Sage Journals: Discover world-class research

Abstract

This article examines heterogeneous forms of human relationalities with algorithms envisioned in the development of a public algorithmic system and their anticipated effects. To do that, we focus on the distinct shapes given to both technologies and people by discourses and practices, together with their underlying logics and associated values. Analysing the blog posts documenting the emergence of Omaolo, a digital platform for healthcare and social welfare in Finland, we identify two algorithmic configurations: the ‘service engine’, which aligns with the public administration goals of standardising social and healthcare services in order to provide financial benefits; and the ‘treatment facilitator’, which advances the prevalent goals of social and healthcare professionals preoccupied with the fulfilment of situated care needs. We demonstrate that each of the configurations has different implications for social and healthcare organisations in which algorithmic technologies are deployed, for the professionals working there and for the people seeking public support. While the service engine might seem to undermine the collective bases of public service delivery, the treatment facilitator evidently supports them. Our findings remind us of the importance of research endeavours that acknowledge the complex and creative nature of development work, and consider the various parties and interests involved, in an attempt to attain more caring arrangements for the uses of data and algorithmic techniques in the public sector and beyond.

Keywords

Algorithmic systems care services human-technology relations public health social welfare Omaolo

Introduction

In February 2019, the Finnish state-owned company SoteDigi, now known as DigiFinland, launched Omaolo, a service platform for public healthcare and social welfare. According to a promotional video, Omaolo was developed to support basic social and healthcare services, with a focus on facilitating people's interaction with the system when they might not even know what is troubling them or what they need. The initial release of Omaolo included symptom checkers, wellbeing check-ups and service need assessments, all deploying algorithmic techniques. Typically, these services comprise an online questionnaire, freely available to anyone with access to a browser, the answers to which are processed computationally in order to give an instant assessment of people's condition and instructions on how to proceed, either by means of self-care or by getting in touch with a professional.

Shortly after its launch, a SoteDigi project manager presented Omaolo at a seminar hosted by the Ministry of Social Affairs and Health. Contrary to the general perception of the service platform as a mere ‘IT project’, the project manager emphasised that those involved in developing it saw the initiative as centred on ‘the renewal of operating models’ within the social and healthcare systems. The primary goal was to improve the quality of social and health services in Finland by making them simultaneously more personalised for service users and more standardised for service providers. Since Omaolo enables data gathering, the project manager explained, it is also possible to follow the platform's impact, gather population-level data on the health and wellbeing of Finnish citizens and use the insights when planning care services.

Omaolo did not gain significant traction in its first year, although SoteDigi continued to develop the system, providing additional services. This radically changed in mid-March 2020, however, when the coronavirus symptom checker was added to Omaolo as part of the national initiative to slow the spread of COVID-19. By the end of 2020, the coronavirus symptom checker had been utilised nearly two million times in a country of five and a half million people (Jormanainen and Soininen, 2021). Notably, no public concerns were raised in Finland pertaining to the COVID-19 symptom checker, unlike in some other European countries (Calatayud, 2020; Kaun, 2020). What is even more striking is that the symptom checker has become widely accepted among healthcare practitioners, although previous studies indicate that the diagnostic accuracy of such checkers is questionable (Wallace et al., 2022). Unlike the Finnish contact tracing application, Koronavilkku, whose effectiveness was challenged by public health professionals despite its initial high uptake (Rannikko et al., 2022), the COVID-19 symptom checker remained uncontroversial throughout the pandemic.

The general acceptance of the coronavirus symptom checker in Finland raised our interest in Omaolo's algorithmic system, which we approach as a socio-material assemblage (cf. Bowker, 1994; Schwennesen, 2019) comprising ‘dynamic arrangements of people and code’ (Seaver, 2019: 419). A number of scholars have been interested in failure and breakdown as an antidote to the unilinear narrative of technological progress (Appadurai and Alexander, 2020; Jackson, 2014), but it is similarly important to explore the conditions under which algorithmic systems succeed in supporting public sector aims. Amidst cancelled projects and those ‘put on hold’, attempts to repair and renew the collective bases of public service delivery by introducing algorithmic technologies call for research that asks what forms of human relationalities with algorithms make organisational sense (Choroszewicz, 2024; Reutter, 2022).

To answer this question, we explore a weblog created with the explicit purpose of conveying to the public the different stages in the Omaolo development, including the design choices negotiated and made along the way, as well as practitioners’ responses.¹ This article is a continuation of our efforts to engage with algorithmic systems prior to their operations in the world in order to bring their nascent ideas and underlying ideologies into the conversation (Ruckenstein and Trifuljesko, 2022). The studied blog posts can be read as highly curated testimonies which, nonetheless, reveal the different parties and interests involved in the renewal of public care services. In their posts, public administrators and social and healthcare professionals actively scrutinise how automation might work for them and for the people with whom they work – a role that is becoming more commonplace in the context of digital transformations and restructuring of expertise (Bergquist and Rolandsson, 2022).

Our research is informed by earlier work that examines how datafication processes and developing algorithmic systems are locally mediated and responded to (Kazansky and Milan, 2021; Lehtiniemi and Ruckenstein, 2019; Ratner and Elmholdt, 2023; Reutter, 2022) within the broader trends of providing personalised public services and standardising professional practices (Hoeyer and Wadmann, 2020; Langstrup, 2019). Acknowledging the call for conceptual and methodological approaches that can ‘grasp the obscure developments, the cultural richness, and the vibrant creativity’ associated with processes of datafication (Milan and Treré, 2019: 321), we turn to the notion of configuration. Configuration, Lucy Suchman (2012: 48) argues, is ‘a device for studying technologies with particular attention to the imaginaries and materialities that they join together’. To capture the different relationalities embedded in the development of Omaolo and their implications for public service delivery, we strengthen our analysis with Annemarie Mol's work on the different logics shaping healthcare and welfare discourses and practices (Mol, 2008).

This allows us to argue that there are two distinct forms of human relationalities inherent to the algorithmic technologies involved in the development of Omaolo: one focused on the renewal of service models linked to public administration, the other on the renewal of treatment models associated with the care profession. Each of these human-technology configurations arranges elements of the algorithmic system differently, producing diverse effects. Our research reaffirms the need to investigate the complex and creative nature of development work. By engaging with the different relationalities and rationales expressed in the blog posts documenting the development of Omaolo, we can achieve a more comprehensive understanding of digital advancements and of their intended and unintended consequences.

A new paradigm: ‘Knowing’ public care services

The development of Omaolo is rooted in the current transformation of qualitative aspects of medical practice into quantified data – a process known as the datafication of health (Hoeyer and Wadmann, 2020; Ruckenstein and Schüll, 2017) – which ties in with questions connected to the datafied welfare state (Dencik and Kaun, 2020). Describing the latest developments in public administration, Margetts and Dunleavy (2013) talk about Digital Era Governance, while Karen Yeung (2022) speaks of New Public Analytics. Understood as the successor to New Public Management (NPM), this novel modality of public sector governance entails a move from NPM's focus on organisational disaggregation, competition and incentivisation to providing holistic services for citizens through technologically enabled reform (Margetts and Dunleavy, 2013). The ultimate goal is the creation of what Ben Williamson (2014) calls ‘knowing public services’, envisioned as a model that processes citizens' personal as well as aggregated behavioural data algorithmically so that the authorities may better respond to people's demand for support.

