em\u003e2007, \u003cstrong\u003e4\u003c/strong\u003e(1):121-131.\u003c/li\u003e\n\u003cli\u003eBeauchamp TL, Childress JF: \u003cstrong\u003ePrinciples of biomedical ethics\u003c/strong\u003e: Edicoes Loyola; 1994.\u003c/li\u003e\n\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":true,"hideJournal":true,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true},"keywords":"Real-World Data, Real-World Evidence, electronic health records, predictive medicine, secure data processing, endpoint predictors, European Health Data Space, public-private partnerships, machine learning, data privacy.","lastPublishedDoi":"10.21203/rs.3.rs-5389651/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-5389651/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003e\u003cstrong\u003eBackground\u003c/strong\u003e: Data protection is often regarded as a major hurdle in clinical research, especially when handling sensitive patient information. However, by managing sensitive data within a secure environment, it becomes possible to mitigate these challenges and promote evidence generation through clinical data studies. This manuscript introduces a collaborative environment which facilitates the secure and ethical secondary use of clinical data for medical research. The environment combines a comprehensive health population database, integrating real-world data (RWD) from over 15 million patients, with a legal framework for patient privacy and data security and a secure processing environment (SPE) for ethically compliant clinical data research.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eMethods\u003c/strong\u003e: The SPE of this collaborative environment is a computational infrastructure (IRWD), located within the healthcare corporative network, that enables large-scale studies essential for real-world evidence (RWE) generation. Within iRWD, diverse studies can be conducted, including treatment efficacy assessments, survival analyses, and the development of predictive models. These studies leverage RWD to perform robust analyses while maintaining compliance with stringent regulatory standards, such as the European Health Data Space (EHDS) and the General Data Protection Regulation (GDPR), which govern data security and patient privacy.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eResults\u003c/strong\u003e: One of the key outcomes facilitated by this infrastructure is the systematic development of early endpoint predictors. These predictive algorithms can identify high-risk patients before symptoms emerge, enabling preventive interventions. The approach promotes a shift in healthcare from a reactive model to a preventive one, allowing for early, efficient, and cost-effective treatments that improve patient outcomes. Additionally, the model supports public-private partnerships, generating revenues that sustain this collaborative environment while reinforcing its capacity for continuous healthcare innovation.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConclusions\u003c/strong\u003e: The integration of clinical big data within a legally compliant and secure framework provided by the SPE offers a sustainable and proactive model for healthcare improvement. The infrastructure supports the development of cost-effective predictive models that are particularly valuable for an aging population, ultimately transforming healthcare delivery into a proactive, data-driven system. By positioning the health system as a central player in knowledge generation and wealth creation, this collaborative initiative sets a new benchmark for data-driven healthcare models across Europe and beyond, highlighting its potential for global adaptation and impact.\u003c/p\u003e","manuscriptTitle":"Ethical and Secure Evidence Generation from Regionwide Clinical Data through a Collaborative Environment for Advancing Predictive Care","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2024-11-06 11:12:07","doi":"10.21203/rs.3.rs-5389651/v1","editorialEvents":[{"type":"communityComments","content":0}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"f49634ca-93fb-45f2-8161-8afdd86bc2cb","owner":[],"postedDate":"November 6th, 2024","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[],"tags":[],"updatedAt":"2025-02-03T12:53:09+00:00","versionOfRecord":[],"versionCreatedAt":"2024-11-06 11:12:07","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-5389651","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-5389651","identity":"rs-5389651","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}