Infraestructuras de datos y conocimiento para la investigación científica sobre humanos y salud en Uruguay: relato de experiencia sobre un proceso de diseño constructivo crítico Data and knowledge infrastructures for scientific research on humans and health in Uruguay: an experience report on a critical constructive design process

Data-centrism, precision health, data-science and open science are factors transforming health research, healthcare and health policy. The COVID-19 pandemic interacted with the former and revealed some deficiencies of the latter. This has in turn stimulated disciplinary clashes, interdisciplinary elaborations and political conflicts, concentrated around the techno-scientific, political and societal changes needed to cope with the complex health problems that were and will be challenging our societies (inequity, poverty, pandemics, climate change, reemerging diseases, etc). Herein, as a continuation of an interdisciplinary dialogue initiated in 2018 with the ―Data and Information Sciences Applied to Human Health‖ project (CIDASH for short), we put forward the idea of planning a data-centric, open, health research infrastructure, interoperable with the integral national health system and with national health registries. We aimed to identify and analyze from interdisciplinary and trans-disciplinary perspectives and based on our experiences in Uruguay the opportunities, concerns and challenges regarding the implementation of an open health research infrastructure (OSHRI). Our specific objectives were: i) to provide a brief account of our experience in human health research in Uruguay; ii) to reflect on the perceived opportunities, challenges, issues and potential solutions in human health research from the scientific, methodological, technical, ethical and regulatory point of view; iii) to elaborate on and propose structuring pillars for future open data infrastructures for human health research in Uruguay. The methodology we followed to report our experience was qualitative, interpretative and deliberative. Results include a list of problems and design pillars identified in relation of the creation of an OSHRI. To discuss these results we took the Centre for the Integration of Health Data and Knowledge (CIDACS) in Salvador de Bahía (Brazil) as an example of an installed and running open science health research infrastructure that warrants access to population-size data with proper ethical and scientific standards. After deliberation on the above-mentioned experiences we found that this report could bring us a step closer in the way of engaging with participatory research and change from within the science community.


I. Introduction
Data centrism, big data, precision X (X can be almost anything, medicine, management, etc.), data science (including autonomization of algorithms) and open science are a few processes that are currently impacting health sciences and healthcare at a global scale (Leonelli, 2016;Yu, Beam, & Kohane, 2018). These processes interact and generate philosophical, regulatory, technological, economic, political, governance and social problems. While there is an obvious need for critical evaluation of results, as well as for participation on these processes aiming at promoting social control and well balanced impacts, mounting evidence suggests that they can benefit from the creation of health research infrastructures (Hummel & Braun, 2020).
Data-centrism in scientific research brings new challenges and opportunities. One well established data-centric science model is the -consortium‖ in which data sharing regulation procedures emerge from consensus and confrontation between parties (Leonelli, 2016). Consensus typically concern the modalities of use of technologies that need to be shared within the consortium and then further disseminated, developed and expanded with the active participation of data producers and users over the globe (Leonelli, 2016). Governments require national statistics (an ancient form of data-centrism). National health registries  27(1), 202227(1), , pp. 7-54 ISSN: 230127(1), -1378 are organizations which are legitimated by law to collect and prepare clinical data on specific health problems with population width repurposing it to provide unbiased epidemiological descriptions and surveillance, in order to inform policy and the public. Although their main commitment is with the national level, they collaborate with similar organizations, frequently constituting international networks. Nevertheless, different registries within the national level are not necessarily interoperable and the stored data do not travel from one to another jurisdiction or are linked together. Neither, health registries are able to contribute their population´s personal data to international public repositories for obvious national security, economic and political reasons.
Open Science (OS) is a social movement that promises to make scientific research and the benefits derived from it larger and more accessible to all levels of society, amateur or professional (Woelfle, Olliaro, & Todd, 2011). This movement, began in the 1990s as an open access movement, and more recently it has been promoted as OS, in all continents. Above all, this idea points to an improvement in individual and collaborative research processes, in their communication and reproducibility, in order to achieve rapid production, access to data and use of new knowledge (Babini & Rovelli, 2020). In general, a social movement is conceived as a -network of informal interactions between a plurality of individuals, groups and or organizations, engaged in political or cultural conflicts, on the basis of shared collective identities‖ (Diani, 1992). In addition, social movements interact with interest groups, coalitions and other organizations, of particular relevance with public universities. Hence, the OS agenda and values both overlap and collide with those of Universidad de la República, which has been mandated to increase, spread and defend culture, promote and protect scientific research, to be in charge of the qualification for the exercise of scientific professions and to contribute to the study of problems of general interest and promote their public understanding (Uruguay, 1958). Therefore, OS is creating new problems, challenges and risks, particularly in the biology, sociology, technology and health domains related to research quality evaluation and promotion, intellectual creation and property rights and knowledge dissemination (Holzmeyer, 2019;Randall, 2021;UDELAR, 2020).