At the heart of knowing public services, Williamson (2014: 309) argues, is ‘a particular form of human-computer interaction in which public services are to be co-produced by citizens interacting with algorithms’. Placing citizens at the centre of public sector reforms and assigning them the role of co-producers of the services they use by virtue of their interactions with algorithmic technologies presents particular challenges to the welfare state. Lina Dencik (2022), for instance, argues that such endeavours affect the relationship between the state and citizens in two ways: they advance ‘an actuarial logic based on personalised risk and the individualisation of social problems’ and entrench ‘a dependency on an economic model that perpetuates the circulation of data accumulation’ (Dencik, 2022: 147). Attempts to datafy public services, therefore, fundamentally undermine the social conditions that enabled the existence of the modern welfare state in the first place. Furthermore, the introduction of data-driven algorithmic technologies to the public sector might directly jeopardise the values of decommodification, universal access and social solidarity upon which welfare states are built (Dencik and Kaun, 2020).

To deal with the erosion of civic values, critical data and algorithm studies scholars have advocated measures and policies that enable the continued development of the datafied welfare state. They include justice and non-bias, decommodification, data diversity and transparency in the datafication process (Andreassen et al., 2021: 210). These are important aims and should be considered in the development of public services that rely on the utilisation of data and algorithms. Yet such measures and policies approach digital applications on a general level that tends to take human-technology relationalities in algorithmic systems and their implications for the welfare state as a given. Finland, like other Nordic countries, has a long tradition of collecting data about its citizens and using various forms of data to improve governing (Alastalo and Helén, 2022), which means that new initiatives do not land on terra nullius but operate in a welfare state context where institutional relations and aspirations are already in place (Lehtiniemi and Ruckenstein, 2022; Reutter, 2022).

In what follows, we demonstrate how the goals of public administration, articulated in the studied blog posts, confirm the individualising tendencies in state-citizen relations. Indeed, the aim of digital services is to enable proactive self-care, reducing interactions with social and healthcare professionals (Hoeyer and Bødker, 2020; Langstrup, 2019; Lupton, 2013). Yet the materials suggest that Omaolo services are not merely built as a unilateral care solution for cost-cutting; rather, they also aim to enhance social and healthcare practices by identifying how personal data streams and communication could be activated in delivering public support. This kind of balancing of interests is familiar from other Nordic contexts as well (Reutter, 2022). As Torenholt and Langstrup (2023: 54) argue, algorithmic techniques gain legitimacy at the local level only if they support practitioners’ aims. Thus, through the studied blog posts, we can also observe efforts to strengthen rather than undermine the collective foundations of relations between the state and its citizens through the introduction of algorithmic technologies.

Configurations of humans and algorithms

The starting point for using configuration as an analytical device lies in paying attention to the distinct shape given to an algorithmic system by discourses and practices. This enables us to recover the system's constituent elements and examine how the agencies of humans and technologies are ordered in relation to one another and the effect of such arrangements (Suchman, 2012: 49). By foregrounding the constitutive imaginaries and materialities of an algorithmic system, we are not only able to understand how people and technologies are figured together but also how they might be reconfigured, that is, figured together differently (Suchman, 2007).

To gain a better understanding of the specificities of a particular human-technology arrangement and its effects, we also pay attention to the general rationales underlying the discourses and practices shaping algorithmic configurations. Within the context of healthcare and social welfare, two rationales have been identified as having primacy. One of them revolves around individual will, focusing on the act of decision making, which is why Mol (2008) refers to it as ‘the logic of choice’. The alternative ‘logic of care’ focuses on addressing people's needs, which encompasses activities that make daily life more bearable. As an ideal, ‘individual choice’ results from the process of weighing up the advantages and disadvantages of a particular course of action, conceptualised in standardised terms. Conversely, ‘good care’ is always situated, continuously being reinvented and adapted by people in their everyday practices, as it is a matter of ‘attuning the many viscous variables of a life to each other’ (Mol, 2008: 54).

Our empirical work led to attempts to specify the significance of ‘individual choice’ and ‘good care’ in the context of Omaolo, and the human-technology relationalities that they involve. This opened new pathways for examining the different effects that algorithmic technologies are envisioned to have on social and healthcare systems. By combining Suchman's and Mol's work, we identify two distinct configurations of humans and algorithms, which we call the service engine and the treatment facilitator. We suggest that the service engine aligns with the managerial goals of standardising social and healthcare services in order to provide financial benefits. The treatment facilitator, in turn, legitimises renewal initiatives by advancing the goals of the social and healthcare professionals preoccupied with the fulfilment of situated care needs (Torenholt and Langstrup, 2023). As we detail below, each of the configurations has different implications for social and healthcare organisations in which algorithmic technologies are deployed, for the professionals working in them and for the people seeking public support.

While the two configurations might seem contradictory, much like their underlying logics (Mol, 2008), they are not necessarily mutually exclusive. In fact, we suggest that incorporating both into Omaolo's development was crucial to offering functioning support in the highly dysfunctional times of the COVID-19 pandemic. Both standardisation and situatedness – respectively, promoted by the service engine and the treatment facilitator – are important for the organisation of public care services, indicating that the coexistence of different forms of human relationalities with technologies might be precisely what is necessary for repairing and renewing operating models within the social and healthcare systems.

The making of Omaolo

The origins of Omaolo can be traced back to January 2016 with the beginning of a national project titled ‘Self-care and digital value services’ (Fin. Omahoito-ja digitaaliset arvopalvelut) known by its abbreviation, ODA. To capture its aims more accurately, the project's name was later changed to ‘My digital age wellbeing services’ (Fin. Omat digiajan hyvinvointipalvelut), but the ODA acronym was kept. The name Omaolo, which refers to personal wellbeing, only surfaced during the final stages of the project. For most of the development stage, the initiative was primarily referred to as ODA, which is why the two names appear in the material presented below interchangeably.

The ODA project was led by the city of Espoo, situated in the capital region, while the project staff was also dispersed across 13 other municipalities and hospital districts throughout Finland. The project began with the inventory of social and healthcare services across participating organisations, followed by the development of collaborative technological solutions in its second year. This phase transformed local initiatives into 38 pilots, including multiple symptom checkers – precursors to the coronavirus checker – which explored and tested various features of the emerging service, reflecting the specific aims and practices of participating organisations.

As is usually the case in digital developments in the Finnish public sector, ODA was a product of public-private partnerships. In June 2016, following a competitive bidding process managed by a municipal procurement entity, ODA purchased the medical knowledge database and algorithms from the company of the Finnish Medical Society (Kustannus Oy Duodecim). Six months later, a consortium of two private companies (Mediconsult and Solita) was hired to ensure technical implementation, support and maintenance. Finally, data centre services were procured by a multinational telecommunications company (Telia) originating in Sweden.

While the local and regional authorities played the main role in ODA, the project also promoted the aims of the central administration. In fact, ODA was a flagship project for Finland's Ministry for Social Affairs and Health, which was attempting to implement goals outlined in a program published in 2015 by the newly formed right-centre government. The government program emphasised the creation of ‘customer-centric services’ in public healthcare and social welfare by ‘exploiting more efficient electronic services in self-care and counselling’ through ‘the better use of the opportunities of health technologies’ (Valtioneuvoston kanslia, 2015: 18). These aims align with the personalisation narrative (Needham, 2011), which is often used to justify public service reforms by promoting five intersecting claims: (1) personalisation improves people's lives; (2) personalisation saves money; (3) personalisation reflects how people live their lives; (4) personalisation is applicable to everyone; and (5) personalisation acknowledges the role of people as experts on their own lives.

