Digital tools

數位工具Digital tools

把研究證據做成實際可用的工具，並用臨床試驗檢驗它們是否真的有幫助。We turn research evidence into usable tools — and test in clinical trials whether they actually help.

待補 · 截圖（不可含真實病人對話）

JanusChoice

論文投稿中Manuscript under review

陪伴新診斷早期肝癌病人的 AI 衛教對話機器人An AI health-education chatbot for people newly diagnosed with early-stage liver cancer

剛確診肝癌的病人，往往在門診後才想到一連串問題。JanusChoice 透過 LINE 回答病人的疑問，回答內容以專家審閱過的肝癌衛教資料為依據（檢索增強生成，RAG），而非讓 AI 自由發揮。我們以多中心隨機對照試驗評估它能否提升病人的健康識能。 People newly diagnosed with liver cancer often think of their questions only after the clinic visit. JanusChoice answers them on LINE, grounding every reply in expert-reviewed patient-education content (retrieval-augmented generation) rather than letting the model improvise. A multicentre randomised controlled trial evaluates whether it improves patients' health literacy.

試驗登錄 NCT06384950Trial registration NCT06384950

待補 · 截圖（不可含真實病人對話）

JitBot

開發中In development

生成式 AI × 即時適應性介入（JITAI）的 LINE 聊天機器人A generative-AI LINE chatbot for just-in-time adaptive interventions (JITAI)

在對的時間給對的支持：JitBot 依使用者當下的狀況提供個人化的提醒與陪伴，協助代謝症候群與糖尿病患者的日常自我管理。 The right support at the right moment: JitBot offers personalised prompts and companionship based on the user's situation, helping people with metabolic syndrome and diabetes manage their health day to day.

聰明做研究：實驗室的 AI 工作流程Working smarter: our AI-powered research workflow

我們自己開發 AI 工具來處理研究中重複、耗時的環節，把時間留給真正需要思考的部分：提出好問題、設計嚴謹的研究、解讀結果。加入實驗室，你也會學會用這些工具工作。We build our own AI tools for the repetitive, time-consuming parts of research, so our time goes to what needs thinking: asking good questions, designing rigorous studies and interpreting results. Join us and you will learn to work this way too.

LINE 研究室助理Lab assistant on LINE

Jarvis the LabbotJarvis the Labbot

老師用一句話交辦工作，Jarvis 自動整理成任務、通知負責的人，進度更新後再回報老師，研究室的大小事不再掉球。Assign a task in plain words; Jarvis turns it into a tracked task, notifies the right person, and reports back as progress is made.

LINE 文獻研讀Paper reading on LINE

老三陪讀Laosan Reading Buddy

傳一篇論文 PDF，老三會判斷研究類型，產出結構化的中文研讀筆記，存進實驗室共用知識庫，讓大家一起讀得更快、更深。Send a paper as a PDF; it identifies the study type, writes structured reading notes and files them in the lab's shared knowledge base.

文獻地圖與選題Evidence mapping

證據缺口地圖Gap Map

針對「介入 × 疾病」等主題自動檢索 PubMed，產生互動式證據缺口地圖，一眼看出哪裡研究多、哪裡還是空白。Searches PubMed for any "A × B" topic and builds an interactive evidence gap map showing where research is dense — and where it is missing.

期刊投稿準備Journal submission

投稿儀表板Submit Dashboard

整理目標期刊的規範，自動檢查格式與參考文獻、草擬 cover letter，把投稿前的瑣事變成一張清單。Collects a journal's requirements, checks formatting and references, and drafts the cover letter — submission prep as one checklist.

AI 協助流程，不取代判斷。研究設計、分析決策與論文內容由研究者負責；可識別的個人資料與健保資料不送入外部 AI 服務；投稿時遵守各期刊的 AI 使用規範。AI supports the workflow; it does not replace judgement. Researchers own the study design, analytic decisions and writing; identifiable personal and insurance data never go to external AI services; and we follow each journal's AI-use policy.

待補 · 請老師確認：工具名稱與描述、上方「AI 使用原則」三點是否符合實際做法