Developing an AI Enabled Salutogenic Leadership Model for Abadan Oil Refinery

Abstract

To develop a context-sensitive model of artificial intelligence (AI) enabled salutogenic leadership for Abadan Oil Refinery, integrating occupational well-being with the organizational requirements of digital transformation. This applied qualitative study integrated theoretical evidence with semi-structured interviews with 18 academic and organizational experts selected purposively for their knowledge of leadership, management, and AI applications. Audio-recorded interviews were transcribed verbatim and analyzed in ATLAS.ti using open, axial, and selective coding. Two independent coders reconciled disagreements through discussion; the reported Cohen’s kappa was 0.79. Preliminary categories were also reviewed by participating experts. The analysis generated 140 initial open codes and a six-category paradigm model. The model comprised causal conditions, intervening conditions, contextual conditions, strategies, a core phenomenon, and anticipated outcomes. Its core combined health-promoting and sustainable leadership, psychological coherence and occupational health, and readiness for human–AI collaboration. Causal conditions concerned leadership relationships, organizational and technological drivers, and attitudes toward technology. Contextual and intervening conditions encompassed health culture, job demands and resources, digital infrastructure, psychological capacities, and digital leadership. Strategies emphasized health support, stress management, leadership development, employee learning, and responsible technological adaptation. Anticipated outcomes concerned employee health and well-being, work attitudes and engagement, and sustainable organizational performance. The model frames AI as a resource for human decision-making within a health-promoting leadership approach. It provides an exploratory framework for refinery management rather than evidence that AI implementation improves health, safety, or productivity. The proposed relationships require prospective evaluation.

Keywords: salutogenic leadership; artificial intelligence; sense of coherence; occupational well-being; digital transformation; oil refining.

Introduction

Employee health is closely connected to the organization of work. Leadership shapes the resources available to employees, the clarity of expectations, and the ways in which occupational pressures are recognized and addressed. Evidence on health-oriented leadership links staff care with employee well-being, while also indicating that these relationships depend on working conditions and employees’ own health-related behavior (Kaluza et al., 2021; Hauff et al., 2022). A health-promoting leadership model therefore needs to address both interpersonal behavior and the wider organizational conditions within which that behavior occurs.

Salutogenesis offers a useful foundation for this task. Rather than considering health only through the prevention of disease, the salutogenic perspective examines the resources that help people maintain and develop health under demanding conditions (Antonovsky, 1987, 1996). Its central concept, sense of coherence, concerns the extent to which experience is comprehensible, manageable, and meaningful. Applied to work, this perspective directs attention to whether employees understand their tasks, can access adequate resources, and regard their contribution as worthwhile (Jenny et al., 2022). It does not imply that personal resilience can compensate for unsafe or persistently excessive demands.

The relevance of this perspective extends to industrial organizations. Halling et al. (2021) examined a leadership approach informed by sense of coherence and Lean principles in a manufacturing unit, providing an industrial example of the connection between work organization and health-oriented management. Research on managers facing technological change likewise suggests that comprehensibility, manageability, and meaningfulness can help explain how leaders interpret and respond to new working conditions (Mayer et al., 2021a, 2021b). These studies support examining salutogenic leadership in a refinery, while leaving open the question of how its components should be configured for that setting.

AI introduces additional demands and resources into this relationship. Data analysis and decision support may assist managers in recognizing operational patterns, coordinating work, and identifying emerging problems. However, the organizational consequences of AI are not inherently beneficial. Studies of AI adoption address anxiety, job stress, perceived opportunities, and organizational support, indicating that employee experiences depend on how technology is introduced and governed (Xu et al., 2023; Jeong et al., 2024; Soomro et al., 2026). Digitalized work may also expose employees to intensified monitoring and constrained discretion (Konuk et al., 2023).

At Abadan Oil Refinery, the focal issue is how leadership can connect occupational health with the human and organizational requirements of digital transformation. The refinery setting brings together operational complexity, safety-sensitive responsibilities, and pressure to maintain reliable production. Related oil and gas research shows that complex energy-sector decisions may require explicit balancing of competing technical criteria; for example, Mofid et al. (2024) used multi-objective optimization to support offshore LNG process design selection. These characteristics make the relationship between technology and employee well-being particularly relevant. Nevertheless, the present study did not measure the prevalence of occupational stress or document the performance of an existing AI system at the refinery. Its purpose was to develop a framework based on expert perspectives and relevant theory.