In addition, ODA followed the broader trend of datafication of health and was designed as an implementation of the ‘Strategy for benefiting the usage of social and health care data’, launched by the previous, centre-left government's Ministry for Social Affairs and Health (Sosiaali-ja terveysministeriö, 2014), in cooperation with the Association of Finnish Local and Regional Authorities. Here, too, the emphasis was placed on making the social welfare and healthcare system ‘customer-centric’ by using data-driven algorithmic systems. It could, therefore, be said that ODA unites various public sector datafication endeavours across the political spectrum in Finland.

Originally planned for public release at the end of the ODA project in October 2018, the Omaolo platform was still undergoing testing at that time. Before the project concluded, development responsibility was passed to SoteDigi, which eventually launched Omaolo for public use. Despite this transition, Omaolo is above all a legacy of the ODA project. This connection is evident not only in the platform's design and purpose but also in the transition of key ODA's personnel to Sotedigi, thereby assigning particular importance to the Omaolo development phase.

Materials and methods

To investigate the distinct shapes and implications of the human-technology relations embedded in Omaolo by algorithmic configurations mobilising the logic of choice and the logic of care, we conducted an analysis of all 55 posts from the official blog of the ODA project, initiated by its lead manager in May 2016. The blog was actively updated until the development of Omaolo was transferred to SoteDigi in September 2018, with a final post published in January 2019, a month before the official release of Omaolo. This last post was the only one signed by the Omaolo team in SoteDigi. Several other posts were written by administrative staff from the central project office who were responsible for oversight of the ODA project and running the blog. Representatives of private organisations involved in building Omaolo and collaborators from the public organisations also authored a few posts; however, the majority were created by ODA project members representing municipalities and hospital districts across Finland. These contributions varied from individual to collective efforts, with some authors writing a single post and others a couple or more.

The length of posts generally runs to between 200 and 500 words. The posts appear exclusively in Finnish – the excerpts presented below are our translations. The content varies widely, but all posts relate to the overall progress of the ODA project or its localised instantiations across Finland. It was intended that the posts serve as project documentation, which makes them particularly suitable source material for analysing the development of algorithmic systems. However, rather than regarding these blog posts as unmediated representations of the ODA reality, we consider them to reflect their authors’ aspirations, providing insights into how things are perceived to be and how they should be.

Despite the blog's comment feature, only two out of 55 posts elicited responses. One comment, signed by ‘a line nurse in the public sector’, praised the bottom-up approach of the ODA project and commended the inclusion of nurses in the development work. Another comment, related to a post on oral healthcare, came from an ‘older person’ inquiring whether the ODA project would ease access to dental services. Neither of the comments led to further discussion. Therefore, given the lack of engagement, these blog posts appear more as fixed narratives than interactive exchanges, underscoring the presentation of different visions involved in Omaolo development rather than negotiating them.

In addition to the blog posts, we examined other documentary materials to gain a more comprehensive understanding of the ODA project. These include governmental policy papers, the Omaolo guidebook, PowerPoint presentations and electronic reports detailing the ODA project, and promotional and demonstration videos and recordings of seminars where Omaolo was discussed, all publicly available on YouTube. These materials have allowed us to navigate the complex realms of digital health and welfare development and to identify the numerous parties and interests involved in the renewal of public care services. However, for the purposes of our analysis, we focused primarily on blog posts, as they constituted the most coherent set of documents for examination.

The first author studied the posts for shared features in how algorithmic systems are promoted, and then analysed the distinct shapes that algorithmic technologies are envisioned to have in the context of public care services. Here, the focus was on explicit designators, such as moottori (Eng. motor or engine) and väline (Eng. means or medium), which blog post authors assigned to algorithmic technologies, as well as on the descriptions associated with such designators, even if the authors did not invoke them directly. This is how the service engine and the treatment facilitator configurations emerged. We then used these findings to examine the underlying rationales and assumptions the posts made about the consequences of the two identified configurations for social and healthcare organisations in which algorithmic technologies are deployed, for the professionals working there and for the people seeking public support. Bearing in mind the principles of abductive analysis (Timmermans and Tavory, 2012), we looked both for examples that would verify the configurations and evidence that challenges them.

During iterative discussions, we settled on the specifying features of the two configurations we identified, which often appear simultaneously in the material; authors can invoke both within the same post, depending on whether they focus on service provision or treatment delivery. For analytical purposes, however, we distinguish between the two in order to shed light on the different visions of public algorithmic systems within the Finnish healthcare and social welfare context emerging through the blog posts.

Advocating the service engine

The prevalent figure of technologies at the centre of Omaolo, according to the studied blog posts, is that of ‘an intelligent decision-making engine’, as the Head Physician of the Health Stations in Helsinki (Mäkinen, 2016) notes. The engine's ‘oxygen and steering mechanism’ are medical knowledge and the regulations of the social sector, while its ‘fuel’ is data about individuals. ‘The result is a proposal on what would be the solution to the customer service needs and what they should do’, the Head Physician explains in his blog post, which is why we name this conjoining of people and technologies a service engine.

The service engine mobilises the logic of choice (Mol, 2008), which, in general, positions individuals as decision makers, and algorithmic technologies as tools assisting them in this endeavour. In this particular case, people are figured as consumers of public services and Omaolo enables them to make optimal choices pertaining to their public service consumption by assessing their specific situation and suggesting a particular course of action based on such assessment. The overarching aim is one of efficient and frictionless living with computational tools (Schüll, 2018).

The primary motivation for introducing the service engine in the blog posts stems from identified problems within public service provision. The ODA's lead project manager outlines in the inaugural blog post (Nordlund, 2016) the existing difficulties in accessing health services – having to queue in line on the phone, waiting for a call back, then making an appointment, again waiting in line – before finally reaching a healthcare professional. The post concludes with a promise for the future:

In 2018, we will act in a completely different manner: we will have an electronic service package at our disposal, which gives us access to social welfare and healthcare services regardless of time and place. Things will advance and answers will be received without queuing, just when we need them.

The recurring mention of problems with public services in the blog posts is no coincidence. The overarching aim of reforming the health field with the aid of data-driven and algorithmic techniques has built on assumptions of organisational inefficiency, including unsystematic and incomplete use of customer data (Tarkkala et al., 2019; Torenholt and Langstrup, 2023). In a similar vein, the integration of databases is advocated as a generic solution for the challenges associated with knowing and governing individuals and populations (Ruppert, 2012).

In the public sector, however, customers are not just seen as consumers of social and healthcare services but also as citizens. Several posts claim that opting for the service engine would not merely deliver personally beneficial convenience, it would presumably also allow people to fulfil their civic duty by helping in the appropriate allocation of care services. When people turn to electronic services, it allegedly allows healthcare professionals to dedicate their time to critical cases and focus on ‘those situations in which their professional skills are most needed’, as suggested in a blog post written by members of the ODA team from Central Finland (Keski-Suomen ODA-projektiryhmä, 2016).

The way the blog posts emphasise the importance of reserving professional attention for the most urgent needs counteracts the notion that digital services reduce access to social and healthcare professionals. The goal of the service engine, the authors claim, is not to shrink public services, but to aid in reallocating expertise and human work to cases where their deployment makes most sense. Ideally, this reallocation includes methods like proactive self-care and diagnostics: predicting illnesses in risk groups or employing early warning systems in mental health care. A tension remains, however, as digital applications are typically introduced with the promise of cutting costs and doing more with reduced human resources, raising pressing questions about the quality and accessibility of care (Hoeyer, 2019; Rahimi, 2019; Torenholt and Langstrup, 2023).

