HANOI, VIETNAM, SEPTEMBER 29, 2026 - OmiGroup, a Vietnam-based healthcare IT company with AI research capabilities, today announced the successful completion of a research and development proof-of-concept (POC) called OmiKG, an evidence-grounded causal reasoning engine for chronic disease built on Amazon Web Services (AWS). The research and POC will be presented at the AMIA 2026 Annual Symposium, a leading global forum for biomedical and health informatics, where it will introduce OmiKG's evidence-grounded methodology to the international biomedical informatics community.
The global AI in healthcare market is projected to grow from USD 36.67 billion in 2026 to USD 194.79 billion by 2031 (MarketsandMarkets), driven by growing demand for AI-powered diagnostics, drug discovery, and clinical decision support. Yet most AI systems operate as "black boxes," unable to explain their reasoning or trace conclusions back to evidence. Clinicians need transparency before they trust AI with patient outcomes. OmiKG’s causal reasoning engine bridges that gap, linking every recommendation to verifiable evidence for medical teams in Vietnam and globally.
Solving Healthcare AI's Biggest Challenge: Verifiability
OmiKG originated as a nutrition advice app for diabetic patients before pivoting to causal inference. Built and hosted entirely on AWS, Amazon Bedrock powers two core capabilities: extracting causal relationships from biomedical literature during knowledge graph construction, and analyzing patient-specific causal pathways at runtime. Amazon EC2 provides the compute infrastructure and database hosting.
“We started by building a nutrition advice system for diabetic patients and quickly discovered that the hardest problem was not accuracy, but verifiability,” said Nhung Nguyen, Head of AI Research, OmiGroup. “If clinicians cannot trace how an AI model reached its conclusions, they will not trust it. That realization drove us to pivot toward structured, evidence-grounded reasoning. Amazon Bedrock made it possible for a small research team to develop and validate this at the pace that would otherwise require significantly more resources.”
Redesigned as a discovery-and-reasoning framework supporting healthcare practitioners, OmiKG goes beyond listing symptoms and conditions. Rather than just generating free-text explanations, OmiKG maps patient clinical inputs onto a domain-specific, evidence-graded knowledge graph of disease mechanisms. Every causal relationship carries its published source and evidence grade, enabling full traceability from conclusion back to peer-reviewed literature.
OmiKG’s causal pathway visualization is able to trace upstream factors through biological dysfunctions and physiological systems all the way to clinical phenotypes.Synthetic patient data shown above.
OmiKG helps clinicians dive beyond surface-level diagnosis to root causes. Practitioners can inspect every causal link, verify its source, and override the system's reasoning before any output reaches a patient. A patient's fatigue, for instance, may trace back through independent causal pathways — such as hypocortisolemia, IL-6 elevation, and chronic inflammation — each pointing to a different molecular mechanism that needs a different intervention target. Two identical patients with type 2 diabetes diagnoses may have entirely different root causes: such as toxic environment exposure versus a high-stress, ultra-processed food diet. By making these molecular pathways visible and evidence-linked, OmiKG helps practitioners understand the biological reason "why" behind a condition, not just the "what”.
How OmiKG Fared In Testing
In a controlled research evaluation, two independent clinical experts blindly assessed OmiKG's outputs across 20 clinical questions related to Type 2 Diabetes mechanisms. OmiKG successfully identified 15 of 17 representative molecular mechanisms, versus one and none for two alternative AI approaches. OmiKG also was on par with general-purpose AI when tested for breadth of coverage, which meant its structured, evidence-grounded approach does not sacrifice completeness for depth.
The evaluation also identified clear priorities for OmiKG’s next bound of development. The system's transparent architecture made it possible to pinpoint specific areas for improvement in citation faithfulness, something opaque AI systems cannot provide. OmiKG is also now focused on adding clinical utility and intervention knowledge, the “what to do” to complement its “why” answers, as the primary objective of OmiGroup's next research phase.
Taken together, the evaluation demonstrated OmiKG’s core advantage: mechanistic specificity, the ability to surface and trace molecular-level disease pathways that general-purpose AI approaches did not identify. It also demonstrated that structured, evidence-grounded reasoning adds measurable value for explaining complex disease mechanisms, while identifying clear priorities for continued research.
“OmiGroup’s work tackles one of healthcare AI’s most important challenges: making AI reasoning transparent, traceable, and explainable to clinicians,” said Eric Yeo, Country General Manager for AWS Vietnam. “Their proof-of-concept demonstrates that a small, focused research team can use AWS to build something new and potentially impactful in healthcare, and we look forward to supporting them as they move from research validation to real-world application.”
What Comes Next: From Research Lab to Global Stage
With the proof-of-concept validated by independent clinical experts, OmiGroup is pursuing a phased path from research to real-world application:
- Global academic presentation: OmiKG has been accepted for oral presentation at the AMIA 2026 Annual Symposium, a leading global forum for biomedical and health informatics. The presentation will introduce OmiKG's evidence-grounded methodology to the international biomedical informatics community.
- Clinical pilot: OmiKG is exploring integration pilots with health data platform partners to test the system within real-world clinical workflows under clinician supervision.
- Commercialization planning: OmiGroup is in discussion with potential partners to bring OmiKG to market, with Vietnam, Japan, and Korea as priority markets given the company’s existing healthcare IT presence in these countries.
###
About OmiGroup
OmiGroup is a Vietnam-based healthcare IT company with more than 14 years of experience in healthcare IT. Combining digital health expertise with strong technology capabilities, OmiGroup supports healthcare transformation across Vietnam, Japan, and South Korea. Its areas of focus include global healthcare digital transformation, smart healthcare solutions, and healthcare products & services. Building on this foundation, OmiGroup also invests in AI research to explore new approaches to healthcare intelligence and evidence-based reasoning. OmiKG is one of the outcomes of this research, developed by OmiGroup’s AI research team to make AI-generated medical reasoning more transparent, traceable, and grounded in published evidence.