The research question was: How can an AI-enabled salutogenic leadership model be developed for Abadan Oil Refinery that integrates occupational well-being, organizational conditions, and responsible human–AI collaboration? The contribution lies in organizing these elements into a context-sensitive qualitative model. AI-enabled leadership is used here to describe the proposed integration of AI capabilities into leadership practice; it does not indicate that an AI algorithm was trained, deployed, or validated during the study.

Conceptual background

Salutogenesis and health oriented leadership

Sense of coherence provides three related lenses for interpreting work. Comprehensibility concerns the clarity and predictability of information, roles, and organizational change. Manageability concerns whether employees have the authority, skills, social support, and material resources needed to meet demands. Meaningfulness concerns whether work and participation in change are experienced as worthwhile (Antonovsky, 1987, 1996; Jenny et al., 2022). These dimensions connect individual experience to organizational arrangements rather than treating health solely as an individual responsibility.

Health-oriented leadership provides a complementary behavioral perspective. Staff care concerns leaders’ attention to employees’ health and their actions to protect it. Self-care concerns leaders’ management of their own health-related needs. Kaluza et al. (2021) considered expectations, self-care, and leader–member relationships, whereas Klug et al. (2022) examined pathways linking leader self-care with staff care and employee health. Hauff et al. (2022) situated health-oriented leadership within the relationship between high-performance work practices and employee well-being. These contributions support treating leader capability and organizational resources as connected components.

The job demands–resources model further clarifies why leadership should address work design. Job demands require sustained effort, whereas resources support task achievement and can help employees cope with demands (Demerouti et al., 2001). In a salutogenic interpretation, clarity, participation, learning opportunities, and access to support can strengthen the conditions through which work becomes manageable and meaningful. Lutz et al. (2023) also examined health-oriented leadership in relation to job satisfaction and self-rated health, although associations from survey research should not be interpreted as proof of intervention effectiveness.

AI adoption and the experience of work

A salutogenic approach to AI requires attention to how employees interpret technological change. An AI system can make information more accessible while simultaneously making managerial decisions harder to understand. It can reduce repetitive work while increasing uncertainty about job security or accountability. Xu et al. (2023) examined perceived AI opportunities in relation to workplace well-being, while Jeong et al. (2024) considered job stress and coaching leadership during AI adoption. Together, these studies provide a rationale for examining both opportunity and strain.

Human–AI collaboration also depends on the fit between technological tools and work tasks. Aulia and Lin (2025) examined AI-related collaboration and employee outcomes in remote work, offering evidence from a setting distinct from oil refining. Soomro et al. (2026) considered managerial capability and organizational support in AI adoption. These findings inform the conceptual framework but cannot be transferred directly to refinery operations. In particular, evidence from remote workplaces, call centers, or small and medium-sized enterprises does not establish that comparable outcomes will occur in a safety-sensitive industrial setting.

The resulting conceptual position is that AI should be examined as part of a sociotechnical work environment. Leadership decisions determine how outputs are interpreted, how employees can question decisions, and whether data are used for support or punitive surveillance. The present model therefore brings together health resources, digital capability, employee participation, and responsible governance, while retaining human judgment as a central feature.

Methods

Design and setting

The study used an applied qualitative design integrating expert interviews with theoretical evidence. The original methodological label, multicontextual, referred to the interaction between two sources of understanding: the empirical context of participants’ accounts and the theoretical context of salutogenesis, health-oriented leadership, and AI-enabled management. It did not denote a multisite design. Abadan Oil Refinery was the organizational focus, and the analysis followed the open, axial, and selective coding sequence described in the study record.

The approach was theory-informed rather than wholly independent of prior concepts. The literature provided sensitizing concepts for the interview protocol and the six-part paradigm structure. Interview material was then coded and compared with this framework. Acknowledging this relationship is necessary because using a predefined interview structure and subsequently comparing codes with theory limits any claim that the model emerged without theoretical influence. The study was exploratory and did not test prespecified hypotheses.

Participants and recruitment

Participants were selected through purposive, judgment-based sampling. Eligibility concerned relevant experience or knowledge in management and leadership, familiarity with salutogenic leadership, and knowledge or experience of AI applications in management. The sample included academic and organizational experts with experience relevant to public organizations and Abadan Oil Refinery. Eighteen in-depth, semi-structured interviews were completed. Recruitment ended when the researchers judged that further interviews were no longer yielding new concepts and the category structure had stabilized.

The sample size of 18 is retained as reported. The original demographic percentages cannot be reconciled with that denominator, so exact counts for sex, educational qualifications, and experience categories are not reported here. This omission prevents the presentation of internally inconsistent participant characteristics. The sampling account also does not establish how many invited individuals declined participation or how the academic and organizational participants were distributed across roles.

