隨著醫療技術發展與人口高齡化趨勢加劇，醫療人力需求日益攀升，導致人力缺口逐年擴大。為提升醫事人員留任意願，建立具彈性且高滿意度的排班模式，對降低人員流動率與緩解人力壓力具有關鍵意義。
本研究運用混合整數規劃 ( Mixed-Integer Programming, MIP ) 模型，分別採用精確方法（Gurobi 標準參數）與結合啟發式調整參數，進一步探討搭配大型鄰域搜尋 ( Large Neighborhood Search, LNS ) 與否對排班效能之影響。研究模擬不同人力配置情境（20人、10人、9人與8人），以全面評估模型於各種營運條件下之表現。
以10位醫事放射師為對象進行個案模擬，結果顯示儘管結合 LNS 策略有助於提升緊縮人力情境下之可行性，但整體滿意度問卷與績效指標顯示，MIP 模型於 Gurobi 標準參數下之精確方法，於排班公平性與偏好滿足程度方面評價最高，展現良好之實務應用潛力。
整體而言，本研究所提出之模型可有效縮短人工排班所需時間約 50%，由原先 2小時以上縮短至約 1 小時內，顯著提升排班效率，並兼顧公平性與員工滿意度。研究成果提供具體可行之排班優化策略，對醫療單位人力資源管理具高度應用價值。

With rising demands driven by medical advancements and an aging population, healthcare institutions face increasing workforce shortages. To improve staff retention and scheduling efficiency, this study develops a flexible, satisfaction-oriented scheduling model based on Mixed-Integer Programming (MIP).
We compare exact methods using default Gurobi parameters with heuristic-enhanced approaches incorporating parameter tuning and Large Neighborhood Search (LNS). Simulations across different staffing levels (20, 10, 9, and 8 personnel) evaluate the model's adaptability under various constraints.
A case study with 10 radiologic technologists shows that while LNS improves feasibility in shortage scenarios, the exact MIP method consistently achieves the highest ratings in fairness and preference satisfaction, supported by survey responses and performance metrics.
Additionally, the proposed model significantly reduces manual scheduling time by about 50%, cutting it from over two hours to under one. These results demonstrate the model’s potential for real-world application in healthcare workforce planning, offering a balance between operational efficiency and human-centered design.