I've spent the past week reviewing the literature on vocational rehabilitation models - ICF frameworks, biopsychosocial approaches, Swedish return-to-work research. Yesterday I documented that synthesis in a concept modelling post. But as I worked through those models, something kept surfacing that I want to think through more carefully: the centrality of social networks to health outcomes, and how poorly equipped our services seem to be to address them.
This resonates with work I did before this PhD - at Region Värmland, where I was involved in service design projects addressing social isolation and loneliness. I want to use this post to reflect on those connections, because I'm curious whether they'll be relevant to what I'm now being asked to explore.
The I2I Project: Isolation to Inclusion
Before joining this research position, I worked at Region Värmland on a project called I2I - Isolation to Inclusion. The project tackled social isolation and loneliness among older people, exploring how services might be designed to strengthen social connections rather than simply delivering care transactions.
The literature we engaged with was unambiguous about the stakes. Social isolation and loneliness are risk factors for poor mental and physical health, particularly problematic in old age due to "reduced social network size, loss of loved ones, and increased prevalence of chronic conditions" (Courtin & Knapp, 2015, p. 1). The relationship between social support and health is well-documented: people with stronger social networks tend to be stronger, healthier, and happier (Mittelmark et al., 2016).
But what struck me during that work wasn't the evidence base - it was the difficulty of designing services that actually addressed relational needs. Healthcare services are structured to provide care according to specialty and function (Tsekleves & Cooper, 2017). They're good at delivering interventions to individuals. They're less good at nurturing the social fabric that makes those individuals resilient.
The I2I project explored social prescribing approaches - connecting people to community activities, groups, and networks rather than (or alongside) clinical interventions. As Polley and Fleming put it, "social prescribing is about aligning the services that are available to a person in different sectors and identifying the need for new services" (Polley & Fleming, 2017, p. 30). The approach recognises that health isn't just about bodies - it's about relationships, belonging, and participation.
Rehabilitation and the ICF Framework
Now, reading the vocational rehabilitation literature, I'm seeing similar themes from a different angle.
The International Classification of Functioning, Disability and Health (ICF) has become the dominant framework for understanding rehabilitation. It represents a shift from purely biomedical models toward what's called the biopsychosocial approach - recognising that functioning and disability emerge from the interplay of physical conditions, psychological factors, and social environments.
Escorpizo et al. describe how the ICF provides a conceptual foundation for vocational rehabilitation, emphasising that disability "is composed of the interplay of physical and mental conditions, perceptions, environments and social experiences" (Escorpizo et al., 2011). The framework includes "participation" as a core domain - acknowledging that health isn't just about body functions but about engagement in life situations.
What interests me is how the ICF explicitly includes environmental factors, including social relationships and support systems. The framework recognises that a person's capacity to function depends not just on their individual characteristics but on the resources available to them - including informal networks of family, friends, and community.
This connects directly to the salutogenic perspective developed by Aaron Antonovsky. Rather than asking "what makes people sick?", salutogenesis asks "what keeps people healthy?" The answer, substantially, involves what Antonovsky called "generalised resistance resources" - including social support, cultural stability, and meaningful engagement (Mittelmark et al., 2022). A person's "sense of coherence" - their ability to comprehend, manage, and find meaning in life's challenges - is strengthened or weakened by their social environment.
Gender, Networks, and Recovery
One thread that keeps appearing in the Swedish literature is the gendered nature of sickness absence and rehabilitation.
The Experio Worklab project, which I was initially involved with at Region Värmland, was predicated on feminist research indicating significant gender differences in long-term sick leave. Women in the public sector are more likely than men to have long-term sick leave, and women have mental health diagnoses to a greater extent (Region Värmland, 2018). The project explored how these patterns connect to workplace structures, care responsibilities, and the kinds of support available during recovery.
Research on return-to-work outcomes shows that social factors - including support from supervisors, colleagues, family, and healthcare providers - significantly influence recovery trajectories (Schultz & Gatchel, 2016). But these social factors operate differently for women and men, shaped by different expectations, responsibilities, and network structures.
Alexanderson and Marklund's research on women's and men's sick leave in Sweden found that "few of the long-term sick-listed in this study returned to work within a two-year period" (Alexanderson & Marklund, 2010). The factors predicting return were complex, involving not just diagnosis but social circumstances, workplace conditions, and the quality of rehabilitation support.