Informatio 27(1), 202227(1), , pp. 7-54 ISSN: 230127(1), -1378 Studying health and healthcare problems through the re-use of population size sample data disseminated through massive communication media was a trending mode of research by ad-hoc and already constituted multidisciplinary research teams, by isolated citizens and groups performing citizen-science like activities as well as things in-between, in attempts to better respond to the COVID-19 pandemic (S. Méndez & Botti, 2021). A major problem we observed is that not all these researchers and citizens engaged in HR during the pandemic were aware and compliant to regulations of research on human subjects, which means that a greater and perhaps different effort from the academy is needed to achieve responsible science in an open and citizen science scenario or in future national emergencies. In addition, we observed that no significant regulatory or procedural changes better responding to new contexts and bioethical issues were elaborated during COVID-19 national emergency (S. Méndez & Botti, 2021). Similar tensions and mismatches were reported to take place in all regions and some authors have proposed that changes in the research evaluation system were needed in underdeveloped countries (Aarons, 2018). In 2018 we started an interdisciplinary dialogue centered around the applications of information and data sciences in human health (CIDASH, 2019). Herein, we continue that interdisciplinary process, that was shaken by the COVID-19 pandemic, proposing to openly start the critical design a Health Research Infrastructure (HRI) in Uruguay, potentially integrating a new health research commons that could bring support for better science, including the promotion of more legitimate, ethical and secure re-use of sensitive personal, health and administrative data (including population size data samples). We wish to consider ideas, problems, etc. related to an HRI that could in principle be compatible with values that we speculate OS and Universidad de la República could share in relation to health research.

II. Objectives
The main objective of this work was to identify and analyze from interdisciplinary

III. Methods
Methodology. The methodological approach herein employed is qualitative. An experience report is described by Daltro and de Faria as a postmodern scientific narrative (Daltro & de Faria, 2019). This concept is especially significant to us, because the experience report methodology departs from strict positivist (modern) approaches and can overcome some of their limitations. In agreement with Daltro and de Faria (Daltro & de Faria, 2019), we think that studying complex real systems involving humans requires accepting the possibility that researchers form observer-participator junctures and therefore, the subjective, perspectival and dynamical nature of the knowledge created through the report of scientific experiences must be acknowledged. Therefore, elaborating generalizations is out of focus in experience reports, rather contextual insights and positioning are probably among its valuable products. Communicating the reader about reasons for key choices and reflections that characterize the experience so far is highly valuable as well. Arriving at conclusions in experience reports is probably not a certain yield. In fact, Daltro and de Faria propose that conclusions should not be  27(1), 202227(1), , pp. 7-54 ISSN: 230127(1), -1378 part of experience reports (Daltro & de Faria, 2019). In that respect we depart from their view. We think conclusions are possible and valuable if correctly communicated and understood as contextual, historical and dynamic.  (Atkins et al., 2003) and has overlapping meanings with RI (Anderson, 2013;Stewart et al., 2010). A Knowledge Infrastructure (KI) is also a related concept, emphasizing the value of knowledge creation and dissemination, which may be more connected to the profile of activities of universities and data-driven research domains (Borgman, 2019). We herein take the European Commission definition of RIs as sufficiently flexible to encompass the others.
Open science has been defined as -the movement to make scientific research (including publications, data, physical samples, and software) and its dissemination accessible to all levels of an inquiring society, amateur or professional‖ (Woelfle et al., 2011).

IV. Results
Brief account of the experience-shaping research and development activities.
The authors participated in five activities that can be considered central for the and 2) to contribute to the development of intersectoral infrastructures necessary for quality inter/disciplinary research in computer, data and information sciences applied to human health. This experience report is one way of advancing in the last direction. CIDASH-II focuses also in bridging the humanities-sciences gap, which is reflected in the proposal at many levels, including its institutional support. CIDASH-II was paused because of funding difficulties and the prioritization of working on the COVID-19 pandemic. CIDASH-II will be  the URUGENOMES project (Naya, 2021), which obtained non-refundable financing from the Inter-American Development Bank (IDB). Importantly, the project helped to unveil a part of the history of Uruguay, shedding light on Uruguayan ancestry genomics, which is also key for understanding genomic variability in Uruguay and informing clinical genomics (Spangenberg et al., 2016). The URUGENOMES project contributes to science and healthcare in the field of rare diseases (which also serves human resources training  Urugenomes project. The results showed that more than half of the cases analyzed were diagnosed at the molecular level from the complete genome (Raggio et al., 2021;Spangenberg et al., 2021;Spangenberg et al., 2016;Spangenberg et al., 2019). International figures show diagnostic rates of 41% with whole genome sequencing (WGS) and 36% with whole exome sequencing (WES) (Clark et al., 2018) and a study has shown that the use of whole genome sequencing (WGS) can have a diagnostic success of 62.5% (Liu et al., 2019). individuals. Adequate social control through, for example external ethical evaluation is not considered a structural limitation (see below). These kinds of issues can self-perpetuate in pernicious contexts putting at risk the quality of research, the performance and behavior of researchers and the health and wellbeing of all, principally of the most vulnerable and those already harmed (vulnerated) (Kottow, 2012). We felt ethically compelled to work on uncovering structural limitations in health research education, social control and research infrastructure.

Issues in social control of health research and health research education.