In advocating the use of the service engine, the blog posts appeal time and again to the value of autonomy, treated as a precondition for making better personal choices (Mol, 2008). The service engine supposedly enables people to choose according to their preferences, as another paragraph from the above-quoted post written by the members of the ODA team from Central Finland suggests:

When we talk about the possibilities of electronic services, we often hear the objection that a computer ‘can’t treat me’ and, as a patient, I need to be in contact with those who can. We completely agree with that. It should always be possible to visit a healthcare professional if there is a genuine need for it. However, a lot of patients want to deal with their affairs when it best suits them. Not only while booking appointments but also by taking care of their health using electronic services.

In a related manner, as a configuration mobilising the logic of choice, the service engine is heralded for promoting equality. The blog posts envision how digital self-care services will address the needs of an ageing and dispersed population. Specifically, writing about their own municipality, Sodankylä, located in Lapland, the ODA project manager and coordinator (Hoppula and Eskola-Tuoma, 2017) elaborate in a joint blog post on how such applications could ensure access to public care for those who live in remote areas, have mobility challenges or face financial or logistical barriers. Notably, about half the residents in Sodankylä live in the central area, while the other half, mainly older people, still live in 30 peripheral villages, which have seen a gradual decline in services, including the shops, schools and post offices that used to keep the villages lively. These ideas have broader resonance, as Finland is the most sparsely populated country in the European Union, and one of the drivers of digitalisation is the need to ensure public service delivery for all.

Responding to the needs of the service engine

In order to function independently of time and location, the service engine needs to be figured as a discrete, self-standing entity. Nonetheless, while the engine is independent of its operational context, the context itself is profoundly influenced by the engine as it must enable the course of action that the algorithmic system proposes. As explained by the project and account managers of the Finnish Medical Society in their blog post, ‘[A]n electronic questionnaire alone cannot direct the customer to the services; the questionnaire needs to be backed up with a set of carefully planned services, which only social and healthcare professionals and organisations can produce’ (Lehto and Suurnäkki, 2016).

The blog posts describe the ambivalent effects of organisational changes on both healthcare and social welfare professionals and consumer-citizens. In order to serve the diverse needs of individuals, the authors argue, the service engine requires the support of organisations and skilled professionals in adapting to the engine's demands. At the same time, professionals are required to relinquish part of their control over their interactions with clients, as the engine increasingly dictates the terms of service delivery. ODA team members associated with the children's clinical and school healthcare in Espoo (Espoon lastenneuvolan ja kouluterveydenhuollon ODA-tiimit, 2017) suggest that the role of practitioners in service provision would diminish:

Before, the public health nurse ‘knew better’ and gave advice. In the future the client will lead the way and the nurse will offer support when needed. If the client is doing well and there is no need for support, the digital ODA survey could at times replace reception visits.

With this shift, the service engine has the greatest impact on public service customers. With the nurse stepping back into a supportive role, the client is expected to take a more active role in their own healthcare. The blog posts detail how the envisioned engine, with its aims of providing a ‘customised electronic service’, places explicit demands on consumer-citizens by opening up ‘new opportunities to find help and solutions to promote your own health’, as described by the ODA team members from a province in Southern Finland (Päijät-Hämeen hyvinvointiyhtymän ODA-ryhmä, 2017). The emphasis on self-care aligns with the neoliberal tendencies elaborated on in studies of datafication of health that describe how digital tools accelerate the withdrawal of the welfare state from citizens' lives, turning ‘empowerment’ into obligation (Lupton, 2013; Rich and Miah, 2014).

Furthermore, as public service customers, consumer-citizens are not only expected to interact with the service engine and follow its instructions, but also to contribute to its operation by ‘fuelling’ the engine with their data, a point highlighted by the Head Physician of the Health Stations in Helsinki (Mäkinen, 2016). In the analysed blog posts, however, the new role of public service customers as ‘data producers’ (Langstrup, 2019) is treated as unproblematic. Despite scholarly attempts to direct attention to the data politics that intertwine with the governance of our everyday lives (Ruppert et al., 2017), promoters of the service engine argue in a straightforward manner for the usefulness of services that ‘direct their users to find the right information, help solve problematic situations, give action recommendations’, as the procurement specialist in the company of the Association of Finnish Municipalities promises in his blog post (Ojala, 2016).

Before further discussing the implications of such human-technology arrangements, we introduce the second algorithmic configuration present in the analysed material.

Promoting the treatment facilitator

The alternative configuration of Omaolo emerging from the blog posts is that of a treatment facilitator that helps in ‘finding a common language and understanding’ within the social and healthcare system to meet people's broad and specific support needs, as the Director of Health Services in the City of Turku's Welfare Department puts it (Korkeila, 2016). In this respect, Omaolo's aim is to improve collaboration and enhance expertise. In doing so, it can ‘strengthen the continuity of care and dedication to treatment’, as ODA project participants from Oulu (Kriikkula et al., 2017), the largest city in Northern Finland, attested during the pilot phase.

In accordance with the underlining logic of care (Mol, 2008), individuals who engage with Omaolo, configured as the treatment facilitator, are positioned not as consumer-citizens but as people in need of care, even though they may still be referred to as ‘customers’. In this context, algorithmic technologies are envisioned as serving as mediators in the care process. The flow of information they enable between public service providers and receivers is considered from the perspective of improving treatment rather than merely optimising specific service outcomes.

The primary motivation for introducing the treatment facilitator, as discussed in the blog posts, stems from asserted problems in delivering public care due to a lack of communicative channels that would promote ‘a more holistic approach’, as the ODA project manager from Joensuu, a municipality in Eastern Finland, explains in her blog post (Kurki, 2016). She details the segmented way in which people purportedly discuss their lives: the doctor hears about health problems, the social worker about debt issues, and work-related matters are discussed at the unemployment office. Yet all these facets, the project manager argues, are intertwined with overall wellbeing and all matter when addressing care needs. ‘As the customer's real-life path usually crosses different administrative interfaces, we should develop services together with different organisations’, she concludes in her second post, published in the blog the following year (Kurki, 2017).

A physician from Tampere, a city in Southern Finland, further emphasises the importance of designing treatment that reflects the interconnected nature of a person's life and care requirements. He claims in his blog post (Karjalainen, 2016) that people often struggle to understand basic but crucial information about their health communicated to them by medical professionals. For instance, after years of treatment, people with diabetes might still fail to comprehend the relationship between eating and blood sugar levels. Since the general level of education, including digital literacy, is considered high in Finland, this lack of knowledge tends to amaze healthcare providers, the physician notes. However, he explains that illnesses in general are not widely understood, and people can be nervous and overwhelmed in care situations, especially when they first receive a diagnosis. To understand their condition fully, then, they need extra coaching, the physician argues, which is exactly what Omaolo is envisioned to provide.

Unlike the service engine configuration, the treatment facilitator aims to deal with the specific problems of specific people in specific circumstances. This is clearly articulated in a blog post collectively written by the ODA project team from Hämeenlinna, a municipality in Southern Finland (Hämeenlinnan ODA-tiimi, 2016). The post describes the fictional case of Veera, who gets symptoms of a urinary tract infection, and, as usual, calls the health station, which promises to call her back. Yet it is difficult for her to answer the phone because she works at the checkout of a large retail chain. Calls come that she cannot answer, and her anxiety increases with the felt symptoms of infection. When she finally manages to get a referral to the laboratory, after jumping through an additional set of hurdles, Veera must leave work early to give her urine sample, while continuing to feel uncomfortable, as she must drink a lot and repeatedly use the toilet.