What this suggests is that rehabilitation isn't just about treating conditions - it's about rebuilding capacity within social contexts. And those contexts are gendered, classed, and shaped by the kinds of networks people have access to.
Kinship and Care Networks
The concept of "kin networks" comes from anthropology but has relevance for how we think about rehabilitation.
Research on informal care emphasises the importance of interhousehold ties - the connections between family members, neighbours, and community members that provide practical and emotional support outside formal services (Hu & Qian, 2021). These networks vary enormously in their density, reach, and reliability. Some people are embedded in thick webs of mutual support; others are isolated, with few people they can call on.
Social capital theory helps explain why these networks matter. As Portes (2002) describes, social capital influences transfers of resources - not just money, but information, practical help, emotional support. People with stronger networks have more resources to draw on during crises, including health crises.
For rehabilitation, this has concrete implications. A person recovering from illness or injury doesn't just need clinical treatment - they need help with daily tasks, emotional support during setbacks, information about options, and encouragement to persist. If their network provides these things, recovery is more likely. If not, even good clinical care may be insufficient.
The challenge is that formal services are poorly designed to assess or strengthen these informal networks. We assess individuals, diagnose conditions, prescribe interventions. We don't typically map someone's social resources or design interventions to strengthen their relational infrastructure.
Service Design and Social Material
This brings me back to service design - the practice I've worked within or found myself at the periphery of research around.
There's been increasing recognition that service design needs to engage with relational and social dimensions, not just transactional ones. Sangiorgi and Prendiville argue that "relational and collaborative services need to be nurtured over time by a clear social value proposition" (Sangiorgi & Prendiville, 2017, p. 18). The journey to good services involves understanding the "dynamic relationships between people, physical elements, systems and resources" (Parker & Heapy, 2006, p. 57).
Van der Bijl-Brouwer's work on "designing for social infrastructures in complex service systems" points to the challenge of addressing complex societal issues through design (van der Bijl-Brouwer, 2018). Service design has tools for mapping journeys, identifying touchpoints, prototyping interactions. It's less clear what tools we have for strengthening the social fabric that makes services work.
The I2I project taught me that designing for social connection requires different approaches than designing transactions. You can't just create a "loneliness service" and expect it to work. You have to understand existing networks, identify where connections might form, create conditions for relationships to develop organically. It's slower, messier, less controllable than traditional service design.
Questions About Automation and Relationship
This brings me to something I'm trying to think through as I read about data science and AI applications in public services.
The research position I've taken up is explicitly about exploring advanced computational approaches - federated learning, machine learning, data science - in vocational rehabilitation contexts. There are obvious potential benefits: pattern recognition across large datasets might identify which interventions work for whom; automated systems might reduce administrative burden; predictive models might enable earlier, more targeted support.
But if social networks and human relationships are as central to rehabilitation as the evidence suggests, I'm wondering how computational approaches fit in. The "digital welfare state" is rapidly becoming prominent, with datafication and algorithmic processing reshaping how public services operate (Dencik & Kaun, 2020). This transformation tends to emphasise efficiency, standardisation, and scalability. Are these the right qualities for relational, contextual care? I genuinely don't know.
Julia Unwin's work on kindness in public services caught my attention. As she puts it, "the evidence is clear that personal relationships improve care and that human contact and engagement is important both at points of crisis and in the longer term" (Unwin, 2018, p. 9). This isn't sentimentality - it's recognition that recovery happens through relationships, not despite them.
The Swedish rehabilitation literature echoes this. Getting good information and being treated well - bemötande - is important for people to be able to draw on their resources and take responsibility for their recovery (Ekberg & Eklund, 2015). This requires human judgement, contextual sensitivity, and the kind of rapport that develops through repeated interaction. Could algorithmic systems support this, or would they work against it?
Wastell's critique of "technomagic" - the tendency of managers to see technology as a magic bullet that will solve complex social problems (Wastell, 2011) - seems worth holding in mind. The appeal is understandable: technology promises efficiency, consistency, objectivity. But I'm wondering whether rehabilitation is a domain where those qualities help or hinder. Each person's situation is different; what works depends on context; recovery is non-linear and unpredictable.
I'm also curious about what automation does to the remaining human interactions. If routine tasks are automated, do the interactions that remain become more transactional, more focused on edge cases and exceptions? Does relationship-building that happens through ordinary contact get squeezed out?