Social control of science is sometimes perceived by researchers as a structural limitation (Wolpe, 2006), but those constraints are purposed at protecting individuals, institutions, collectives and the society as a whole from researchrelated risks and harms. On the other hand, social control and bioethical evaluation are not monolithic or stationary and must respond to new issues and contexts. We understand that the poor education of researchers is a structural reason for disvaluing the role of social control in health research but also that there are deficits in regulation. These issues are intertwined. The quality-ethics evaluation processes in Uruguay is suffering from several problems in our understanding. On one side, Uruguay has a fragmented regulation covering many facets of health research (MSP, 2008;Uruguay, 2006Uruguay, , 2008aUruguay, , 2008bUruguay, , 2009aUruguay, , 2010Uruguay, , 2012Uruguay, , 2014Uruguay, , 2019Uruguay, , 2021, etc. and the application and control of the actions implemented under these regulations is heterogeneous (with respect to institutional support, accreditation status, integration, etc). It must be stressed that the performances and characteristics of research ethics committees vary widely in Uruguay as well as in other countries, with also varying roles for central structures (Vidal, 2017 were formal accreditation of animal research scientists currently exist (Uruguay, 2009b). An additional cause of educational heterogeneity arises probably because the programs at the graduate level in different careers may not provide today the experience in human research needed. Therefore, the system is highly heterogeneous, both at the researcher/research team and external evaluator/research ethics committee sides. Another aspect that deserves special attention in Uruguay is the lack of an effective research follow-up process on participant safety and data security after the initial ethical approval, which is a established activity of the research and innovation system in more developed countries.
We suggest that harm associated with risks identified in relation to research data and information management, processing and/or analysis must also be more stringently addressed in order to protect participants and ensure that researchers, institutions and funders remain accountable, as the responsible health research principle indicates. However higher stringency can only be effectively achieved without compromising the volume of research after increasing human training opportunities (formal education and proper experiences), better research infrastructures and adequate incentives. Instruments for participant safety and data security monitoring requires public investment, probably in several forms (UNICEF, 2005). The case of the creation and functioning of an Open Science Health Research Infrastructure will surely require a specific law with derogative effects on several regulatory texts, but more importantly, will give to academics and the government the opportunity to co-design and co-run an OSHRI, one of the most relevant pieces of the health knowledge generation machinery, one that has a high structuring potential, but nevertheless, just a piece of a system.
We think that to co-design and co-run an OSHRI, there are risks and risks balances we need to better understand. Particularly, those involving the already recognized personal, group, community and social risks associated with either a) not pursuing health research on humans (Jones et al., 2017), b) performing health research extrapolating Mertonian norms (universalism, communism, disinterestedness and organized skepticism) (Merton, 1973) or c) controlling human and health research in different ways (including but not only, independent bioethical evaluation). These balances can be considered either in the traditional that are typical of nascent true interdisciplinary work, which we think should be followed from a slower and safer learning curve. The health research and healthcare communities have already acknowledged the problem of biased databases (sampling bias), artificial intelligence/data science algorithms and resulting models (Knight et al., 2021;Panch, Mattie, & Atun, 2019). However, a key understanding we wish to provoke is that a bias-aware Open Science community is not enough. We agree with other authors who had already concluded that the previous -anonymize, release and forget‖ (ARF) praxis concerning personal data handling should be abandoned (Lomas, 2019;Nowakowski, 2016;Ohm, 2009 (Franks, 2020).
Issues deriving from a low prioritization of research activities in the academic, public and private health sectors and an avid global and/or professional market for engineers and PhDs. Stronger public funding and strategic academic and public research agenda building are critically relevant for a more productive government-academy interaction, particularly when it comes to build new better facilities and to utilize knowledge for health and development.
Unfortunately, raising levels of investment in science in Uruguay has proven to be very difficult, and will probably be a limiting factor in the coming years; therefore, global and regional funds could be needed. It is widely acknowledged that a RI proposal has to be evaluated, in part, in terms of its costs, benefits, and sustainability (Florio, 2019), but there are other dimensions that, in our opinion, need the same level of attention. In addition to the identification of medium-and long-term health problems and health objectives, work must also be done to identify the possible rationales and general characteristics of the projects that will be supported by the HRI, including reflection on the multiplicity of health research and action paradigms that are available and most valuable (see below).
To this end, it is necessary to consider already advanced experiences as well as relevant characteristics at the local level (i.e. the structure of our health system, our geography, epidemiology, culture, etc.), with internal and external histories.
We are convinced that planning a HRI must also explicitly include from the start the consideration for the health, human development, bioethical and biopolitical issues and challenges that, as a double-edged sword, the HRI will address and generate. Critical to any HRI is the founding human capital and a strong user community, and both may well be scarce in Uruguay (see for instance (L. Méndez, Pellegrino, Robaina, & Vigorito, 2019)). Among related phenomena, we consider that the -Brain Drain‖ that affects the academy, that is researchers and professionals (including data administrators and experts in new computing processes) either leaving the country, leaving early the academic circuit to the private sector or working for abroad (directly or indirectly), will be a great limitation that should be correctly addressed. Repatriation and/or nationalization has to be engineered, tested and adopted (mostly in unintended or forced ways) by end consumers. One aspect of this prominent change might be the imposition of Open Data Science itself. We note that some reports dedicated to criteria that should be observed prior to opening government data do not consider bioethics and biopolitical issues at all (Yang, Lo, & Shiang, 2015 the HBP above) and risks in particular subject areas (Herington & Tanona, 2020).