With this example, the post mentions specificity and attentiveness as key values of the treatment facilitator; indeed, they are introduced as preconditions of good care, which extends beyond mere acts of diagnostic decision making to the entire continuum of care. Drawing on insights from various blog posts, the treatment facilitator queries what can be done to make a difficult situation more bearable. Yet the benefits of the treatment facilitator are presumably not limited to caring for specific people with specific problems in specific circumstances. ‘Trust in the service system grows, the feeling of security strengthens and the quality of life improves’, ODA project participants from Oulu (Kriikkula et al., 2017) point out in their assessment of the benefits of Omaolo as a treatment facilitator. Moreover, they add, those providing care become more motivated when they can exercise attentiveness. By promoting attentiveness and specificity, the treatment facilitator is, thus, understood to lead to both professional renewal and societal reproduction in which the benefits transcend current healthcare and social welfare systems; the overall aim is to ensure that the Finnish welfare society continues to thrive.

Strengthening existing care practices

In the blog posts, good care calls for improved communication and collaboration in attending to people's health and welfare, such as when helping a family caregiver living in a remote area to find support for her mother's difficult situation (Mikkola, 2017), or a social worker to deal with the issues of unemployed people in effective ways (Kurki, 2016). As with the service engine, Omaolo configured as the treatment facilitator is thus shaped by professional knowledge and practices. It is not, however, figured as a distinct entity or tool; rather, it takes different forms depending on the context of its use. Importantly, it might not even be of interest to decide beforehand whether a particular algorithmic setup is required. Because it advances an ethos rather than a standardised workflow, the treatment facilitator does not fit as comfortably into policy visions of the digital welfare society as the service engine. The promotion and development of the treatment facilitator evidently focuses on identifying ways to prevent and alleviate various kinds of troubles in people's lives, but it does not begin with the building of predefined solutions like symptom checkers or automated screening tools.

Rather than prompting profound changes in professional practices, like the service engine, the blog post authors argue that the treatment facilitator strengthens already existing organisational practices by enhancing informational exchanges between the different parties involved in the care process. The proclaimed goal is to establish readily accessible communication channels. Consequently, follow-up visits can be easily agreed upon and various support services, currently dispersed and difficult to find, can be combined, as the project team members from Kuopio, a municipality in Eastern Finland, claim in their blog post (Kuopion projektitiimi, 2016). By promoting continuity of care, the treatment facilitator configuration places algorithmic techniques at people's service rather than the other way around. Exemplifying this, the team members from Kuopio envision in their own post how a young person could fill out a wellbeing questionnaire before a check-up with the public health nurse so that the survey information would be available in the face-to-face meeting; the time otherwise spent on this task could be used to explore how the youngster is doing based on already provided answers, they conclude (Kuopion projektitiimi, 2016).

The post further envisions how it might be easier for young people to share difficult issues in an electronic format, which is not an unfounded expectation. As has been suggested by other studies, a bladder problem objectified in the form of a spreadsheet documenting bathroom visits renders aspects of the ‘private, subjective and somewhat inaccessible world of feelings and problems more tangible and comparable’ (Sharon and Zandbergen, 2017: 1705). When shared, the digital information – in whatever format – gives the professional the opportunity to tackle the most pressing issues in a dialogic healthcare encounter. Thus, the goals outlined in the post of the Kuopio ODA project team respond to the policy aims of promoting datafied and customer-centric services but emphasise that the gathered data needs to be used in a clinically meaningful manner.

Rather than being separate from clinical objectives, therefore, the treatment facilitator is envisioned as complementing and supporting them. Configured in this manner, algorithmic technologies ostensibly strengthen the role of practitioners. Here, we can explicitly point out why social and healthcare professionals might prefer such a development: it does not diminish their professional authority. As the ODA project planner from Oulu claims in her blog post (Hietala, 2016):

The development and implementation of ODA services is not intended as a replacement for face-to-face contact between the customer and the employee. They are new, modern services alongside the traditional ones. The ODA service provides an indicative action recommendation to the client, but the final responsibility to make decisions still rests with the professional.

Thus, the overall aim of the treatment facilitator, it seems, is not to create closed-off situations in terms of client-professional positions but, rather, to open the life of a care seeker to guidance. The facilitator should allow social and healthcare professionals to participate and intervene in people's daily lives, rather than the latter providing knowledge purely for the purposes of diagnosis. This kind of openness doubtlessly approaches digital care in a manner that affects all the parties involved; the human care providers are presumed to be still in charge of the situation, but the algorithmic system is envisioned as introducing elements that distribute some of the agency to the care seekers and the technology involved. In the proposed new division of labour, the blog posts call for nurturing conditions that enhance professional responsibility and judgement, along with the engaged usage of algorithmic techniques. Whereas the service engine configuration strives to reduce the role of professionals, the treatment facilitator aims at strengthening their ability to attune all the variables relevant in care to each other, in collaboration with the people seeking public support.

Delivering standardised service and situated treatment

Our study has employed the analytical lens of configuration to examine the different visions involved in the development and implementation of Omaolo. We have suggested that paying attention to the distinct shapes of human-technological relationalities, together with their underlying logics and associated values, helps us to gain a better understanding of the different goals of public algorithmic systems. As the above analysis demonstrates, we can distinguish between the two configurations of Omaolo (summarised in Table 1), noting important differences regarding forms that algorithmic technologies take, the activities in which they are involved and the organisational goals they promote.

Table 1.

Differences in the algorithmic configurations of public care services.

Algorithmic configuration	Service engine	Treatment facilitator
Form	Sturdy, independent of the context of its use	Malleable, dependent on the context of its use
Activity	Guides individuals to make appropriate choices pertaining to the allocation of public services by proposing a particular course of action based on personal data	Enables interaction between the different parties involved in the care process by gathering and distributing information about people's conditions
Goal	Standardised service provision yielding financial benefits	Situated treatment delivery effectuating comprehensive care

As a service engine, the goal is for Omaolo to offer standardised action recommendations that direct people to operate in a predefined, albeit personalised manner. The service beneficiaries become the engine's primary targets, and people are expected to act in accordance with the engine's proposal. This vision aligns with traditional standardisation efforts in clinical care (Berg, 1997), while also introducing disruptive tendencies. The aim is to streamline the provision of public healthcare and welfare services by taking over triaging work from practitioners and assigning self-care tasks to service users who should contact a social or healthcare professional only if they cannot care for themselves.

As a treatment facilitator, Omaolo is envisioned to strengthen caregiving and related developments in terms of public health and welfare. Care is not a matter of weighing up advantages and disadvantages, but a comprehensive process of attuning that calls for positioning algorithmic technologies as modifiers and enablers of care. The blog posts ask, for instance, how professionals can gain insights into aspects of people's lives that have previously remained unobserved or unaddressed. In an ideal scenario, the treatment facilitator enhances informational exchanges between the different parties involved in the care process, reframing caregiving in a manner that pushes social and healthcare professionals into new territories of negotiation, collaboration and discovery. Potentially, this can open innovative and unforeseen possibilities for the development of algorithmic systems. Initiating its input from an understanding of specific needs, the configuration of the treatment facilitator seemingly encourages professionals to be attentive in evaluating how technology could be adapted to become more responsive to what is required for caring social interactions in public organisations. Ideally, this would also open up avenues for recognising and cultivating emergent forms of care within algorithmic systems (Ruckenstein, 2023; Seaver, 2021).