I don't want to prejudge this. Pattern recognition across outcomes data could genuinely help identify what works. Administrative automation could free up time for the relational work that matters. Maybe the question isn't whether to use technology, but how to design it so it supports human relationships rather than replaces them.
These are questions I'm carrying into the fieldwork. I'm not sure yet what "data science for rehabilitation" means in practice, or whether the tension I'm sensing between computational approaches and relational care is real or resolvable.
Questions for This New Work
As I begin exploring vocational rehabilitation at SCÖ, these reflections raise questions:
How do rehabilitation services currently account for social networks? The ICF framework includes environmental factors, but do the actual services assess and address them? Or do they focus on individual capacities while assuming social context as background?
What role could federated data approaches play in understanding network effects? The research premise I've been hired to explore involves data science and federated learning. Could these approaches help identify patterns in how social factors influence rehabilitation outcomes? Or is the data too individualised to capture relational dynamics?
What would it mean to design rehabilitation services that strengthen networks rather than just treating individuals? This seems important but also difficult - it would require different assessment approaches, different intervention logics, different success metrics.
How do gender and other social differences shape network resources in rehabilitation contexts? The Experio Worklab premises - and the work that I was involved in there suggest this matters significantly. Are current services designed with these differences in mind?
I don't have answers yet. I'm still orienting, still reading, still trying to understand what's possible in this context. But I wanted to capture these early reflections, because I suspect the social and relational dimensions of rehabilitation will keep surfacing as I go deeper into this work.
References
Alexanderson, K. and Marklund, S. (2010). Studier om kvinnors och mäns sjukfrånvaro. Försäkringskassan.
Courtin, E. and Knapp, M. (2015). Social isolation, loneliness and health in old age: a scoping review. Health and Social Care in the Community, 25(3), 799-812.
Dencik, L. and Kaun, A. (2020). Datafication and the Welfare State. Global Perspectives, 1(1), 12912.
Ekberg, K. and Eklund, M. (2015). Återgång i arbete. Studentlitteratur.
Escorpizo, R., Reneman, M.F., Ekholm, J., Fritz, J., Krupa, T., Marnetoft, S.U., Maroun, C.E., Rodriguez Guzman, J., Suzuki, Y., Stucki, G. and Chan, C.C.H. (2011). A Conceptual Definition of Vocational Rehabilitation Based on the ICF: Building a Shared Global Model. Journal of Occupational Rehabilitation, 21, 126-133.
Hu, Y. and Qian, Y. (2021). COVID-19, Inter-household Contact and Mental Well-Being Among Older Adults in the US and the UK. Frontiers in Sociology, 6, 714626.
Mittelmark, M.B., Bauer, G.F., Vaandrager, L., Pelikan, J.M., Sagy, S., Eriksson, M., Lindström, B. and Meier Magistretti, C. (eds.) (2022). The Handbook of Salutogenesis (2nd ed.). Springer.
Mittelmark, M.B., Sagy, S., Eriksson, M., Bauer, G.F., Pelikan, J.M., Lindström, B. and Espnes, G.A. (eds.) (2016). The Handbook of Salutogenesis. Springer.
Parker, S. and Heapy, J. (2006). The Journey to the Interface. Demos.
Polley, M. and Fleming, J. (2017). Making Sense of Social Prescribing. University of Westminster.
Portes, A. (2002). Social Capital: Its Origins and Applications in Modern Sociology. Annual Review of Sociology, 24, 1-24.
Region Värmland (2018). Experio Worklab Project Proposal.
Sangiorgi, D. and Prendiville, A. (2017). Designing for Service: Key Issues and New Directions. Bloomsbury.
Schultz, I.Z. and Gatchel, R.J. (eds.) (2016). Handbook of Return to Work: From Research to Practice. Springer.
Tsekleves, E. and Cooper, R. (eds.) (2017). Design for Health. Routledge.
Unwin, J. (2018). Kindness, Emotions and Human Relationships: The Blind Spot in Public Policy. Carnegie UK Trust.
van der Bijl-Brouwer, M. (2018). Designing for Social Infrastructures in Complex Service Systems: A Human-Centered and Social Systems Perspective on Service Design. She Ji: The Journal of Design, Economics, and Innovation, 3(3), 183-197.
Wastell, D. (2011). Managers as Designers in the Public Services: Beyond Technomagic. Triarchy Press.