Haem.io
Intelligent Diagnostics for Precision Haematology
Haem.io
Intelligent Diagnostics for Precision Haematology
The Problem
Haematology diagnosis is too slow and too complex
Unstructured Data
Genetics, cytogenetics, flow cytometry, morphology, and clinical context arrive in fragmented formats.
Overwhelmed Clinician
A specialist has to reconcile every source before diagnostic reasoning can even begin.
Complex Guidelines
WHO 2022 and ICC 2022 contain hundreds of interconnected pathways that are hard to apply consistently.
The Solution
LLM feature extraction plus formal diagnostic algorithms
Haem.io is the reasoning layer that turns fragmented diagnostic reports into a traceable clinical decision.
An LLM extracts genetic and clinical features. Haem.io's deterministic algorithms use those features for diagnosis, risk stratification, and clinical-trial matching.
Input
PDFs, scans, text reports, genomics, flow cytometry, morphology, and clinical context.
Extraction
An LLM extracts mutations, fusions, cytogenetics, flow markers, morphology, and clinical features.
Algorithms
Structured features are passed into Haem.io algorithms for diagnosis, risk, and trial matching.
Output
Diagnosis, derivation trace, risk stratification, treatment context, and eligible trials.
Product Proof
Production beta, simplified workflow, auditable output

Production beta with 10 haematologists
AML and MDS classifiers are built and being tested before formal NHS pilot validation.
Product proof
- LLM extraction: unstructured reports become structured genetic and clinical features.
- Diagnostic algorithms: extracted features feed diagnosis, risk, and trial matching.
- Full trace: every output can be reviewed and challenged by the clinician.
- Trials layer: leukemia trial matching engine built for UK studies.
Haematology Is The Wedge
Genomics-based diagnosis starts in blood cancer, then expands
Blood cancer is the first wedge for a broader genomics diagnosis platform.
The same architecture can extract molecular features, apply disease-specific algorithms, and connect patients to the next clinical action.
AML / MDS
Production beta in testing with 10 haematologists.
Leukemia trials
UK trial matching engine built from patient genetics and diagnosis.
Lung cancer
Genomics-driven solid tumour diagnosis prototype already built.
More diseases
Further cancers first, then other genomics-led diseases over time.
Business Model + Market
Bottom-up NHS licensing with strategic pull from labs and pharma
Annual SaaS licensing
NHS Trusts
£50k-£100k/year
Private Hospitals
£75k-£150k/year
Diagnostic Labs
£100k-£200k/year
Clinical Validation
Endorsed by leading NHS haematologists
The Team
Founder-market fit across AI, haematology, and NHS diagnostics

Robert Lee
CEO & Co-Founder
Computer science and financial technology background, with experience at Coinbase, LSEG, and FlexTrade. Works across product development, software, regulatory planning, and research coordination.

Dr. Daniel Clarke
CTO & Co-Founder
PhD Physics, University of Manchester and CERN. Former UK Civil Service statistician. Leads cloud architecture, AI strategy, and secure data systems.

Dr. John Burthem
Chief Medical Officer & Co-Founder
FRCP, FRCPath. Senior NHS consultant at Manchester Foundation Trust, regional diagnostic service lead, 50+ publications, and UK NEQAS digital advisory lead.

Dr. Luke Carter-Brzezinski
Clinical Director & Co-Founder
Consultant Haematologist at MFT's Regional Diagnostic Service. Leads clinical outreach, validation strategy, and real-world workflow feedback from practicing clinicians.
Clinical logic co-developed directly with specialist NHS haematologists and translated into auditable software.
The Ask
£750k seed round. 18 months to Series A.
Seed Investment | 18-Month Runway
Team
Regulatory & Pilots
Operations
Series A unlock: regulatory registration + NHS validation + first paid contracts = a repeatable commercial platform for haematology and genomic oncology.
Haem.io
Precision diagnostics for every haematologist, everywhere.
Built by clinicians. LLM-assisted extraction. Algorithmic by design.