01
Chie Weng Tham
Chief Executive Officer
Privacy-first · Clinician-reviewed · ExplainableFrom scattered PDFs
From scattered PDFs
to clinical clarity.
medX helps clinics extract, organize, compare, and explain blood test results across patient visits — without relying on a black-box diagnosis engine.
Sarah K., 34F
Results
Trends
Hypotheses
Complete Blood Count
Haemoglobin10.8 g/dL12.0–16.0
MCV71.2 fL80–100
WBC6.4 ×10⁹/L4.0–11.0
Metabolic Panel
Ferritin8 µg/L20–200
Serum Iron9.1 µmol/L9–30
B12412 pmol/L180–900
Hypothesis · pending review
Low Hgb + low MCV + depleted ferritin + borderline serum iron suggests iron-deficiency pattern. Consistent across 2 of 3 visits.
Early access validation
Validated with a real clinic workflow
medX is being shaped with feedback from clinical teams reviewing blood test PDFs, longitudinal results, and explainable interpretation flows.
Verified withTCM Clinic MalaysiaAdditional clinic pilots in progress
Interactive workflow
See the report split into reviewable clinical layers
Select each layer to see how medX moves from a flat blood test PDF to structured evidence a clinician can inspect.
01Source report
Original PDF
A clinic uploads the blood test report exactly as it was received.
Lab report pagePatient visit dateSource trace retained
The original document remains the source of truth for every extracted value.
How it works
Five steps from PDF to explainable insight
medX is not a black box. Each step is traceable back to the original report, value, reference range, and source data.
01
Extract
Convert blood test PDFs into structured biomarker records — values, units, reference ranges, and report metadata — regardless of lab format.
02
Organise
Build longitudinal patient records across visits. Compare current and historical results side by side with biomarker trend visualisations.
03
Flag
Identify markers that are outside range, close to clinical boundaries, or changing meaningfully across repeated reports — not just the latest snapshot.
04
Pattern
Group related marker signals into intermediate reasoning artifacts. Clinicians see how individual findings connect before a hypothesis is formed.
05
Hypothesise
Suggest possible clinical interpretations with visible supporting evidence, contradictions, missing data, and source traceability. Clinicians review, edit, and approve.
Built for clinical trust
Every output is traceable
AI can generate a polished report. medX is built to show the reasoning behind it.
Deterministic reasoning layer
The LLM is not the decision-maker. Structured biomarker logic, deterministic reference-range checks, and knowledge graph relationships form the core. The LLM explains and communicates what the system found — it does not invent it.
Longitudinal records
Blood test interpretation requires context. medX tracks every patient across visits, making it possible to detect trends that a single-report view would miss.
Privacy-first architecture
Patient data never leaves your clinic's environment. medX is designed for clinical-grade data handling from the ground up, not retrofitted for compliance.
Clinician review at every step
No output is used without clinician approval. Every hypothesis is editable. Every flag is reviewable. medX works with clinical judgment, not around it.
Our principle
Don't hide the reasoning.
Every hypothesis should be traceable. Every pattern should show the markers behind it. Every marker should link back to the original value, unit, reference range, and source report.
The reasoning pathway
What clinicians need to trust an interpretation
Early AI tools generated impressive summaries. But summaries alone are not enough. A clinician needs to see the chain.
See a live example01What value was extracted from the original report?
02Which reference range was used?
03Why was this marker flagged?
04Was the marker abnormal, borderline, or trending?
05Which markers formed a pattern?
06What evidence supports the hypothesis?
07What evidence contradicts it?
08What information is missing?
See medX in your clinic's workflow
Request a demo and we'll walk you through a full blood test review — extraction, flagging, patterns, and hypotheses — using de-identified sample data.
No sales pressure. 30-minute session.
Our Team
The people building medX
A focused founding team combining product, engineering, and clinical workflow thinking.