Data collection

The principal data collection method was semi-structured interviewing. The protocol introduced the study, obtained general participant information, explained its purpose, and emphasized confidentiality. Questions addressed causal conditions, the core phenomenon, contextual conditions, intervening conditions, strategies, and outcomes relevant to AI-enabled salutogenic leadership. Interviews were audio-recorded with informed consent and transcribed verbatim. No interview dates, durations, interview locations, or interviewer characteristics were available in the study record.

The literature component organized concepts from prior research into a preliminary framework. Because the record did not specify reproducible database searches, eligibility rules, or a screening process, this component is described as a targeted theoretical review rather than a systematic review. Table 1 presents a concise evidence map of relevant, traceable publications used to interpret the model. It is not an exhaustive inventory or a quantitative synthesis of the literature.

Analysis and integration

Interview data were managed and analyzed in ATLAS.ti. During open coding, transcripts were examined line by line to identify concepts and meanings. The study reported 140 initial open codes after conceptual refinement and removal of redundant expressions. During axial coding, related concepts were grouped according to semantic similarity and their relationships within the paradigm structure. During selective coding, these categories were integrated into broader themes and related to the core phenomenon.

Theoretical and empirical categories were compared iteratively. The literature-derived framework acted as a flexible guide, while the interview findings informed its refinement and contextual interpretation. Table 2 summarizes the six analytical categories and their organizing themes. Table 3 translates the detailed interview code structure into English, preserving the 29 axial categories presented in the source. Table 4 reports the final model components and the interview identifiers linked to them in the study record. Figure 1 visualizes the proposed relationships.

Two independent coders undertook coding, and disagreements were resolved through discussion. The reported Cohen’s kappa was 0.79. This value is retained as a reported agreement statistic; the coding unit, number of double-coded units, and calculation details were not provided, so it cannot be independently reproduced from the manuscript. Preliminary findings and categories were also shared with some participating experts, whose feedback informed refinement. The number of participants involved in this review and the resulting revisions were not specified.

Reporting and ethical procedures

The study record states that informed consent, confidentiality, and faithful recording and interpretation of participants’ views were maintained. It does not identify an approving ethics committee, an approval number or date, or a documented exemption. Accordingly, formal ethics approval is not asserted. This distinction preserves the reported consent procedures without substituting a general ethical statement for institutional review documentation.

COREQ and the Standards for Reporting Qualitative Research identify information needed to appraise interview-based studies, including researcher characteristics, context, analytical procedures, and evidence supporting interpretations (Tong et al., 2007; O’Brien et al., 2014). They are used here to clarify reporting boundaries, not to claim that the original study met every checklist item. Verbatim participant quotations and a complete coding audit trail were not supplied; none have been reconstructed or invented.

Results

Overview of the model

The analysis of 18 interviews yielded 140 initial open codes, organized into axial categories and integrated into six paradigm categories. The core phenomenon was AI-enabled salutogenic leadership. Its three organizing themes were health-promoting and sustainable leadership, psychological coherence and occupational health, and readiness and maturity for human–AI interaction. The other categories described the perceived antecedents, enabling or constraining conditions, strategies, and anticipated outcomes of this leadership approach.

The categories represent the researchers’ synthesis of expert accounts and theoretical concepts. Labels such as causal conditions and outcomes retain the terminology of the original qualitative paradigm; they do not indicate statistically demonstrated causal effects. The interview identifiers in Table 4 preserve the links reported in the source, but the absence of quotations limits readers’ ability to examine how individual statements supported the final interpretation.

Causal conditions

The first category concerned leadership and interpersonal relationships, organizational and technological drivers, and individual attitudes toward technology. At the leadership level, the codes emphasized self-awareness, self-care, emotional regulation, resilience, and attention to employee needs. At the organizational level, the model incorporated technological capability, the availability of information, and readiness for digital transformation. At the individual level, it included AI literacy, willingness to learn, and attitudes toward technological change.

These elements position leaders as interpreters of technological change and providers of health-related resources. The coding framework connected knowledge of AI with the capacity to make its use understandable and relevant to work. It also linked self-care and resilience with the ability to respond consistently under pressure. The findings describe perceived prerequisites for the proposed model rather than observed predictors of adoption or health outcomes.

