Triage the problem. Geek the solution.

CH.01 · ABOUT

看見痛點，打造解方See the pain, build the cure.

CH.02 · PRODUCTS

兩個產品，一條主線Two products, one thread

我們從急診室出發，發現高齡長者的多重用藥正是造成非預期急診就醫的重要原因——於是把戰線往前推：在傷害發生之前介入，而不是在急診室補救。 We started in the emergency room — and found that polypharmacy among older adults is a major driver of unplanned ER visits. So we moved upstream: intervene before harm happens, instead of fixing it in the ER.

● ACTIVE LONG-TERM CARE MVP

記藥又藥JiYao YouYao

長照安全用藥 AI 決策支援系統AI medication-safety support for long-term care

拍一張藥袋，少 key 一百筆。為住宿型長照機構的護理師與照服員設計的安全給藥 App：AI 讀懂跨院、格式不一的藥袋，再交給規則與藥證資料庫把關。 Snap the drug bag, skip a hundred keystrokes. A medication-safety app for nurses and caregivers in residential care: AI reads drug bags from any hospital, then rules and a licensed-drug database double-check the result.

• 01藥袋／藥單拍照 AI 辨識：藥名、劑量、頻率、時段一次建檔Photo-to-record: AI extracts drug name, dose, frequency and timing
• 02自動比對衛福部約 7.2 萬筆藥證資料，防止 AI 幻覺Cross-checked against ~72,000 MOHW drug licenses to prevent AI hallucination
• 03依早、中、晚、睡前列出給藥清單，記錄完成／拒服／暫緩Time-slot medication rounds with done / refused / on-hold logging
• 04劑型磨粉警示（緩釋、腸溶等不宜磨粉）與重複醫囑檢查Crush-warning for extended-release / enteric-coated forms, duplicate-order checks
• 05多機構資料隔離、角色權限、HL7 FHIR 相容Multi-facility isolation, role-based access, HL7 FHIR compatible

// 定位：記錄 · 提示 · 軌跡——輔助專業判斷，不做醫療決策 // Record · Remind · Trace — supports professional judgment, never replaces it

◆ ORIGIN EMERGENCY TTAS

急刻診答TriaGeek ER

智慧急診 AI 分流輔助系統AI-assisted emergency triage support system

我們的第一個產品，也是技術的起點。病患在候診時以手機或平板完成 AI 問診，系統依台灣檢傷分類標準（TTAS）產出建議級數、鑑別診斷、分科與檢查；醫師端可檢視、修正，修正結果再回饋給模型持續學習。 Our first product and technical foundation. Patients complete an AI-guided intake on a phone or tablet while waiting; following Taiwan’s triage standard (TTAS), the system suggests acuity level, differential diagnoses, department and tests. Physicians review and correct — and every correction feeds back into the model.

01檢傷級數 1–5Acuity 1–5

02鑑別診斷Differentials

03建議分科Department

04建議檢查Suggested tests

CH.03 · TECHNOLOGY

Neuro-Symbolic 雙引擎The Neuro-Symbolic dual engine

大型語言模型擅長理解，卻可能產生幻覺；規則與資料庫可靠，卻不懂人話。我們讓兩者各司其職：AI 負責理解，規則負責把關，人負責決定。 Large language models understand but can hallucinate; rules and databases are reliable but can’t read human language. We let each do what it does best: AI understands, rules verify, humans decide.

"resourceType": "Bundle",
"entry": [
  { "Patient" },
  { "Condition" },
  { "AllergyIntolerance" },
  { "MedicationStatement" }
]

CH.04 · JOURNEY

我們的心跳紀錄Our heartbeat, so far

往下捲動，每一次跳動都是一個里程碑。Scroll down — every beat is a milestone.

CH.05 · AWARDS & EXHIBITIONS

獎項與參展Awards & exhibitions

CH.06 · TEAM

跨領域的五人小隊A cross-disciplinary crew of five

來自中原大學生物醫學工程學系，橫跨 AI、臨床、法規、商業與設計，並持有 BLS、EMT-1 等急救證照。From CYCU Biomedical Engineering, spanning AI, clinical work, regulation, business and design — with BLS and EMT-1 certifications on the team.

CH.07 · CONTACT