01
Repeated manual work
Each result takes about three minutes to enter. More tests mean more repetitive work and input risk.
LAB-TO-MARKET · Healthcare Data AX
Turn hospital documents into data assets
Born from multi-institution research, MediPipe structures hospital documents in two seconds. Recent paid validation at Konyang University Hospital generated our first revenue.
MediPipe is our focus today. DeepBreath continues in parallel with clinical and regulatory preparation.
Pulmonary-test PDFPaper and PDF results
OCR + NLP validation3 minutes manual → 2 seconds
Structured PFT dataReady for EMR, research and AI 99% validated
The data foundation for DeepBreath
3 min → 2 sec Processing time per PFT record
99% MediPipe OCR + NLP accuracy
4 institutions Started with multi-institution research
Initial revenue Recent paid validation at Konyang University Hospital
Grand Prize 2026 Joint Lab-Startup Camp IR pitch
Where hospital AX begins
Before AI, data is the bottleneck
The workflow bottleneck
Pulmonary-test results still accumulate as paper and PDF. Clinical staff enter them manually. EMR, research and AI need structured data.
02
Generic OCR falls short
Medical abbreviations, tables, units and reference values require context. Generic OCR falls short of hospital-grade quality.
03
Data that cannot be used
Results trapped in documents cannot support EMR, research or analytics. Hospital AX begins with data transformation.
MediPipe
Practical AX for medical documents
Multi-institution research validated the technology. Recent paid validation moved it into commercialization.
- 01
Collect documents
Bring in pulmonary-test results stored as paper or PDF.
- 02
Recognize and validate
OCR and medical NLP identify and validate values, units and table structures.
- 03
Connect workflows
Standardized data flows directly into EMR, research and medical AI.
Healthcare document AX Flagship · initial revenue generated
MediPipe
We productized data-cleansing technology from multi-institution research. Recent paid validation at Konyang University Hospital confirmed field value.
- OCR + NLP accuracy of 99%
- 3 min → 2 sec per page (98.9% faster)
- Paid joint validation research with Konyang University Hospital
- Automatic EMR integration
- Patent pending (App. No. 10-2025-0166208)
MediPipe data-powered SaMD Long-term pipeline · development and regulatory prep
DeepBreath CDSS
MediPipe data powers five-year deterioration prediction. Clinical and regulatory validation will proceed in stages.
- Predicts a 5-year deterioration trajectory
- Automatic Red / Orange / Green triage
- 4× faster reading
Business and technical validation
Proven in research, validated in revenue
Publications validated the technology. Hospital work confirmed cost savings and willingness to pay.
Grand Prize
External validation
IR pitch at the 2026 Joint Lab-Startup Camp
Hosted by Sungkyunkwan University’s Startup-Centered University program · 2026
Pitch topic
The data infrastructure that lets hospitals actually use AI — MediPipe
01 99% accuracy MediPipe OCR and medical-NLP cleansing performance
02 98.9% faster Per-record processing time · 3 minutes to 2 seconds
03 AUROC 0.890–0.950 Research performance of DeepBreath-related prediction models
Research × Business
Rigorous in research. Fast in execution.
- Multi-institution research across Samsung Medical Center, Incheon National University, Ewha Mokdong and Hallym Chuncheon
- Recent paid validation at Konyang University Hospital · initial revenue
- 15 SCI(E) papers · 8 first-author papers · h-index 8
- Hands-on development of an MFDS-approved SaMD
Selected publications
Deep learning models for acute kidney injury prediction: multi-center external validation and evaluation under simulated continuous monitoring conditions
Separate and joint associations of cardiorespiratory fitness and healthy vascular aging with subclinical atherosclerosis in men
Novel artificial intelligence-based technology to diagnose asthma using methacholine challenge tests
Why MediPipe
Three reasons research becomes product
01
Medical-domain cleansing
MediPipe structures pulmonary values, units, tables and clinical context. The product reflects experience cleansing a 30-year, 1M-record cohort.
02
Field commercialization
Multi-institution research validated the technology. Recent Konyang University Hospital validation generated our first revenue.
03
An expansion path to AI
MediPipe turns documents into data assets. The same flow extends to EMR, research and DeepBreath.
| MediPipe differentiation | MediPipe | Generic OCR / Document AI | Manual entry / outsourcing |
|---|---|---|---|
| Purpose | Healthcare document AX | Generic character recognition | Basic data entry |
| Medical context | Pulmonary-domain specific | Limited | Depends on operator experience |
| Cleansing quality | OCR + NLP 99% | Varies by document type | Human-error risk |
| Processing speed | 2 seconds per page | Post-processing required | About 3 minutes per page |
| Extensibility | EMR, research and AI | Requires separate development | Limited data reuse |
Growth strategy
Enter with MediPipe, expand with DeepBreath
MediPipe secures hospital workflows and early revenue. Its data foundation supports predictive AI and global expansion.
Business progress From multi-institution research to revenue and medical AI
RESEARCHMulti-institution study
4 institutions The work began across Samsung Medical Center, Incheon National University, Ewha Mokdong and Hallym Chuncheon.
NOWFirst revenue
Konyang Univ. Recent joint validation generated our first paid revenue and confirmed field value.
SCALEExpansion opportunity
1.42M / year Growing pulmonary-test data demand created by Korea’s screening expansion.
PIPELINELong term
DeepBreath Predictive SaMD development and regulatory preparation continue on MediPipe data.
Roadmap MediPipe → DeepBreath
in three stages
- 2026AX commercialization
- Conduct multi-institution research
- Incorporation (Aug 2026)
- Paid Konyang University Hospital validation · initial revenue
- Grand Prize at the Joint Lab-Startup Camp IR pitch
- MediPipe official launch
- 2027Hospital expansion
- Expand MediPipe hospital adoption
- Extend EMR integration and document coverage
- Accumulate pulmonary data assets
- Prepare DeepBreath clinical and regulatory work
- 2028SaMD expansion
- Advance DeepBreath approval
- Full clinical operation
- Prepare global regulatory pathways
Going global
MediPipe references support overseas expansion. We will pursue DeepBreath approval in Korea and abroad.
Team
From the lab to the field
Breath You is a lab-based startup from Sungkyunkwan University. We keep research rigorous and test products and business quickly.
Kyung Hyun Lee
이경현
Co-founder & CEO He studies clinical problems. He turns the findings into products and business.
- 8 years in Medical AI
- 15 SCI(E) papers · 8 first-author
- h-index 8
- MFDS-approved SaMD experience
Dayoon Lee
이다윤
Co-founder & Chief Quality Officer She translates clinical practice into quality and regulatory systems.
- Clinical insight from nursing
- Medical-device QA
- QMS development
- Regulatory process
Collaboration network
We began with multi-institution research. Recent paid validation at Konyang University Hospital moved the work into revenue.
Contact
Start hospital AX with MediPipe
We welcome MediPipe adoption and PoC inquiries. Research and investment materials are available upon request.
Email us directly lkh256@breathyou.care
- MediPipe adoption / AX
- Research collaboration
- Investment / other
We review product adoption and clinical or research collaboration.