Leveraging a proprietary Clinical Knowledge Graph

Proprietary architecture

Most AI tools pull from the open internet. asksam™ works differently. It runs on a proprietary, verified clinical knowledge environment built specifically for healthcare. At its core is a purpose-built clinical knowledge graph that brings together medical literature, clinical guidelines, and structured health data into a single connected network. The graph integrates close to 15 million medical concepts and over 188 million relationships, spanning 38 distinct entity types (including genes, proteins, diseases, symptoms, drugs, and clinical trials.

Trained with Australian medical literature

The asksam™ Clinical Knowledge Graph is trained on Australian medical literature and nationally recognised clinical guidelines. This means responses reflect the standards, terminology, and care pathways that Australian healthcare professionals actually use every day. The system is also strengthened by the body of knowledge recommended by the World Health Organisation, including the ICD-11 classification framework.

Hallucination mitigation technology

AI scribes typically use large language models to both reason and respond, which can lead to hallucinations. asksam™ takes a different approach. The clinical knowledge graph acts as the clinical brain, not the LLM. Because reasoning is grounded in a densely connected, validated medical knowledge structure, asksam™ can distinguish between causality and correlation, apply temporal reasoning to track how conditions and treatments evolve, and navigate complex multi-factorial interactions that simpler AI systems often miss.

Leveraging Australian & Global Medical Literature & Guidelines

Cochrane

PubMed

UptoDate

BMJ

RACGP

Nice Guidelines

NHMRC

CINAHL

Scopus

Curious how it works?