Abstract
Context: Population segmentation is a critical health system planning activity that enables more integrated, needs-responsive, and sustainable care. This paper describes the development and evaluation of a Systemic Health System Population Segmentation Model based on the person-centred and needs-based Unified Care Model by Yishun Health, a regional population health system in Singapore. We highlight three implications to enhance health systems operational relevance: (i) psychosocial factors as key determinants of outcomes, (ii) accountability and resource allocation across differentiated segments, and (iii) integration of lifelong and episodic care needs. Methods: Three interdependent models were developed, a Lifelong Care Segmentation Model, a Needs-Based Sub-Segmentation Model, and an Episodic Care Segmentation Model, all underpinned by the Unified Care Model. These models systematically stratify residents into mutually exclusive and collectively exhaustive population groups based on biopsychosocial needs across different health system levels. An expert-driven design process was used, supported by integrated administrative and clinical data. Model evaluation examined the ability to stratify patients into distinct risk groups using healthcare utilisation, costs, and readmission outcomes. Findings: In 2022, 78,810 residents were segmented into seven lifelong care segments, with 43,473 residents with chronic conditions further stratified into sub-segments reflecting varying complexity and psychosocial needs. Additionally, 14,335 emergency admissions were categorised into six episodic care segments. Healthcare utilisation and annual healthcare costs differed significantly across needs-based sub-segments (p < 0.001). Higher episodic care needs were associated with longer hospital stays, higher rates of emergency readmissions, and admission costs (p < 0.001). Psychosocial issues consistently emerged as a key determinant of poorer outcomes, underscoring implications for more systemic and systematic accountability assignment and more deliberate resource planning, especially for care integration horizontally. The integration of lifelong and episodic care needs further enabled operational redesign for vertically integrated health systems. Conclusions: By incorporating psychosocial drivers, focusing on clarifying accountability and resource allocation, and lifelong-episodic care integration, our Systemic Health System Population Segmentation Model strengthens the operational utility of segmentation as a foundation for population health system transformation and provided a robust framework for health systems governance and leadership system redesign globally.