Health and Psychosocial Resources

The aim of the research

The research aims to develop an AI-based decision-support system, operating in a closed hospital environment, that analyses clinical data together with psychological, social and communicative factors.

The aim of the research is to develop an AI-based predictive system that analyses patients' clinical data together with psychological, social and communicative factors. The system's task is the early recognition of unfavourable health outcomes, difficulties in psychological adaptation, and increased support needs, as well as laying the groundwork for targeted prevention strategies. According to the biopsychosocial approach, understanding illness and health requires treating biological, psychological and social dimensions within a unified analytical framework [5].

The theoretical starting point of the research is that coping with illness is not determined solely by biological factors. It also matters greatly how well a patient understands their own situation, whether they have mobilisable resources, whether they are able to ask for help, what social support they perceive, and how manageable, comprehensible and meaningful they consider their life situation to be. This approach connects to the model of salutogenesis, which places emphasis on the individual and social resources that support health, as well as on the sense of coherence [1, 6, 7].

Theoretical and publication links

The research concept connects directly to Krisztián Dombrádi's communication- and social-theoretical work. His works treating social capital as a resource created and mobilised through communicative processes lay the groundwork for incorporating social support, trust and communal embeddedness into the hospital prognostic model [2, 3]. The transdisciplinary and systems-theoretical interpretation of communication provides the theoretical framework for linking the clinical, psychological, social and AI modules [4].

The empirical health-sociological basis of the plan is reinforced by the 2026 study by Krisztián Dombrádi and András Székely, which examines the correlations between stress, social support and coherence within a sociological-theoretical framework using Hungarostudy 2021 data [1].

Main research question

To what extent does the accuracy of hospital prognoses improve if, alongside traditional clinical and demographic data, indicators of psychological coping, social support, communicative security and sense of coherence are also incorporated into the predictive models?

Data examined

The research can be built on properly pseudonymised clinical data available within the hospital's IT system:

• diagnoses, comorbidities and previous hospital events;
• laboratory results, medication and treatment data;
• length of stay, complications and readmissions;
• age, sex and justified socio-demographic variables;
• psychological coping strategies;
• sense of coherence;
• perceived social support and social embeddedness;
• loneliness, stress, anxiety and depressive symptoms;
• the patient's perceived level of information and communicative security;
• participation in and cooperation with treatment.

Psychosocial factors can be measured with short, validated questionnaires, for example standardised instruments for sense of coherence, coping, and perceived social support.

Modules of the AI system

The researcher programs the various analytic units into a unified system operating within a closed hospital environment.

• Data integration module: connects and organises clinical, psychological and social data into a unified format.
• Communication analysis module: with appropriate authorisation and data protection, can analyse communicative patterns appearing in pseudonymised clinical notes, such as signals indicating uncertainty, support needs or difficulty in cooperation.
• Prediction module: using machine learning methods, estimates the probability of, among other things, psychological decompensation, extended hospital stay, complications, readmission, or an increased need for follow-up care.
• Explanatory module: shows which clinical and psychosocial factors played the most important role in a given prediction. The system must provide not merely a risk score but a professionally interpretable justification.
• Prevention module: links risk patterns to predefined support options, such as psychological consultation, involvement of a social worker, activation of family support, patient education, or more intensive follow-up.

Research design

The first phase of development can take place on retrospective, pseudonymised data. This would be followed by a temporally separated validation and then a prospective, observational study. The model's performance must be evaluated not only for accuracy but also for calibration, explainability, and fairness across different patient groups.

During the research phase, the system may not make an automatic clinical decision. It can provide only a decision-support signal, which is evaluated by a physician, psychologist, or other designated professional. The goal is not to label patients but to recognise who may need additional support, and in what form.

Data protection and ethical conditions

Hospital employment alone does not authorise the researcher to use patient data for research purposes. The research requires prior institutional and research-ethics approval, a documented legal basis for data processing, role-based access, data minimisation and pseudonymisation. The GDPR requires appropriate technical and organisational safeguards for scientific research, in particular data minimisation and, where possible, pseudonymisation [8].

Given the large volume of health data and the profiling involved, a data protection impact assessment is also warranted, and in certain cases may be mandatory [9]. In Hungary, non-interventional medical research may also fall within the approval authority of the ETT TUKEB (Scientific and Research Ethics Committee) or the competent regional research ethics committee [10].

All data processing must take place within the hospital's controlled IT environment. Identifiable patient data may not be uploaded to a public or consumer AI service. The model development data, the re-identification key, and the researcher's analytic environment must be kept separate from one another.

The research must follow the 2024 version of the Declaration of Helsinki, which also extends to research using identifiable health data and prescribes protection of participants' rights, privacy and wellbeing [11].

Expected outcome

The research is expected to result in an integrated decision-support system that treats a patient's condition not merely as a sum of clinical risks. It also takes communicative, psychological and social resources into account, thereby creating an opportunity for earlier recognition of vulnerability and for personalised, salutogenically oriented prevention.

Bibliography and ethical sources

Related works by Krisztián Dombrádi

• Dombrádi, K. & Székely, A. (2026). A Sociological-Theoretical Description of the Bio-Psycho-Social Background Factors of Health. Belvedere Meridionale, 38(1), 164–181. DOI: 10.14232/belv.2026.1.12 · MTMT record
• Dombrádi, K. (2011). Familiar Stranger: Communication and Social Capital. Budapest: Századvég Kiadó. 167 p. ISBN 978-963-7340-97-0. MTMT record
• Dombrádi, K. (2009). The Communicative Aspect of Social Capital. In Jancsák, C., Nagy, G. D. & Szűcs, N. (eds.), Céhem vándorkönyvei: Studies for the 60th Birthday of Imre Pászka (pp. 27–43). Szeged: Belvedere Meridionale. MTMT record, full text
• Dombrádi, K. (2025). Communication Theory a Transdisciplinary Science. Budapest–Szeged: Belvedere Meridionale. 177 p. ISBN 978-615-6060-95-2. MTMT record

Theoretical and methodological background

• Engel, G. L. (1977). The Need for a New Medical Model: A Challenge for Biomedicine. Science, 196(4286), 129–136. doi.org/10.1126/science.847460
• Antonovsky, A. (1996). The Salutogenic Model as a Theory to Guide Health Promotion. Health Promotion International, 11(1), 11–18. doi.org/10.1093/heapro/11.1.11
• World Health Organization (2021). Health Promotion Glossary of Terms 2021. Geneva: WHO.

Data protection and research-ethics sources

• European Parliament and Council (2016). Regulation (EU) 2016/679 (GDPR), in particular Articles 9 and 89.
• Hungarian National Authority for Data Protection and Freedom of Information. Data protection impact assessment list.
• Hungarian Medical Research Council. Scientific and Research Ethics Committee (ETT TUKEB) – research authorisation.
• World Medical Association (2024). Declaration of Helsinki: Ethical Principles for Medical Research Involving Human Participants.