Traditionally, independent evaluation of proposals for health research on humans follows independent ethical reflection and deliberation by members of institutional research ethics review boards (RECs), in which accepted general principles and rules of wisdom are used for tailored case-specific decision making. Under this framework, the -first do no harm principle‖ with roots in the Corpus hippocraticum (-to do good or at least to do no harm‖) (Craik, 2015) means that the risk of (physical, moral, psychological, social, legal, and financial)  , 2005). While this and other principles are widely accepted in the health research on humans community, a turn to the -do no net harm‖ is being processed in biological and social sciences and in evidence-based social action only in recent years (Davies & Bawa, 2012;Putzel, 2010). What some authors propose is that extending the traditional bioethical evaluation by RECs to projects into the new contexts of research on humans (i.e. big data, complex problems) should be done but is and will be increasingly less encompassing (Herington & Tanona, 2020;Vayena & Blasimme, 2021). Harm to third parties due to the findings of research are also of concern (Hausman, 2007), but less appreciated in the literature. It could be the case of a review of the literature or studies using animal models. This might be a relevant problem deserving research given that as scientists we are currently exposed to and misguided by wrong incentives (i.e. the -publish or perish‖ normative). The current practice conditions that these kinds of research risks can only be evaluated by peers before publication, an evaluation process which concentrates on the scientific validity of the work already done and does not take explicitly into account the possibility of harm to bystanders, social groups or the society as a whole before and after publication. Researchers, reviewers, editors and scientific journals are currently unaccountable for harm to non-participants due to research findings (which may be even more important if using biased datasets and/or black box automatic methods). Some authors have agreed that institutional research review boards are not prepared for the evaluation of third party risks (Hausman, 2007;S. K. Shah et al., 2018). Also, the do no net harm principle in some instances requires that the risks to which participants are exposed should be outweighed by social benefits (the social value of a research).
It has been proposed that the social value of a research project is something often difficult to weight and well beyond the capabilities of most institutional review boards (S. K. Shah et al., 2018). Social control of science is not stationary as we mentioned above. Consider for example that until recently, epidemiological observational research was widely regarded as not raising significant ethical issues and was commonly carried out without approval of an ethical review committee, but that has changed already (CIOMS, 2009). So looking ahead, expanding the scope and improving the systematization of project evaluation and monitoring in a few more dimensions is an eventually relevant direction worth following for the sake of health research and open science".
Issues in the access to the benefits of health research. The inequity in access to the benefits of health research will likely continue to increase. The majority of large-scale genomic research projects have been conducted in Europe, the USA and Canada. Therefore, we predict that a large inequity in access to this type of analyses and knowledge will continue to develop in Uruguay, which compels action. In fact, it has been concluded that the benefits of genomic studies hardly reach patients around the world equally. Even in developed countries, low-income  Emanuel et al., 2008). Since a single and certain answer to this question may not exist and the evaluations involved might be difficult (Herington & Tanona, 2020) particularly in resource-limited contexts, the issue requires collective compromise and reflexive discussion, rather than avoidance.
The privileged place the health research community gives to ethics and the fact that OS has been considered a necessity for confronting with professional ethics issues (i.e. in relation to climate change data and knowledge) (Cai, Judd, & Informatio 27(1), 2022, pp. 7-54 ISSN: 2301-1378 Lontzek, 2012), may be indications of a larger than expected compatibility between OS, health research and health policy. Recent experiences related to the COVID-19 pandemic give us a lot think about in these respects. Academies of Sciences & Medicine, 2017); 2) factors related to human agency, individual choice, will, self-determination and control (Deci & Ryan, 2008;Tsey, 2008); 3) biological factors from within the individual or in relation to hostparasite relations, mostly genes and genomics that determine general or differential susceptibility (or virulence); 4) physical and biological environmental factors (National Research Council, 2006)  These models can be deterministic, stochastic or mixed. -X determines Y‖ is sometimes used in place of -X causes Y‖, but the former should be read today as -X causes Y deterministically‖ (not probabilistically or with probability of 1). The explanation for that can be traced back to Hume who only considered deterministic causality, hence our inherited notion generally assumes deterministic relationships, homogeneity and then law-like behaviors (Dupré & Cartwright, 1988;Neumayer & Plümper, 2017). Causal stochastic models may not rely on explicit detailed chains of events and cause and effect relationships and cannot predict at the individual (single event) level (Pearl, 2009).

Pillar 3. Provide bridges for the health data repository archipelago and
provide a "hub" for integrative knowledge creation. The advancement of genomics (regionally as well as globally) requires a quote of scientist modesty, and genomic data alone (without identifiers) under the category of -individual identification code information‖ -a category reserved for identification information, such as passport numbers, fingerprint scan data, social security numbers, etc.- (Yamamoto et al., 2018). Interestingly, direct-to-consumer genetic testing services were becoming more common in Japan at that time, and some researchers and medical professionals were worried about how personal information including genomic data accumulated at those companies would be handled (Yamamoto et al., 2018), a process which is happening now in Uruguay.
