月旦知識庫
月旦知識庫 會員登入|元照網路書店|月旦品評家
 
 
  1. 熱門:
首頁 臺灣期刊   法律   公行政治   醫事相關   財經   社會學   教育   其他 大陸期刊   核心   重要期刊 DOI文章
醫療品質雜誌 本站僅提供期刊文獻檢索。
  【月旦知識庫】是否收錄該篇全文,敬請【登入】查詢為準。
最新【購點活動】


篇名
生成式 AI 優化醫學圖書館諮詢服務之專案實踐
並列篇名
Optimizing Medical Library Consultation Services Through Generative Artificial Intelligence: A Project Implementation Study
作者 汪秀玲 (Hsiu-Ling Wang)、徐嫚伽、柯宏龍、黃志仁
中文摘要

目的:面對醫療數位轉型與教學醫院評鑑要求,醫學圖書館傳統人工諮詢受限於人力與時效,難以滿足臨床全天候需求。本專案旨在導入生成式AI技術,建構智能諮詢系統以優化服務效能。方法:以高醫岡山醫院圖書室為場域,組成跨部門團隊,識別出服務時間、回應速度、人力負荷及技術應用四項服務落差。團隊採用 Anthropic 的Claude與檢索增強生成(Retrieval-Augmented Generation, RAG)技術,透過 Claude Artifacts 低程式碼環境,整合院內圖書館規章、電子資源及學術倫理指引,開發具「主題選單」與「自然語言對話」響應式雙模組介面的「LibraryChatbot」,並內建防錯與人工轉接機制抑制AI幻覺。結果:系統上線後三個月評估顯示,每月由 AI 成功處理的諮詢案約 420 件,人工處理之簡易諮詢案由 500 件降至 80 件(降幅 84%);圖書館員得以專注於進階學術諮詢,服務時間由每週 47.5 小時延伸至 168 小時全天候涵蓋。實測顯示系統在事實性、邏輯性及資源整合皆具準確度。結論:本專案建立具成本效益且可複製的醫學圖書館AI模式,有效釋放館員人力並實現全天候服務。未來以 API 擴充系統功能,進一步深化數位賦能,為行政單位數位轉型提供實務參考。

英文摘要

Objective: Medical libraries face growing demands related to digital transformation in healthcare and to teaching hospital accreditation requirements. However, traditional manual consultation services are often constrained by limited staffing and service hours and struggle to meet the 24/7 demands of clinical settings. This study developed an intelligent consultation system based on generative artificial intelligence (AI) to optimize service performance in medical libraries.Methods: This study was conducted in a library setting at Kaohsiung Medical University Gangshan Hospital, Kaohsiung, Taiwan. A cross-functional team identified four primary service gaps: availability, response speed, workload, and technical application. By using Claude and retrieval-augmented generation within the Claude Artifacts low-code environment, the team integrated institutional regulations, electronic resources, and academic integrity guidelines to develop LibraryChatbot. This responsive system features a dual-module interface that combines a “topic menu” with “natural language dialogue” and includes builtin error prevention and human handover mechanisms to mitigate AI hallucinations.Results: A 3-month postlaunch evaluation revealed that the system successfully handled approximately 420 consultations per month. It reduced the number of simple consultations handled manually from 500 to 80 cases per month (an 84% reduction), which allowed librarians to focus on advanced academic inquiries. It also increased service availability from 47.5 hours per week to 24/7 coverage (168 hours per week). Testing confirmed the system’s high accuracy in factual consistency, logical reasoning, and resource integration.Conclusion: This study constructed a cost-effective and replicable AI model for medical libraries that effectively frees librarians’ time and ensures 24/7 service availability. Expanding system functions through the application programming interface (API) is expected to further strengthen these digital capabilities and provide a practical reference for the digital transformation of other administrative units.

起訖頁 036-045
關鍵詞 生成式人工智慧、醫學圖書館、智能諮詢系統、品質改善、數位轉型、generativeartificialintelligence(GenAI)、medicallibrary、intelligentconsultationsystem、digitaltransformation、qualityimprovement
刊名 醫療品質雜誌  
期數 202611 (20:6期)
出版單位 財團法人醫院評鑑暨醫療品質策進會
該期刊-上一篇 慢性病照護中的醫病共享決策:敘述性文獻回顧與概念框架初探
該期刊-下一篇 腦傷照護一生守護:疾病照護品質認證經驗分享
 

新書閱讀



最新影音


優惠活動




讀者服務專線:+886-2-23756688 傳真:+886-2-23318496
地址:臺北市館前路28 號 7 樓 客服信箱
Copyright © 元照出版 All rights reserved. 版權所有,禁止轉貼節錄