The two configurations, therefore, have diverging effects on the context of their implementation (summarised in Table 2). Since the service engine has a sturdy form, it places demands on caregivers and care receivers to accommodate it. The care facilitator, on the other hand, is envisioned as a malleable technology and, as such, it can be adapted to strengthen existing care practices. Nonetheless, the two configurations are not necessarily separate or incompatible in terms of their overall goals, whether individual or collective. By improving the flow of information, as the treatment facilitator suggests, it is possible to prevent work tasks from overlapping, reduce unnecessary interpersonal exchanges, and accelerate access to services. Similarly, providing personal guidance with the service engine could enhance the quality of caregiving and strengthen the trust in the public services promoted by the treatment facilitator. The case of Omaolo, therefore, illustrates that the adoption of algorithmic technologies tailored to individuals does not necessarily jeopardise the public good (Nikunen and Hokka, 2020); rather, adverse outcomes are likely to stem from the demands of self-responsibilisation and the ways of employing algorithmic techniques to reduce much needed collective support.

Table 2.

Differences in the effects of two algorithmic configurations.

Algorithmic configuration	Service engine	Treatment facilitator
Effect on public organisations	Prompts changes of professional practices	Strengthens existing professional practices
Effect on social and healthcare professionals	Displaces professionals in certain cases	Highlights the role of professionals
Effect on people seeking public support	Places service demands on individuals	Distributes some of the agency of care to its recipients

In terms of protecting public outcomes, a key theme that emerges in the analysis of blog posts is the contextual understanding that guides specific development ideas of health and social care professionals. We get a strong sense that professionals in different parts of Finland are carefully trying to elaborate on what would work for them in terms of new technologies. This underscores how professionals' own preferences might differ from expectations of the generic value and performance of algorithmic systems (Lehtiniemi, 2024). Hoeyer and Wadmann (2020) call for an analysis of what professional judgement means in organisations that build practices based on the use of data and algorithms. They advocate a distancing from forces of datafication that aim to transform human ends to match technological means (also see Winner, 1978). Obtaining care through the service engine model requires that people must, to a large extent, attend to their own needs. From this perspective, data-driven algorithmic technologies within the public sector indeed become the means of furthering responsibilisation and rentierism, as Dencik (2022) puts it; the relationship between the state and the citizen weakens if it is mediated by technology that interrupts communication between care seekers and caregivers.

A thoughtful and caring combination of people and code, however, suggests that the latter development is neither necessary nor inevitable. The posts written by social and healthcare professionals offer contextual understanding that communicates an ability to envision how their work practices and judgement push back on any simple solutions. They explain how, by providing comprehensive public care through enabling information flows, data-driven algorithmic technologies could serve both social and healthcare professionals and people seeking public support. While this scenario appears idealistic in the face of pressure to govern through data and build new audit regimes (Dahler-Larsen, 2012; Power, 1997; Strathern, 2000), it is a concrete reminder that different paths can be taken in digital welfare service development. As Tamar Sharon (2015: 295) notes, scholarly criticism can obscure ‘the many ways in which individuals engage with healthy citizenship discourse that are not governed by principles of autonomous choice and that do not corroborate fears of normalisation and discipline’. By delivering personal care with the aid of data-driven algorithmic technologies that strengthen different capabilities and respond to diverse needs, welfare state values could continue to thrive (Kaun et al., 2023: 6). Our study suggests that one way to accomplish this is by means of expanding and multiplying the communicative channels between care seekers and caregivers and augmenting the role of professionals in the care process. For this to occur, however, algorithmic techniques need to be deployed carefully.

Concluding remarks

The inaugural blog post of the ODA's lead project manager promised that the initiative would bring about effortless access to social welfare and healthcare services in Finland. Nonetheless, although Omaolo has been up and running for five years, this vision has not materialised. Accessing public social and healthcare services continues to be an intensifying problem in Finland, while the implementation of customer-centric uses of data and algorithmic techniques is messy and incomplete rather than a well-planned and carefully considered process (Choroszewicz, 2024). The piloting that went into the building of Omaolo is not over; indeed, the platform is in perpetual motion, continuously evolving through new projects, the outcomes of which might remain questionable. During the pandemic, however, we could witness how the two configurations of Omaolo came together in the form of the coronavirus symptom checker. The service engine materialised as a diagnostic tool to query the possibility of an infection, while the treatment facilitator opened a communicative channel with care professionals. This dual task – in decision making and in enabling interaction – made the symptom checker an effective public health support at the heart of the pandemic.

With the help of the analytical device of configuration, strengthened through the work on the logic of choice and the logic of care, we were able to identify forms of human-technology relationalities in the Omaolo development that might both undermine and support the collective bases of public service delivery. This provides a concrete reminder that discussions about future technologies and their critiques can ignore and distort the complex and creative nature of development work, and the many parties and interests involved. We highlight attentiveness and specificity as core values for care providers, demonstrating how perceptions of good care transcend discrete decision-making instances to encompass the entire care continuum. Professionals in the field emphasise technology's role as an enabler of enhanced care, calling for its deliberate use. Simultaneously, there is an appreciation of algorithmic techniques that support self-care, provided they promote rather than undermine autonomy and equality in social and healthcare sectors.

Our findings call for more research that goes beyond readily available notions of algorithmic systems and identifies the local specificity of human-technology arrangements. In order to reconfigure human relationalities with algorithmic technologies within the public sector, we need to intervene in the current developments. One way to do this is to ensure distance, whenever possible, from the detrimental forces of dataism and automation whereby algorithmic aids start to transform professional aims into ‘meaningless work’ (Hoeyer and Wadmann, 2020). Another path we have demonstrated here is that of foregrounding other available options. This work includes seeking local alternatives to prevailing policy-driven solutions, and recognising, with the aid of professionals, practices that could be strengthened within existing systems. For those alternatives to grow and become effective, they need to be made visible, acknowledged and publicly valued.

Footnotes

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

This work was supported by the Research Council of Finland, Funding Decision no 332993 (Re-humanizing automated decision-making project) and the University of Helsinki, Finland, under CHANSE ERA-NET Co-fund programme, which has received funding from the European Union's Horizon 2020 Research and Innovation Programme, under Grant Agreement no 101004509 (Reimagine ADM project).

ORCID iD

Sonja Trifuljesko

Notes

References

Alastalo

Helén

(2022) A code for care and control: The PIN as an operator of interoperability in the Nordic welfare state. History of the Human Sciences 35(1): 242–265.

Andreassen

Kaun

Nikunen

(2021) Fostering the data welfare state: A Nordic perspective on datafication. Nordicom Review 42(2): 207–223.

Appadurai

Alexander

(2020) Failure. Cambridge: Polity Press.

Berg

(1997) Rationalizing Medical Work: Decision-Support Techniques and Medical Practices. Cambridge, MA: MIT Press.

Bergquist

Rolandsson

(2022) Exploring ADM in clinical decision-making: healthcare experts encountering digital automation. In: Pink

Berg

Lupton

Ruckenstein

(eds) Everyday Automation: Experiencing and Anticipating Emerging Technologies. London and New York: Routledge, pp.140–153.