In parallel to regulatory changes securing personal genomes, Japan was also building biobanking infrastructures promoting genomics and precision health at the bench, computer and bedside (Nagai et al., 2017;Takayama et al., 2021). We think personal genomic data is identificatory per se even after anonymization, and therefore it shouldn't be treated as open data or be freely shared by researchers.
The path followed by Japan is of course not the only possibility; there are different flavors of biobanks and genomebanks (O'Doherty et al., 2021). In our opinion, the advantage of integrating genomes and EHRs is huge if the information system is properly developed, because it helps bridging the genomephenome gap (i.e. the identification of potentially causal relations) (Linder et al., 2021). Recommendations on how to integrate genomic (and in general omics) data into EHRs have been produced and some implementations following them are available (Grebe et al., 2020;Lau-Min et al., 2021). Multiple, small (institutions) to bigger (public/private sector) jurisdictions within a country will clearly generate difficulties. Interoperability constrains are typically the higher barriers in home grown EHRs (EHRs developed and tailored locally to each health institution, a characteristic of our EHR system). Using genome and phenome data has two extreme solutions; either i) participants in a diseasespecific cohort consent and renounce to their right to privacy, and data is shared by selected researchers with clear purposes (project evaluation and monitoring needed) within a national or international system (Chan et al., 2017;Network, 2021); or ii) data is protected from researchers but used within a federated secure edge data sharing and computing research system, where participants are not asked to consent any rights renouncement (at bigger implementation costs, but lower harm risks) (Rehm et al., 2021;Voisin et al., 2021).   (Rehm et al., 2021;Sanmarchi, Toscano, Fattorini, Bucci, & Golinelli, 2021). The OSHRI could provide the data integration, processing and analysis environment needed for secure/safe and ethical open health science without personal data dissemination.  (Foucault, 2001(Foucault, , 2007(Foucault, , 2008. In his works, biopolitics appears as a concept that refers to the way in which the  (Foucault, 2000(Foucault, , 2001(Foucault, , 2007(Foucault, , 2008. Surveillance and control of birth, disease, life expectancy, mortality, migration, and today we would add human mobility become part, according to the author, of a form of exercise of power, a "biopower," and a fundamental pillar for the development of the capitalist production model (Foucault, 2001). for this process and that were developed by the author in his genealogy (Foucault, 2001). In the XX century some ideologies supported genetics as the basis of the purity of a race and "quality" as criterion for descent took the stage. Under the supremacy of one race over another, crimes have been perpetrated, in the name of that sought-after good. In this sense, we understand that, depending on the definition of health and criteria of normality from which these parameters are access is widely known to be dramatically inequitable and globally dangerous (Herzog, Norheim, Emanuel, & McCoy, 2021). Likewise, this scenario promoted a so-called open science policy, putting it at the center of the debate (Molldrem, Informatio 27(1), 2022, pp. 7-54 ISSN: 2301-1378 Hussain, & Smith, 2021). We can identify in the OS proposals the principles that guide it: the conception of knowledge as a public good, openness of access to publications and data managed by academia in conjunction with government institutions, and the absence of profit (Babini & Rovelli, 2020). fair subject selection, favorable risk-benefit ratio, independent evaluation, informed consent and respect for enrolled subjects were systematized by Emanuel (E. Emanuel, 1999) and are part of a rational framework for the ethical analysis of biomedical research with human beings. According to the author's proposal, they are all universally applicable and are ordered in a logical sequence, constituting declarations of value that must be specified in each case and particular context.
We will focus especially on two of them that may shed some light on aspects that seem central to this reflection: value or relevance and the independent evaluation of research proposals.
The value of an investigation represents a judgment on its social, scientific or clinical importance where the contribution to the health, well-being or knowledge of the population should be valued from the impact of its results and the collectivization of its benefits, a requirement that stems from the need of a responsible use of finite resources and the imperative to avoid exploitation (E. Emanuel, 1999). In other words, the estimation of the value of an investigation does not correspond to scientific validity, a necessary but not sufficient requirement. Good research from a methodological point of view is not necessarily ethical, relevant or good quality health research from this perspective.
According to Emanuel, independent evaluation is a requirement that is based on the fact that researchers have the potential for conflict of interest and therefore they must remain socially accountable (E. Emanuel, 1999). These conflicts of interests not only relate to lucrative aims, but can be associated with academic prestige, struggle for funding and positions of power or respond to corporate pressure (E. Emanuel, 1999). Although many of them may be legitimate, they can affect various aspects of the development of an investigation. In order to minimize these biases, the independent evaluation seeks to ensure that participants are treated ethically, preventing them from being used as mere means (human exploitation) and preventing damage or the disclosure of sensitive data and information which should remain confidential (E. Emanuel, 1999). Of particular importance in the context of this reflection, Emanuel suggests that in addition to scientific review boards and research ethics committees there should be data and safety monitoring boards (E. Emanuel, 1999). These instruments that should give support to health research have not been implemented in Uruguay yet. It is, in our opinion, an ethico-political imperative to build these competencies, something the Universidad de la República, the national government, and the research system as a whole should acknowledge.