Intervening and contextual conditions

Intervening conditions included psychological resources, digital leadership competencies, team dynamics, and job-related capacities. The detailed codes addressed resistance to AI, change leadership, trust, transparency, fairness, and privacy. These codes explain how the broad themes could operate in practice: technological change may be easier to manage when employees understand its purpose, have opportunities to learn, and retain confidence that their information will be treated responsibly.

Contextual conditions concerned the organizational health climate and culture, job design and operational pressure, and the digital transformation ecosystem. The interview code structure emphasized psychological safety, access to human and technical resources, trust, cooperation, and continuous learning. Together, these categories locate leadership within an organizational environment. Supportive interpersonal conduct alone was not presented as sufficient when work demands or infrastructure prevent employees from using available resources.

Strategies and the core phenomenon

Strategies combined health-oriented interventions, stress management, leadership development, employee learning, and cognitive and technological adaptation. The codes included managerial support, participation in decisions, empowerment, open communication, data-informed decision-making, early recognition of strain, and human–AI collaboration. Proposed monitoring activities were framed as preventive resources, while the governance codes emphasized confidentiality and fair treatment. These were components of the model, not systems documented as having been implemented.

The core phenomenon connected these strategies with the salutogenic aim of strengthening health resources and sense of coherence. AI-enabled leadership was therefore conceptualized through the quality of human judgment and organizational support surrounding technology. Its preventive dimension concerned recognizing pressures early and responding with appropriate resources. Its human-centered dimension concerned preserving discretion and making technology responsive to employee needs.

Note. The entries translate the code labels presented in the original manuscript. They are summarized code content, not verbatim interview quotations. The source reported 140 initial open codes; a separate 140-item codebook was not supplied.

Anticipated outcomes and integration

The outcome category encompassed multidimensional health and well-being, work attitudes and engagement, and organizational performance and sustainability. More detailed codes included resilience, belonging, job satisfaction, reduced stress and burnout, productivity, and reduced absence and error. These outcomes describe what participants and the model anticipated. No pre–post health measures, productivity comparisons, safety records, or observed AI performance results were reported.

Integration retained distinct roles for the six categories while linking them around the core phenomenon. The source’s final table grouped psychological and team resources under intervening conditions, whereas its detailed interview coding placed change management and ethical governance in that category. Both levels are retained: Table 2 presents the final organizing themes, and Table 3 presents the more specific interview categories. Their relationship is interpretive rather than a demonstrated measurement hierarchy.

Note. Identifiers reproduce the category-to-interview links reported in the original manuscript. They do not represent independent participants per category, frequency estimates, or a verified quotation-level audit trail.

Discussion

Leadership as a source of coherence during digital change

The model places the human experience of work at the center of AI-enabled leadership. Its contribution is the integration of salutogenic concepts with leadership behavior, work design, and technological governance in a refinery-focused framework. Comprehensibility is reflected in transparent communication and clarity about AI-related decisions. Manageability is reflected in skills, autonomy, support, and adequate resources. Meaningfulness is reflected in participation, belonging, and the perceived value of work. These interpretations are consistent with the salutogenic account of health resources (Antonovsky, 1996; Jenny et al., 2022).

The emphasis on leader self-care and staff care also aligns with evidence that the health-related behavior of leaders is relevant to employees’ experience of support (Kaluza et al., 2021; Klug et al., 2022). However, the present findings do not establish which leadership behavior has the greatest influence or whether training would change employee health. They identify categories that experts regarded as relevant. The distinction matters because a conceptually coherent model can still require substantial refinement before it becomes an effective intervention.

Work design and organizational resources

The contextual category broadens the model beyond leader behavior. Workload, role clarity, autonomy, technical resources, and organizational culture affect whether employees can act on support offered by managers. This interpretation is compatible with the job demands–resources model and with research situating health-oriented leadership within organizational work practices (Demerouti et al., 2001; Hauff et al., 2022). It also avoids interpreting stress solely as a deficit in employees’ coping capacity. Sector-specific management research also shows that oil-industry organizations face constrained resource-allocation decisions across competing projects, for which formal portfolio optimization has been applied (Razavi & Motavali, 2023).

The refinery focus adds an operational setting to this integration, but its transferability remains conditional. Halling et al. (2021) provide a manufacturing example of linking sense of coherence with leadership and operational indicators. The current study extends the conceptual discussion to AI-related leadership requirements; it does not replicate an industrial intervention. The model should therefore be read as a structured account of perceived relationships at the focal organization rather than a general account of all refineries.

