🏗️

System Architecture – TRIAGE.AI

Versi 1.0 — November 2025 | Penulis: Hasrinata Arya Afendi (MeowLabs / UNIMUS)

← Kembali

FRONTEND LAYER

💻

Pasien - Web App

(Next.js App Router, Tailwind CSS, Supabase Auth)

[Deployment: Vercel]

Input keluhan
→
← Tampilkan hasil
🖥️

Dashboard Dokter/Admin

(Review Hasil, Override AI)

[Deployment: Vercel]

Data triase &
log aktivitas
↕️
Verifikasi hasil AI
& Catatan

BACKEND LAYER

⚙️

Next.JS Backend API

(API Routes, Supabase SDK, Auth Handling)

[Deployment: Vercel]

Autentikasi, Validasi,
Simpan/Baca Data
↕️
🗄️

Supabase Database

(PostgreSQL, RLS Active, Auth, Storage)

[Deployment: Supabase Cloud]

AI LAYER

Kirim data ke AI Engine
(POST /predict)
→
🤖

AI Triage Engine - FastAPI

(Python, Scikit-learn/ONNX, IndoBERT Opsional)

[Deployment: Render / Railway]

← Hasil prediksi kategori
& urgensi (JSON)
Request Ringkasan Natural
↓
☁️

LLM API (Opsional)

(Gemini / ChatGPT - Ringkasan Natural)

INTEGRATION LAYER

Export FHIR
JSON
→
🔄

FHIR/BPJS API Mock

(Standar Data Kesehatan, Simulasi Antrean)

💻 Technology Stack

Frontend:Next.js 16.0.1, TypeScript, Tailwind CSS 3.4.1
Backend:Next.js API Routes, Supabase SDK 2.81.1
AI Engine:Python FastAPI, Scikit-learn, TF-IDF
Database:Supabase PostgreSQL (RLS Enabled)

⭐ Key Features

✓Hybrid AI (ML + 32 Red Flag Rules)
✓3-Level Urgency (Red/Yellow/Green)
✓Role-Based Access Control (RBAC)
✓Row Level Security (RLS)
✓Bahasa Indonesia Native Support
✓FHIR-Ready Integration

🔄 Data Flow Process

1️⃣

Input Pasien

Keluhan, gejala, vital signs

2️⃣

Validasi Backend

Auth & data validation

3️⃣

AI Processing

Classification & urgency

4️⃣

Save to DB

Store hasil triase

5️⃣

Doctor Review

Verifikasi & catatan

System Architecture Documentation - TRIAGE.AI

© 2025 Hasrinata Arya Afendi (MeowLabs / UNIMUS)