Science has been internationally and nationally agreed to be a Universal Human
Right. It is included in the Universal Declaration of Human Rights in its article 27 (Nations, 1948) 27(1), 202227(1), , pp. 7-54 ISSN: 230127(1), -1378 According to this moral minimum of respect for human rights (Cortina, 2000), new developments and ways of doing science should not only pursue the technological imperative, but also pose a challenge of new approaches to think about the values that are at stake in this social practice. We understand that these values are present in Open Science as an instituting proposal, however, as we have tried to show, various tensions and procedural aspects arise that, as a society, we will have to rethink in order to achieve the desired objective. In a similar direction, the recent UNESCO document recommends to consider the dialogue with other systems of knowledge and ethics as a pillar for OS development, specially including the respect for the human rights of indigenous people (UNESCO, 2021), as declared in (Nations, 2007). Developments such as an OSHRI including infrastructures for storage, preparation and analysis of genomic data entail dilemmas and particular ethical problems for the different areas -see Art 3. in (UNESCO, 2005). Finally, under the bioethical principle of responsibility (Jonas, 1985), the human being must act with precaution when deciding which technological objectives to pursue, avoiding decisions based on the technological imperative alone, in other words: not everything possible must be done. Of course, this requires profound reflection. Nevertheless, it might be agreed that this imperative of responsibility might be more prevalent now that there is ample evidence of an endangered future for humanity, as recognized in (UNESCO, 2021).

V. Discussion
No practice takes place in a philosophical vacuum, and all artifacts have politics.
This markedly applies to science and scientific technologies. This work makes some emphasis at analyzing human genomics as a techno-science in connection to health that is both increasing and closing inequity gaps. We agree with the view that we cannot expect to decrease inequity significantly if not acting on the causes of the causes (M. Marmot, 2018). Socio-economic factors are widely and reasonably considered the most relevant factors causing health variability, but by no means the only factors that matter. In addition, we expect that socio-economic factors cause both deterministic (for which we would assume homogeneity and known mechanism), stochastic effects (which we may assume are too complex to be explained and predicted) and perhaps also mixed effects. We should ask: does it matter? We think yes, but of course not knowing the nature of the effect and the mechanism of production does not mean we have our hands tied, we should seek to expose and remove the well-recognized damaging cause, as Breilh claims (Breilh, 2013). In addition, as researchers we can and should be on alert, sensitive to unexpected or counterintuitive results (both positive and negative) at one or many levels, which would give evidence of how uncertain we should feel about projecting patterns and regularities learned in the past or in different geographies, which, in turn should incentivize research on causal mechanisms at local levels, avoiding too Humean approaches (Pearl, 2021).
One thing we wish to consider is that big data and artificial intelligence have been expanding exponentially in the last 20 years, nevertheless, the complexity of real (i.e. biological, sociological, health) problems is still perplexing. Should processed, organized and curated data waiting for accumulation and later explanation then be the main output of data-driven and OS genomic research?
One assumption behind data accumulation in huge electronic stores is that some patterns in data can only be recognized and extracted if more data is available, but what happens if new patterns accumulate too with more data (linearly or not) as the context is also rapidly changing or if, alternatively, more data brings even stronger biases? Therefore, critical research in careful data accumulation and preparation for research and state-of-the-art analytical methodologies are both critical shared necessities of the health research community now and in the future.
Traditional researchers that think critically know that the theoretical lenses we have been using and developing to look at data are imperfect. Skepticism about the scientific method and mainstream programs is not new to epistemologists. The reader may wish to consider that they might be reasonably skeptical about datadriven -artificial research‖ accomplishments and promises.
National regulation adjustments, ethical guidelines and the participative design of responsible research and innovation mechanisms will surely be needed in the near  . In fact, the linkage of various administrative datasets allows carrying out comprehensive analyses that each individual institution contributing their databases would not manage to realize independently. The differential brought by CIDACS has been the utilization of administrative data to study the impact of governmental policies targeting the social determinants of health (e.g., social protection policies) on  27(1), 202227(1), , pp. 7-54 ISSN: 230127(1), -1378 health outcomes, using individual-level data. In fact, previous data linkage initiatives in Brazil generally used this type of data to study the effects of common exposures on health outcomes (M. L. . The employment of administrative data for research can hold significant advantages (including large sample sizes, longitudinal structures, high population coverage and high data quality). Nevertheless, it faces frequent limitations (mostly related to data quality and missing data) (Harron et al., 2017) and a number of challenges. First of all, different datasets must be obtained from different government bodies on a routine basis, in order to maintain the linked datasets up to date. Secondly, an appropriate computational infrastructure needs to be in place, with the capacity to execute data preparation and linkage procedures on databases with millions of entries. The establishment of such an infrastructure requires a significant financial investment in both human and non-human assets. Finally, protocols must be established to ensure that the data remains secure and the privacy of the individuals whose information is being handled is preserved, in accordance with the legislation related to data protection (M. L. . In Brazil, with Federal Law 12.527/2011 (the Information Access Law), access to administrative data containing personal information for research purposes began to require ethical approval by institutional committees, followed by the formal authorization from the relevant government data system administrators. As research involving personal data requires the explicit approval of institutional ethical committees, whenever individual consent is not feasible, ethical assessments are focused on risk-benefit analysis, participants' rights, measures to prevent harm to and discrimination of individuals or groups, researcher responsibilities and the monitoring of approved research (M. L. . However, in 2018 the Brazilian General Data Protection Law was passed (Law 13.709/2018), and became effective in 2020. This legislation, which was largely inspired by the European Union General Data Protection Regulation, established a series of principles and privacy protection measures. One of these determines that any individual whose data is stored in a database can request from the data guardian, at any time, information regarding the sharing of his/her personal data, the criteria and procedures used in the treatment of personal data and request the removal (opt-out) of his/her data from the database in question (Maurício Lima Barreto, Informatio 27(1), 2022, pp. 7-54 ISSN: 2301-1378 Almeida, & Doneda, 2019). One of the strategies employed by CIDACS to preserve the privacy of identifiable data and curb the possibility of reidentification of de-identified data is to separate linkage and analysis processes (Harron et al., 2017). The Centre, in fact, has two separate environments for managing the data. On the one hand, the Data Production Centre is a secure room housing the computational infrastructure for treating and linking original identified databases. In this space are produced the de-identified/anonymized datasets that are then accessible to researchers in the Data Analysis Environment  Barreto et al., 2021). CIDACS databases are currently being mapped to align with The Observational Medical Outcomes Partnership Common Data Model (OMOP CDM), so as to allow the Brazilian data held by the centre to be explored within the scope of international comparison studies (Stang et al., 2010).
