StochStack

prototype 09

Data Agent

面向临床研发的数据目录 Agent:用自然语言知道“有哪些数据、在哪里、谁有权限、字段长什么样”。

向 Data Agent 提问

匹配置信度: 0%

命中数据集

Global Claims Longitudinal

来源: externalClaims

Oncology · Cardiovascular · Immunology · CNS

数据 Owner: RWE Strategy

Access Owner: Data Governance Office

存储: lakehouse table

刷新频率: monthly

粒度: patient-level (de-identified)

Schema 字段

patient_token (string) - Tokenized patient key
diagnosis_code (string) - ICD-10 diagnosis code
procedure_code (string) - CPT/HCPCS procedure code
service_date (date) - Date of claim service

EHR Oncology Outcomes Repository

来源: externalRWE

Oncology

数据 Owner: Translational Medicine Data

Access Owner: RWE Access Committee

存储: secure workspace

刷新频率: quarterly

粒度: patient-level with longitudinal labs

Schema 字段

patient_token (string) - Tokenized patient key
tumor_stage (string) - Clinical stage at diagnosis
biomarker_panel (json) - Molecular marker panel
treatment_line (integer) - Line of therapy index

Site Startup and Activation Ledger

来源: internalClinical Operations

Oncology · Immunology · CNS · Cardiovascular

数据 Owner: Global Clinical Operations

Access Owner: Clinical Ops PMO

存储: warehouse mart

刷新频率: daily

粒度: site-study-week

Schema 字段

study_id (string) - Study identifier
site_id (string) - Site identifier
startup_status (string) - Current startup milestone status
cycle_days (integer) - Days elapsed in startup cycle

Patient Screening Funnel

来源: internalPatient

Oncology · Immunology

数据 Owner: Study Operations Analytics

Access Owner: Patient Data Privacy Board

存储: secure mart

刷新频率: weekly

粒度: patient-screening event

Schema 字段

screening_id (string) - Screening event id
screen_fail_reason (string) - Primary reason for failure
site_id (string) - Site where screening happened
age_band (string) - Age bucket

RBQM Unified Risk Signals

来源: internalQuality

Oncology · CNS · Cardiovascular

数据 Owner: RBQM Center of Excellence

Access Owner: Quality Governance

存储: feature store

刷新频率: daily

粒度: site-study-day

Schema 字段

risk_signal_id (string) - Signal id
kri_name (string) - Key risk indicator name
kri_value (float) - Observed indicator value
severity (string) - Risk severity band

Regulatory Submission Document Graph

来源: internalRegulatory

Oncology · Immunology · Cardiovascular

数据 Owner: Regulatory Operations

Access Owner: Regulatory Document Control

存储: document index + graph

刷新频率: daily

粒度: document-version

Schema 字段

document_id (string) - Document unique id
document_type (string) - Protocol/CSR/IB/etc
version (string) - Version number
study_id (string) - Associated study id

Access 指引

Run a query first. The agent will return owner, access level, and recommended request path.

update log

Prototype Change Log

  1. 2026-03-01 · v0.1.0

    Data Agent 目录原型首版

    • - 新增覆盖内外部域的临床研发数据目录。
    • - 新增自然语言查询接口,返回数据集匹配与置信度。
    • - 新增 access 元数据返回:owner、权限等级、存储形态与 schema 预览。