01
Limits of single-point reading
A single day’s pulmonary measurement must somehow predict five years ahead. Snapshot diagnosis cannot identify high-risk COPD and asthma patients early.
Medical AI · Predictive Pulmonary Solutions
Bringing the silent 5 years forward
Breath You builds medical AI that analyzes pulmonary function test data to predict respiratory deterioration five years ahead. As Korea’s 2026 national screening expansion opens a market of 1.42M reads per year, we deliver AI that both clinicians and patients can feel.
Why now
A respiratory-health market opens at once
Jan 2026 · National screening expansion
Korea is expanding pulmonary function testing to roughly 1.78M citizens aged 56–66. At an 80% participation rate, that is 1.42M new reads per year — a volume the current nationwide reading workforce cannot handle.
02
No data infrastructure
80–90% of medical data is unstructured. Pulmonary results survive only as paper and PDF, leaving no clean data for AI to learn from.
03
No patient-side channel
Patients have no way to know their own 5-year outlook. Diagnosis rates sit at 2.5% — 9 in 10 lose half their lung function unaware.
Solution
Two tracks, one pipeline
Today, Breath You enters the clinical field with MediPipe — turning hospitals’ unstructured data into an asset and generating early revenue — while preparing DeepBreath CDSS to predict five years ahead on top of that cleansed data.
Hospital data-asset solution Flagship · launching soon
MediPipe
Multi-agent OCR and NLP convert paper and PDF pulmonary results into structured data — cutting a 3-minute manual task to 2 seconds and turning a hospital’s scattered unstructured data into a usable asset. Having completed its first proof of concept, it is Breath You’s current flagship solution, with an initial product launch due within two months.
- OCR + NLP accuracy of 99%
- 3 min → 2 sec per page (98.9% faster)
- Automatic EMR integration
- Patent pending (App. No. 10-2025-0166208)
AI prediction medical device (SaMD) In development & regulatory review
DeepBreath CDSS
A time-series prediction model trained on Samsung Medical Center’s 30-year, 1M-record pulmonary cohort. From a single test it forecasts a 5-year deterioration trajectory, delivered with traffic-light triage and explainable reports for clinicians and patients alike.
- Predicts a 5-year deterioration trajectory
- Automatic Red / Orange / Green triage
- SHAP-based explainable reports
- 4× faster reading
Validated
Technology proven by data
Breath You’s technology is validated through peer-reviewed publications and conference presentations, with clinical validation ongoing at partner hospitals.
AUROC 0.950 Bronchial challenge diagnostic model · published in AAIR 2024
AUROC 0.890 Pulmonary-test prediction model · presented at EAACI 2025
80% intent-to-use Usability evaluation, 5 specialists
CEO Kyung Hyun Lee is an 8-year medical-AI researcher with 15 SCI(E) papers (first author on 8), an h-index of 8, and over 400 cumulative citations.
Selected publications
Deep learning models for acute kidney injury prediction: multi-center external validation
Separate and Joint Associations of Cardiorespiratory Fitness and Healthy Vascular Aging with Subclinical Atherosclerosis
Novel AI-Based Technology to Diagnose Asthma Using Methacholine Challenge Tests
Advantage
Three things competitors cannot copy
01
Data moat
The team that first cleansed Samsung Medical Center’s 30-year, 1M-record cohort is our own engineering team. Even with the same data, no one can use it without the cleansing know-how.
02
Regulatory head start
Our CEO personally led MFDS approval and clinical trials for two SaMDs. With QMS and GMP systems in place, we are 1.5–2 years ahead of competitors.
03
Dual safeguard
70% of medical-AI failures stem from bad input. MediPipe validates the input; SHAP XAI validates the output rationale. No other medical-AI company — in Korea or abroad — covers both.
| Competitive landscape | Breath You | Domestic pulmonary-device makers | Overseas pulmonary apps |
|---|---|---|---|
| Core capability | AI prediction SaMD | Pulmonary-measurement hardware | Simple pulmonary apps |
| Analysis method | 3–10 yr time-series trajectory | Single-day cross-section | Short-term trend (months) |
| Prognosis prediction | Yes (5-year trajectory) | Not possible | Limited |
| Diagnostic performance | AUROC 0.890–0.950 | No AI | Some AI (single point) |
| Data asset | SMC 1M records over 30 yrs | Own measurements | Overseas-data centric |
Market
A vast market, a clear entry point
Korea’s 2026 national screening expansion opens the respiratory-health market at once. Breath You starts from the most validated foothold.
TAM
~KRW 65T
Global
WHO estimates 392M COPD patients worldwide.
SAM
KRW 60B+
Korea
Korean serviceable market built on 1.42M national-screening reads per year.
SOM
KRW 20B
Within 3 yrs
Obtainable within three years, centered on tertiary general hospitals.
LAM
KRW 625M
First foothold
Samsung Medical Center’s health-screening center — the 30-year data partner is also the first adopter.
Roadmap
2026–2028 in three stages
2026 Entry
- MediPipe official launch
- DeepBreath GMP & manufacturing license
- IEC 62366 usability evaluation
2027 Validation
- DeepBreath device approval filing & review
- Research-use product launch
- Tertiary-hospital adoption prep
2028 Expansion
- Device approval complete
- Full clinical operation
- European market entry (CE)
Going global
Building on a presentation at ERS 2026 in Barcelona, we are preparing EU CE marking in 2027 and a US FDA 510(k) submission in 2028.
Team
A team bridging clinic, AI and regulation
Kyung Hyun Lee
이경현
Co-founder & CEO PhD student in Digital Health at SAIHST, Sungkyunkwan University. An 8-year medical-AI researcher who led SaMD development and validation at AITRICS — including an MFDS-approved cardiac-arrest prediction device. 15 SCI(E) papers (first author on 8), h-index 8. A full-stack founder who personally covers everything from clinical problem discovery to AI design and regulatory entry.
Dayoon Lee
이다윤
Co-founder & Chief Quality Officer Chief Quality Officer with a nursing background. She owns medical-device quality assurance (QA), the quality management system (QMS), and regulatory processes, taking responsibility for product safety and reliability from a clinical-field perspective.
Collaboration network
Breath You has signed a formal research collaboration agreement with Konyang University Hospital and is validating both MediPipe and its pulmonary-trajectory prediction model with Konyang University Hospital data. We also collaborate with Samsung Medical Center, Hallym University Chuncheon Sacred Heart Hospital, and Ewha Womans University Mokdong Hospital.
Contact
Ready to start together?
Leave an inquiry about adoption, collaboration, or investment and we will get back to you. Detailed materials for investors and partners (IR) are shared individually upon request.
Email us directly lkh256@breathyou.care
Leave an inquiry about adoption, collaboration, or investment and we will get back to you. Detailed materials for investors and partners (IR) are shared individually upon request.