Clearly, CIDACS implements its own version of the FAIR data principles, one that we think is FAIRer than the original when considering the re-identification and manipulation risks, we as society have been building up and must face today.
First: CIDACS requires that the -Persons who wish to receive authorisation must: 1) Be affiliated to the institution or be identified as collaborators; 2) Present a detailed research project together with ethical approval by an appropriate Brazilian institutional review board; 3) Provide a clear data plan restricted to the objectives of the proposed study, and a summary of the analysis plan intended to guide the linkage and/or extraction of a relevant set of records and variables; 4) Sign terms of responsibility regarding the access and use of data; 5) Perform the analysis of datasets provided using the CIDACS data analysis environment, a safe and secure infrastructure that provides remote access to deidentified/anonymized datasets and analysis tools‖ (M. L. . Many of the problems we experienced in Uruguay (see results section) are considered. Importantly, the researchers are not allowed to share, have and much less to -own‖ the analyzed  (Savage, 2019). Some of the key questions the HBP asks are how human brainartificial intelligence interaction can help us to understand what consciousness is and to build intelligent robots to assist people with disabilities. HBP researchers are aware that knowledge and technology produced in these respects are dual-use research of concern (DURC): can be used for right or wrong, including dangerous applications (Savage, 2019). Notoriously, the ethical, political and societal implications of the possible new understandings of the brain and the mind are acknowledged by the researchers, who have given the project a particular  (Owen, Macnaghten, & Stilgoe, 2012). One aspect of this -responsibility by design‖ is that researchers must be transparent about research progress and any one can raise and register an ethical concern against the project on that basis, a concern with which researchers will deal with transparency in a process that involves an ombudsperson and the research ethics committee in charge of the project evaluation (Project, 2021). Beyond this particular framework, we believe that these and other strategies that allow -responsibility by design‖ and participation from the community should be considered when planning and running an HRI.  2021). This recommendation might influence financing for research projects in Uruguay and in underdeveloped regions, but will also require a shift away from considering their axiology and objectives but also the problems they bring to our society and the opportunities for social control that could be created there. It is generally agreed that bioethical reflection on moral problems and the values linked with health research with human participants is essential and this could and should be extended to other types of developments and social modalities of response to problems (i.e. OS, RIs). Two questions that we wish to consider first are from which perspective (paradigm and theoretical assumptions) can we approach the discussion and which are the hegemonic positions that are translated from the evaluative point of view? A partial look would be problematic, as it could lead us to focus only on aspects related to risks, for example, which must be included but should not be the only aspect of concern. According to some authors, open science necessarily brings increased risks to health research, to the point that we have to choose between OS and privacy (Dennis et al., 2019). Two particularly relevant risks, already discussed, are those of re-identification due to data linkage that allows databases to be combined (a process that is not prevented by anonymization) (Ohm, 2009) and that of manipulation by ill-intentioned agents with the potential to harm vulnerable people and/or groups (Wood, 2014 Uruguay is lacking RIs with a capacity comparable to that of CIDACS'. The Centro Nacional de Supercomputación (ClusterUY) provides the highest computing power available in Uruguay and is a relevant antecedent for an HRI.
ClusterUY must be differentiated from an OSHRI. ClusterUY is a high performance computing platform, which emerged as an academic initiative to provide services to solve complex problems in Uruguay (Nesmachnow & Iturriaga, 2019). The main precedent in Uruguay, and motivation for the development of ClusterUY, is ClusterFING, a project that successfully initiated the development of the paradigm of use and centralized management of computing resources. ClusterFING was a High-Performance Computing (HPC) infrastructure of the Facultad de Ingeniería (Universidad de la República). Its initial infrastructure was acquired in 2008, funded by the Comisión Sectorial de Investigación Científica (CSIC), Universidad de la República, Uruguay.