Bowker

(1994) Science on the Run: Information Management and Industrial Geophysics at Schlumberger, 1920–1940. Cambridge, MA: MIT Press.

Calatayud

(2020) Country analysis: Spain. In: Chiusi

Fischer

Spielkamp

(eds) Automated Decision-Making Systems in the COVID-19 Pandemic: A European Perspective. Berlin and Gütersloh: AlgorithmWatch and Bertelsmann Stiftung, pp.29–13.

Choroszewicz

(2024) (In)visible everyday work of fostering a data-driven healthcare and social service organisation. New Technology, Work and Employment 39(1): 1–18.

Dahler-Larsen

(2012) The Evaluation Society. Stanford: Stanford Business Books.

10.

Dencik

(2022) The datafied welfare state: a perspective from the UK. In: Hepp

Jarke

Kramp

(eds) New Perspectives in Critical Data Studies: The Ambivalences of Data Power. Cham: Palgrave Macmillan, pp.145–165.

11.

Dencik

Kaun

(2020) Datafication and the welfare state. Global Perspectives 1(1): 12912.

12.

Espoon lastenneuvolan ja kouluterveydenhuollon ODA-tiimit (2017) Pienin kokeiluin kohti ODA-palveluita Espoon lastenneuvolassa ja kouluterveydenhuollossa. In: Omat digiajan hyvinvointipalvelut. Available at: https://oda-projekti.blogspot.com/2017/02/pienin-kokeiluin-kohti-oda-palveluita.html (accessed 18 August 2023).

13.

Hämeenlinnan ODA-tiimi (2016) Minunterveyteni.fi palvelukokonaisuus asiakaskäytössä – asiakastarina alkutaipaleelta. In: Omat digiajan hyvinvointipalvelut. Available at: https://oda-projekti.blogspot.com/2016/07/minunterveytenifi-palvelukokonaisuus.html (accessed 18 August 2023).

14.

Hietala

(2016) ODA osallistaa asiakkaan. In: Omat digiajan hyvinvointipalvelut. Available at: https://oda-projekti.blogspot.com/2016/06/oda-osallistaa-asiakkaan.html (accessed 18 August 2023).

15.

Hoeyer

(2019) Data as promise: Reconfiguring Danish public health through personalized medicine. Social Studies of Science 49(4): 531–555.

16.

Hoeyer

Bødker

(2020) Weak data: the social biography of a measurement instrument and how it failed to ensure accountability in home care. Medical Anthropology Quarterly 34(3): 420–437.

17.

Hoeyer

Wadmann

(2020) ‘Meaningless work’: How the datafication of health reconfigures knowledge about work and erodes professional judgement. Economy and Society 49(3): 433–454.

18.

Hoppula

Eskola-Tuoma

(2017) Sähköiset omahoitopalvelut helpottavat palvelujen sekamelskaa myös pitkien etäisyyksien Lapissa. In: Omat digiajan hyvinvointipalvelut. Available at: https://oda-projekti.blogspot.com/2017/01/sahkoiset-omahoitopalvelut-helpottavat.html (accessed 18 August 2023).

19.

Jackson

(2014) Rethinking repair. In: Gillespie

Boczkowski

Foot

(eds) Media Technologies: Essays on Communication, Materiality, and Society. Cambridge, MA: The MIT Press, pp.221–239.

20.

Jormanainen

Soininen

(2021) Use and users of the web-based Omaolo COVID-19 symptom self-assessment tool in Finland since March 16, 2020. Public Health and Informatics 281: 739–743.

21.

Karjalainen

(2016) Kaikessa tekemisessä alku on tärkein. In: Omat digiajan hyvinvointipalvelut. Available at: https://oda-projekti.blogspot.com/2016/08/kaikessa-tekemisessa-alku-on-tarkein.html (accessed 18 August 2023).

22.

Kaun

(2020) Country analysis: Sweden. In: Chiusi

Fischer

Spielkamp

(eds) Automated Decision-Making Systems in the COVID-19 Pandemic: A European Perspective. Berlin and Gütersloh: AlgorithmWatch and Bertelsmann Stiftung, pp.31–32.

23.

Kaun

Lomborg

Pentzold

, et al. (2023) Crosscurrents: welfare. Media, Culture & Society 45(4): 877–883.

24.

Kazansky

Milan

(2021) “Bodies not templates”: Contesting dominant algorithmic imaginaries. New Media & Society 23(2): 363–381.

25.

Keski-Suomen ODA-projektiryhmä (2016) Vuosikontrolli kuudessa viikossa tai kuudessa päivässä – kummanko valitset? In: Omat digiajan hyvinvointipalvelut. Available at: https://oda-projekti.blogspot.com/2016/06/vuosikontrolli-kuudessa-viikossa-tai.html (accessed 18 August 2023).

26.

Korkeila

(2016) Kohti Baabelin tornia. In: Omat digiajan hyvinvointipalvelut. Available at: https://oda-projekti.blogspot.com/2016/10/kohti-baabelin-tornia.html (accessed 18 August 2023).

27.

Kriikkula

Kivelä

Hirvasniemi

(2017) Paljon palveluja tarvitsevien prosessin kehittäminen Oulussa. In: Omat digiajan hyvinvointipalvelut. Available at: https://oda-projekti.blogspot.com/2017/03/paljon-palveluja-tarvitsevien-prosessin.html (accessed 18 August 2023).

28.

Kuopion projektitiimi (2016) Mitä uutta ODA tuo yläkouluikäisten oppilashuoltoon? In: Omat digiajan hyvinvointipalvelut. Available at: https://oda-projekti.blogspot.com/2016/09/mita-uutta-oda-tuo-ylakouluikaisten.html (accessed 18 August 2023).

29.

Kurki

(2016) Toimintaky lähestymistapana pitkäaikaistyöttömien hyvinvointiin. In: Omat digiajan hyvinvointipalvelut. Available at: https://oda-projekti.blogspot.com/2016/09/toimintaky-lahestymistapana.html (accessed 18 August 2023).

30.

Kurki

(2017) Yhteistoiminnassa asiakkaan parhaaksi. In: Omat digiajan hyvinvointipalvelut. Available at: https://oda-projekti.blogspot.com/2017/04/yhteistoiminnassa-asiakkaan-parhaaksi.html (accessed 18 August 2023).

31.

Langstrup

(2019) Patient-reported data and the politics of meaningful data work. Health Informatics Journal 25(23): 567–576.

32.

Lehtiniemi

(2024) Contextual social valences for artificial intelligence: Anticipation that matters in social work. Information, Communication & Society 27(6): 1110–1125.

33.

Lehtiniemi

Ruckenstein

(2019) The social imaginaries of data activism. Big Data & Society 6(1): 1–12.

34.

Lehtiniemi

Ruckenstein

(2022) Prisoners training AI. In: Pink

Berg

Lupton

Ruckenstein

(eds) Everyday Automation: Experiencing and Anticipating Emerging Technologies. London and New York: Routledge, pp.184–196.

35.

Lehto

Suurnäkki

(2016) Älykkäät sähköiset sote-palvelut tulevat - mitä sitten? In: Omat digiajan hyvinvointipalvelut. Available at: https://oda-projekti.blogspot.com/2016/11/alykkaat-sahkoiset-sote-palvelut.html (accessed 18 August 2023).

36.

Lupton

(2013) The digitally engaged patient: Self-monitoring and self-care in the digital health era. Social Theory & Health 11: 256–270.

37.

