StochStack
Back to Signal

signal logs

Data Foundation Architecture

六层数据架构映射：Clinical Authoring 2.0 + Site Feasibility + Trial Simulation

Data Foundation 架构

Clinical Authoring 2.0 + Site Feasibility + Trial Simulation

本文档按三个实际项目映射到 Data Foundation 六层架构，每层列出关键数据对象、责任归属、刷新频率、合规要求与质量KPI。 统一治理字段贯穿所有层级，确保数据血缘可追溯、合规可审计。

六层架构概览

L5

Apps / Agents

3 objects • 3 KPIs

L4

Data Products / Feature Store

4 objects • 3 KPIs

L3

Semantic / Canonical

5 objects • 3 KPIs

L2

Conformed / Clean

4 objects • 3 KPIs

L1

Landing / Raw

3 objects • 4 KPIs

L0

Source Systems

3 objects • 3 KPIs

数据流向: Source → Raw → Clean → Canonical → Products → Apps

统一治理字段

(所有层/项目通用)
Source System
Source Object
Extract Batch ID
Extract Timestamp
Study ID (Canonical)
GxP Flag
PII Flag
Region Flag
Data Quality Score
Access Policy ID

详细层级映射

点击展开详情

最小可用 Data Foundation (MVD)

建议为三项目各选 10 个“必须有”的数据对象，定义最小可用 Data Foundation。每个对象需明确： Owner / Refresh / Compliance / KPI，形成可管理的产品 backlog。

Clinical Authoring 2.0

10 critical objects

Focus: Document lineage & traceability

Site Feasibility

10 critical objects

Focus: Site matching & performance

Trial Simulation

10 critical objects

Focus: Event timelines & assumptions

Source Systems
Raw Data
Clean Data
Semantic Model
Data Products
Applications
Published: 2026-03-01
Category: Data Architecture
Projects: Clinical Authoring 2.0, Site Feasibility, Trial Simulation
Layers: 6 (L0-L5)