ClusterUY then arose, scaling the dimension of the computing infrastructure and extending the service to the national level (Nesmachnow & Iturriaga, 2019). The initial investment for the acquisition of the computational equipment was financed by the Agencia Nacional de Investigación e Innovación, through a call launched in 2012, and the counterpart has been provided by CSIC. The project is currently operating in a technically and financially self-sustaining manner. In this way,

ClusterUY positions itself competitively with similar infrastructures in Latin
America, whose services are nowadays accessible to researchers, scientists and technicians of the country, with remote access from anywhere in Uruguay, operating at different levels according to the needs of each problem (Gitler, Gomes, & Nesmachnow, 2020). While ClusterUY is excellently serving a big research and innovation community, it is centered around basic science and technological problems, and adequate solutions to problems like critical and responsible research, person/group respect and protection, integral project There have always existed tensions between social order, scientific inquiry and development (Atkinson, 1978;Nafziger, 2007). Social control (different definitions exist) is a social behavior identified with -how people define and respond to deviant behavior‖, and how we eventually convert ourselves to conformity (Black, 2014;Janowitz, 1975). Social control of research aims at preventing issues related to legitimacy, relevance, responsibility, validity, fairness in participant selection, fair risk exposure, respect of the individual, etc. and at promoting wise individual and collective behavior, tending to the conservation of the social order or to beneficial changes in it. Bioethics is -a discipline of applied ethics and a particular way of ethical reasoning‖ (Gordon, 2021). Wikipedia defines it as -the study of the ethical issues emerging from advances in biology, medicine and technologies.‖ Hottois explains that -Bioethics covers a set of research, discourses and practices, generally multidisciplinary and pluralistic, which aim to clarify and, if possible, solve ethical questions raised by biomedical and biotechnological research and development within the societies characterized, to varying degrees, by being individualistic, multicultural, and evolutionary‖ (Hottois, 2004). Bioethics certainly contributes to improved social control of scientific research, healthcare and public health. The ethical evaluation and monitoring of research projects by institutional review boards is a key aspect in HR. This last aspect (project execution monitoring) is only partly defined in the Uruguayan regulation and has implementation problems. In addition, data privacy, safety and security evaluation boards do not take part in the evaluation and monitoring of most health research projects. Therefore, to improve science, OS not only requires new research infrastructures (RIs) but also revised structures for social and technological control of research.

VI. Concluding remarks
This work, rooted at and inextricable from the reported authors' experiences can be understood as part of a trans-disciplinary process or methodology with similarity to critical and constructive design research, were the -researchers becomes a change agent who is collaboratively developing structures intended to critique and support the transformation of the communities being studied‖ and where -deep relationships between researchers and research participants‖ are highly valuable (Barab, Thomas, Dodge, Squire, & Newell, 2004;Bardzell, Bardzell, Forlizzi, Zimmerman, & Antanitis). Accordingly, although a new health research infrastructure (HRI) for Uruguay has not yet been proposed as far as we know, we conclude that the need, the characteristics of it and the problems it will produce can and should be thoroughly and openly evaluated. We think that Uruguay would benefit from critically designing a HRI, addressing current and projected (i.e. in relation to OS) necessary conditions as well as requirements under future uncertain scientific human and health research scenarios. We also conclude that further deliberations are needed before commitment agreements are signed within or between institutions on the way of creating an OSHRI. This article hopefully will help us share with other researchers, academics and stakeholders our current humble perspectives and experiences on this subject.
Considering the preceding statements, we think one key principle we can all accept is to promote and expand external and independent ex-ante evaluation by peers and non-peers of strongly justified health research proposals. Non-peer evaluation of research and technological projects is something that needs to be taken more seriously by the academic and non-academic community and by the national government. Strong justification would mean that data-centric projects must express the proposed or expected causal connections between research procedures and outputs with potential health improvements as well as with participant's and non-participants dignity, privacy and safety, because that is the general multiple objective of health research. Respecting the dignity, preserving the privacy and protecting the safety at person, group and whole society level requires actions both individual and collective. Individual action for health researchers includes assuming bioethics and biopolitics as a general basis for research and science evaluation and education. Collective actions include, in our opinion, a) planning the best possible OSHRI (one that: is oriented towards closing the inequity gaps concerning access to the benefits of health research and healthcare, ensures the best possible data quality, best possible technologies, gives strong justification for actions, promotes peer plus non-peer evaluation and bioethical and biopolitical reflection, practices socially acceptable and responsible by design health research); b) planning the best possible preparation for responding to a future pandemic or any other health or human crisis (one that includes bioethics and biopolitics among other aspects that needs to be addressed and are not addressed herein); and c) planning the best possible research evaluation system (one that: gives a better support for research ethics committees, is more transparent, effectively respond to health research requirements in times of emergencies, that improves education of scientists and professionals, and that ensures the best possible ex-ante evaluation, participant safety and data security monitoring).
During this work we held difficult deliberations that are still in progress. In them, the role of democracy, participation and respect for the individual and the relevance of the many forms of trust (in persons, institutions, law, democracy itself, etc.) needed for fruitful work and social life were really apparent. In this sense, it is important to recognize that the best design of a HRI will be useless at preventing its degradation and misuse if democracy is broken, institutions corrupted and ethics minimized.