MedSWFlow: An Open-Source LLM Workflow for Drafting Medical Social Work Case Plans

MedSWFlow is an open-source framework designed to assist medical social workers in drafting case plans using Large Language Models (LLMs). Rather than relying on a single, monolithic prompt, the system breaks the complex task of case planning into a structured, six-stage workflow: assessment, problem analysis, goal setting, intervention planning, risk anticipation, and planned effect evaluation. By standardizing inputs and enforcing a logical sequence, the framework aims to improve the consistency and transparency of AI-assisted documentation in clinical settings.

The framework is grounded in established social work and behavioral theories, including the person-in-environment perspective, the biopsychosocial model, and the International Classification of Functioning, Disability and Health (ICF). The system architecture consists of five layers—input normalization, case profiling, workflow execution, prompt generation, and document assembly—which ensure that the LLM operates within predefined professional boundaries. This modular design allows practitioners to use different LLM providers while maintaining a consistent, reproducible planning logic.

Medical social work involves high-stakes decision-making where AI-generated content can be prone to hallucinations or the omission of critical contextual factors. MedSWFlow addresses these risks by forcing the AI to work through a structured, evidence-based sequence rather than generating an entire plan at once. By producing reviewable drafts rather than final decisions, the system supports practitioners in maintaining human oversight, ensuring that ethical considerations, resource availability, and client-specific needs are properly evaluated before any service plan is finalized.