Mäkinen

(2016) ODA – mahdollisuus uudelle ajattelutavalle. In: Omat digiajan hyvinvointipalvelut. Available at: https://oda-projekti.blogspot.com/2016/06/oda-mahdollisuus-uudelle-ajattelutavalle.html (accessed 18 August 2023).

38.

Margetts

Dunleavy

(2013) The second wave of digital-era governance: A quasi-paradigm for government on the web. Philosophical Transactions of the Royal Society A 371(1987): 1–17.

39.

Mikkola

(2017) Minustako omaishoitaja? In: Omat digiajan hyvinvointipalvelut. Available at: https://oda-projekti.blogspot.com/2017/09/minustako-omaishoitaja.html (accessed 18 August 2023).

40.

Milan

Treré

(2019) Big data from the south(s): Beyond data universalism. Television and New Media 20(4): 319–335.

41.

Mol

(2008) The Logic of Care: Health and the Problem of Patient Choice. London: Routledge.

42.

Needham

(2011) Personalising Public Services: Understanding the Personalisation Narrative. Bristol: Policy Press.

43.

Nikunen

Hokka

(2020) Welfare state values and public service media in the era of datafication. Global Perspectives 1(1): 12906.

44.

Nordlund

(2016) ODA sähköistää sosiaali-ja terveyspalveluja – “Mikä ihmeen ODA”? In: Omat digiajan hyvinvointipalvelut. Available at: https://oda-projekti.blogspot.com/2016/05/oda-sahkoistaa-sosiaali-ja.html (accessed 18 August 2023).

45.

Ojala

(2016) Kustannus Oy Duodecim toimittaa tietämyskannan kansalliseen sähköiseen sote-palvelukokonaisuuteen. In: Omat digiajan hyvinvointipalvelut. Available at: https://oda-projekti.blogspot.com/2016/08/kustannus-oy-duodecim-toimittaa.html (accessed 18 August 2023).

46.

Päijät-Hämeen hyvinvointiyhtymän ODA-ryhmä (2017) ODA: sta polvi paranee. In: Omat digiajan hyvinvointipalvelut. Available at: https://oda-projekti.blogspot.com/2017/09/odasta-polvi-paranee.html (accessed 18 August 2023).

47.

Power

(1997) The Audit Society: Rituals of Verification. Oxford: Oxford University Press.

48.

Rahimi

(2019) Digital health and the elusive quest for cost savings. The Lancet Digital Health 1(3): e108–e109.

49.

Rannikko

Tamminen

Hellsten

, et al. (2022) Effectiveness of COVID-19 digital proximity tracing app in Finland. Clinical Microbiology and Infection 28(6): 903–904.

50.

Ratner

Elmholdt

(2023) Algorithmic constructions of risk: Anticipating uncertain futures in child protection services. Big Data & Society 10(2): 1–12.

51.

Reutter

(2022) Constraining context: Situating datafication in public administration. New Media & Society 24(4): 903–921.

52.

Rich

Miah

(2014) Understanding digital health as public pedagogy: A critical framework. Societies 4(2): 296–315.

53.

Ruckenstein

(2023) The Feel of Algorithms. Berkeley: University of California Press.

54.

Ruckenstein

Schüll

(2017) The datafication of health. Annual Review of Anthropology 46: 261–278.

55.

Ruckenstein

Trifuljesko

(2022) Algorithmic futures and the unsettled sense of care. In: Lanzeni

Waltorp

Pink

, et al. (eds) An Anthropology of Futures and Technologies. London: Routledge, pp.79–92.

56.

Ruppert

(2012) The governmental topologies of database devices. Theory, Culture & Society 29(4-5): 116–136.

57.

Ruppert

Isin

Bigo

(2017) Data politics. Big Data & Society 4(2): 1–7.

58.

Schüll

(2018) The sense mother. In: Society for Cultural Anthropology. Available at: https://culanth.org/fieldsights/the-sense-mother (accessed 7 August 2023).

59.

Schwennesen

(2019) Algorithmic assemblages of care: imaginaries, epistemologies and repair work. Sociology of Health & Illness 41(S1): 176–192.

60.

Seaver

(2019) Knowing algorithms. In: Vertesi

Ribes

(eds) digitalSTS: A Field Guide for Science & Technology Studies. Princeton, NJ: Princeton University Press, pp.412–422.

61.

Seaver

(2021) Care and scale: Decorrelative ethics in algorithmic recommendation. Cultural Anthropology 36(3): 509–537.

62.

Sharon

(2015) Healthy citizenship beyond autonomy and discipline: tactical engagements with genetic testing. BioSocieties 10: 295–316.

63.

Sharon

Zandbergen

(2017) From data fetishism to quantifying selves: Self-tracking practices and the other values of data. New Media & Society 19(11): 1695–1709.

64.

Sosiaali-ja terveysministeriö (2014) Tieto hyvinvoinnin ja uudistuvien palveluiden tukena. Sote-tieto hyötykäyttöön -strategia 2020. Available at: http://urn.fi/URN:ISBN:978-952-00-3548-8.

65.

Strathern

(2000) Introduction: new accountabilities. Anthropological studies in accountability, ethics and the academy. In: Strathern

(ed) Audit Cultures: Anthropological Studies in Accountability, Ethics and the Academy. London: Routledge, pp.1–18.

66.

Suchman

(2007) Human-Machine Reconfigurations: Plans and Situated Actions. Cambridge: Cambridge University Press.

67.

Suchman

(2012) Configuration. In: Lury

Wakeford

(eds) Inventive Methods. London: Routledge, pp.48–60.

68.

Tarkkala

Helén

Snell

(2019) From health to wealth: the future of personalized medicine in the making. Futures 109: 142–152.

69.

Timmermans

Tavory

(2012) Theory construction in qualitative research: From grounded theory to abductive analysis. Sociological Theory 30(3): 167–186.

70.

Torenholt

Langstrup

(2023) Between a logic of disruption and a logic of continuation: negotiating the legitimacy of algorithms used in automated clinical decision-making. Health: An Interdisciplinary Journal for the Social Study of Health, Illness and Medicine 27(1): 41–59.

71.

Valtioneuvoston kanslia (2015) Ratkaisujen Suomi: Pääministeri Juha Sipilän hallituksen strateginen ohjelma 29.5.2015. Hallituksen julkaisusarja 10/2015. Helsinki: Edita Prima.

72.

Wallace

Chan

Chidambaram

, et al. (2022) The diagnostic and triage accuracy of digital and online symptom checker tools: A systematic review. NPJ Digital Medicine 5: 118.

73.

Williamson

(2014) Knowing public services: Cross-sector intermediaries and algorithmic governance in public sector reform. Public Policy and Administration 29(4): 292–312.

74.

Winner

(1978) Autonomous Technology: Technics-Out-of-Control as a Theme in Political Thought. Cambridge, MA and London, England: The MIT Press.

75.

Yeung

(2022) The new public analytics as an emerging paradigm in public sector administration. Tilburg Law Review 27(2): 1–32.

Algorithmic configurations in caring arrangements

Abstract

Keywords

Introduction

A new paradigm: ‘Knowing’ public care services

Configurations of humans and algorithms

The making of Omaolo

Materials and methods

Advocating the service engine

Responding to the needs of the service engine

Promoting the treatment facilitator

Strengthening existing care practices

Delivering standardised service and situated treatment

Concluding remarks

Footnotes

Declaration of conflicting interests

Funding

ORCID iD

Notes

References