Explainable Knowledge-Guided Beam-Search Framework for Hospital Layout Generation
Authors:
- Niloufar MasoumzadehNMNiloufar Masoumzadeh
- Rahbar MortezaRahbar Morteza
- Ahmad EkhlassiAEAhmad Ekhlassi
Preprint
- 21 September 2026
- preprint
- Published by Elsevier in SSRN Electronic Journal
Abstract
Hospital layout design requires balancing functional relationships, spatial constraints, circulation, and geometric performance. This study presents a knowledge-guided beam-search framework to support early-stage hospital layout generation. The framework integrates architectural rules, weighted functional relationships, spatial skeletons, and beam-search exploration to generate multiple feasible layout alternatives. Rather than following a single placement sequence, the method preserves promising partial layouts during generation and evaluates alternative spatial arrangements before final selection. Generated layouts are assessed using seven criteria, including functional adjacency, area accuracy, circulation efficiency, compactness, spatial proportion, site-envelope compatibility, and architectural rule compliance. The framework was implemented in Python and evaluated using a representative surgery-center case study through multi-seed experiments and benchmark generation strategies. The results demonstrate that the approach can produce high-performing layouts while maintaining functional and geometric feasibility, providing an interpretable decision-support method for conceptual hospital design.Keywords
This publication has 2 references indexed in Scilit:
- Data-driven interior plan generation for residential buildingsACM Transactions on Graphics, 2019
- An approach to hospital planning and design using discrete event simulation2007