ScienceClaw — Oncology Market Intelligence

Oncology Market Intelligence

An autonomous knowledge graph spanning 8 US oncology market verticals — academic research, drug discovery, clinical trials, manufacturing, sales, distribution, FDA regulatory, and government policy. Updated weekly. No commercial data subscriptions.

Explore the Knowledge Graph View Predictions Methods paper v1.0 (PDF)

746

KG nodes

3591

KG edges

190

Drug × indication predictions

0.884

Approval model CV AUC

—

Verticals loaded

2026-04-04

Last updated

Academic research

PubMed · NIH Reporter

High coverage

Drug discovery

ChEMBL · Open Targets

High coverage

Clinical trials

Thread 2 cross-reference

Daily signal

Manufacturing

openFDA enforcement

Structural only

Sales

CMS Part D / Part B

2yr lag

Distribution

HRSA 340B · NPI

Structural only

FDA regulatory

openFDA · Drugs@FDA

High coverage

Government policy

Fed Register · Congress.gov

Text extraction

⚲

Market Knowledge Graph

Interactive force-directed network across all 8 verticals. Nodes coloured by type: indications, drugs, approvals, sponsors, grants, policy acts, and more. Filter by node type, hover for confidence and source metadata. Tier 3 inferred edges shown as dashed links.

Open KG →

▲

Market Predictions

Sortable predictions table for 190 drug × indication pairs. Four model outputs: approval probability (logistic, CV AUC 0.884), disruption risk (random forest), revenue trajectory index (ridge), and market entry timing (empirical hazard). Expand any row for per-feature contribution bars and raw feature values.

Open Predictions →

06:00 Collect

→

08:15 GLM-5 Tier 2 extract

→

09:00 KG write

→

09:30 Centrality

→

09:45 Lead-lag

→

10:00 GLM-5 classify

→

10:30 Feature matrix

→

10:45 sklearn models

→

11:00 Deploy

Runs every Sunday 06:00–11:00 UK. Staggered before Thread 2 (11:45 UK) to avoid concurrent LLM calls. Single-writer KG with atomic file writes throughout.

Data coverage and model limitations

This platform uses free public data sources only. Manufacturing, distribution, and sales verticals have structural gaps — no drug volume or revenue data is available from public sources at the resolution required for commercial market sizing. CMS Part D spending data has a ~2-year publication lag and covers Medicare-reimbursed spend only. All prediction outputs should be treated as directional signals, not forecasts. This platform is a research initiative and does not substitute for commercial market intelligence services.