Building healthcare AI that earns trust.
Marqi Medical exists because healthcare AI should be validated before it is deployed, not after. We build predictive models with the rigor of clinical research and the practicality of enterprise software.
Our mission: Help health systems and payers identify high-risk patients before costly readmissions, using AI that has been validated, calibrated, and published in peer-reviewed literature.
The Story Behind Marqi
In 2020, founder Xavier Serrano's father survived a widow-maker heart attack. His arteries were over 90% blocked. No one saw it coming - but with today's technology, someone should have.
After 20 years in the group health insurance industry, Xavier had seen how predictive tools fail in practice: scores that do not calibrate, models that do not validate, and claims that do not survive scrutiny.
He founded Marqi Medical to build what should exist: validated AI that health systems and payers can actually trust.
Xavier Serrano
Founder & CEO
"If he survived, I would spend the rest of my life helping others avoid a similar crisis. That meant building something that actually works - with the evidence to prove it."
Our Journey
From personal tragedy to validated product.
Founding insight
Xavier's father survives widow-maker heart attack, sparking the mission
Research begins
Partnership with academic medical centers for model development
First validation
Internal validation on 50K+ patient cohort shows promising results
External validation
Independent validation confirms 0.807 AUC on AMI cohort
Peer review
Methodology paper accepted for peer-reviewed publication
Market entry
First health system deployments begin
What We Believe
The principles that guide how we build and deploy Marqi Index.
Validation Over Claims
Every performance claim should be independently verifiable. We publish our methodology, disclose our limitations, and welcome external scrutiny.
Calibration Matters
A risk score that does not calibrate is a liability, not an asset. We report calibration metrics on every validation because accuracy without calibration is dangerous.
Workflow Integration
The best model in the world is useless if care teams cannot act on it. We build for discharge planning workflows, not academic benchmarks.
Explainability Required
Clinicians need to understand why a patient is flagged as high-risk. Black box models erode trust and make intervention harder.
Peer Review is Non-Negotiable
We submit our methodology to peer-reviewed journals because external validation is the only way to build lasting trust in healthcare AI.
Continuous Monitoring
Models drift. We monitor calibration in production and retrain when performance degrades, because yesterday's accuracy is not today's guarantee.
Built by People Who Understand Healthcare
Our team brings together clinical expertise, data science rigor, and healthcare IT experience. We understand the regulatory environment, the clinical workflows, and the business pressures that shape how healthcare organizations adopt new technology.
12+
Team Members
3
Physicians
4
PhDs
Xavier Serrano
Founder & CEO
20+ years health insurance industry
Dr. Erica Oberg
Chief Medical Officer
Board-certified, clinical informatics
Advisory Board
Stanford, Former FDA, Health System CIOs
Ready to See the Evidence?
Schedule a 30-minute call to learn more about Marqi Index and how we built it.