AI as a resource and a source of strain

The simultaneous presence of AI literacy, technological opportunity, anxiety, and privacy in the coding structure suggests that digital transformation was understood as an organizational process with competing implications. This interpretation is consistent with studies addressing perceived AI opportunities and workplace well-being, as well as the role of supportive leadership during AI-related stress (Xu et al., 2023; Jeong et al., 2024). It also accords with evidence that digitalized work can raise concerns about emotional demands and technological control (Konuk et al., 2023).

The model’s human-centered orientation has a practical implication: the value of a tool depends on the quality of the decisions and work arrangements surrounding its use. For example, identifying a workload concern would provide limited benefit without a managerial response that reallocates resources or adjusts demands. Conversely, using employee information without clear purposes or safeguards could undermine trust. These examples explain the logic of the framework; they are not observations of systems deployed at Abadan Oil Refinery.

The newer AI literature reinforces the importance of organizational support, managerial capability, and task-related collaboration (Aulia & Lin, 2025; Soomro et al., 2026). Differences in sectors and research designs nevertheless limit direct comparison. The present study contributes an integrated vocabulary for examining these issues in refining, while leaving questions of feasibility, algorithm performance, employee acceptance, and health effects unresolved.

Practical interpretation

Within the proposed framework, leadership development would address self-awareness, supportive communication, health awareness, and digital literacy together. Employee participation would help clarify the purpose of technology, identify work-related concerns, and interpret whether a proposed tool fits actual tasks. Governance would specify legitimate data uses, access responsibilities, confidentiality protections, and human accountability. These are implications of the model, not completed organizational changes.

The strategic category also connects learning with the management of uncertainty. Explaining the limitations of AI outputs and preserving routes for questioning decisions could strengthen the comprehensibility of change. Providing time and resources for learning could strengthen manageability. Involving employees in decisions affecting their work could support meaningfulness. The framework thus makes the salutogenic dimensions concrete without assuming that technology adoption automatically produces health benefits.

Limitations

Several limitations constrain interpretation. The sample comprised 18 purposively selected experts and does not represent the refinery workforce statistically. Exact participant demographics could not be reconstructed from the inconsistent percentages in the source. The range of employee perspectives, including those of frontline and shift workers, cannot be assessed because the role distribution was not specified. Claims of saturation are retained as the researchers’ judgment, but a documented saturation assessment was not available.

The analysis combined interviews with a theory-informed framework. This strengthened conceptual integration but may have encouraged the placement of data within predefined categories. The absence of verbatim quotations, a full codebook, and documented examples of category revision reduces the transparency of the analytic chain. The reported agreement statistic is not independently reproducible from the available information. These constraints should be considered alongside the use of independent coding and participant feedback (Tong et al., 2007; O’Brien et al., 2014).

The theoretical review lacked a reproducible search and screening account and should not be treated as systematic or exhaustive. The study also did not evaluate an AI application, measure changes in employee health or operational performance, or estimate the proposed pathways. All outcome claims are therefore prospective. Finally, institutional ethics review details and author declarations were not available in the supplied study record, preventing a claim that every submission requirement has been completed.

Conclusion

This study developed an exploratory model of AI-enabled salutogenic leadership for Abadan Oil Refinery by integrating expert interviews with theoretical evidence. The model connects leadership relationships, psychological and organizational resources, work design, digital readiness, and responsible AI governance around the goal of supporting employee coherence and occupational well-being. Its anticipated benefits concern employee health, engagement, and sustainable performance. These benefits remain propositions rather than demonstrated effects. The framework provides a basis for subsequent operationalization and evaluation while preserving the central roles of human judgment, employee participation, and organizational responsibility.

Declarations

Ethics and informed consent

This study involved semi-structured interviews with academic and organizational experts. Informed consent was obtained from all participants before study participation and audio recording. Participants were informed of the purpose of the study, the voluntary nature of participation, and the confidentiality of their responses. Interview materials were handled confidentially and reported in a manner intended to protect participant identity. No ethics committee name or formal approval identifier was available in the study documentation; accordingly, no ethics approval number is reported in this manuscript.

Data availability

The qualitative data supporting the findings of this study consist of interview transcripts and coding materials. Because these materials may contain information that could permit participant identification or reveal confidential organizational information, they are not publicly available. Requests for access may be directed to the corresponding author and will be considered on a case-by-case basis, subject to participant confidentiality, consent conditions, and applicable institutional requirements.

Funding

The authors declare that this research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors.

Conflicts of interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Author contributions

All authors contributed to the conception and design of the study, interpretation of the findings, development of the proposed model, and critical revision of the manuscript. All authors read and approved the final version of the manuscript and agree to be accountable for the integrity and accuracy of the work.