From
Dr. Alberto Tozzi
How the Vatican's Children's Hospital Uses AI Faster Than Most DACH Region Clinics
With Dr. Alberto Tozzi
Part of
Epilepsy 4 items
Chapter 1 of 8
Hospital introduction
Introduction to Ospedale Pediatrico Bambino Gesù and its AI leadership
Expert statements on this page
No expert statements were drawn from this page.
Host summaries · secondary, not cited in answers
Educational content from recorded physician discussions — not medical advice. Talk to your (or your child's) care team about your situation.
More from this expert
Dr. Alberto TozziPodcast
Ethical and regulatory implications of artificial intelligence in healthcare
Alberto Tozzi · 21 min · Published Oct 2026
Podcast
Data requirements and infrastructure for artificial intelligence in healthcare
Alberto Tozzi · 14 min · Published Oct 2026
Podcast
Evidence-based medicine and artificial intelligence in pediatric healthcare
Alberto Tozzi · 18 min · Published Oct 2026
Podcast
Digital Twins in Healthcare: Virtual Models for Personalized Medicine
Alberto Tozzi · 16 min · Published Oct 2026
Podcast
Le funzioni e gli usi dell'intelligenza artificiale per la salute
Alberto Tozzi · 14 min · Published Oct 2026
Podcast
Healthcare innovation requires cultural change beyond technology adoption
Alberto Tozzi · 21 min · Published Oct 2026
What the experts said
One hour faster onboarding is significant in the context of a children's hospital, where every child with a severe illness, a newborn with a heart defect, or a child with a rare disease may have only minutes—one hour can be the difference between life and death.
The data cannot be used because of cloud architecture issues, data protection concerns, and the fact that patients have not consented to their data being used to save other children's lives—this is not a patient problem but a problem of current legal and technical frameworks.
The Vatican's hospital is extremely progressive, not just compared to large companies but also to research institutes in the DACH region (Germany, Austria, Switzerland).
Many physician practices in Germany, Switzerland, Austria, and Liechtenstein still do not use AI and continue to send things by fax.
Physicians cite concerns about administrative burden, documentation, forms, billing, data protection complexity, patient data sensitivity, and liability questions as reasons for not adopting AI.
The hospital has served humanity for 150 years, and with AI, it can save more lives—one hour faster treatment can save a child's life.
Ospedale Pediatrico Bambino Gesù was founded in 1869 by Duchess Isabella Salviati, who donated it to children after her own children died of scarlet fever.
In 1924, the Salviati family donated the hospital to Pope Pius XI, making it the hospital of the Pope and the Holy See.
The hospital has 120 beds, admits 20,000 patients per year, and handles 90,000 outpatient visits annually.
Three thousand researchers work at the hospital's research institute, producing 31,000 scientific papers per year.
Newsweek ranked the hospital as the number one pediatric hospital in Europe in 2026.
The hospital is the only center in Europe performing all types of organ and tissue transplants in children.
In 2020, the hospital performed 350 transplants.
In 2025, the hospital was the first worldwide to permanently implement artificial hearts in children.
Dr. Alberto Tozzi serves as chief innovation officer at the hospital.
Dr. Tozzi is president of the International Society for Pediatric Innovation, a global network of 40 children's hospitals.
Dr. Tozzi corresponds with the Pontifical Academy for Life, the institution that writes AI ethics guidelines for the Vatican.
Dr. Tozzi has an h-index of 45, placing him among the top potential leaders in the world.
AI should take over boring tasks like digital pathology so that physicians have time for important things: sitting with families, taking medical histories, and making decisions that algorithms cannot make.
When AI is used and an algorithm suggests a diagnosis, it must be based on evidence, and every algorithm must be transparent—not a black box like ChatGPT or Claude.
A physician who leads an AI project must not have fear of AI; they must understand it and lead the project in the hospital, not a company or external firm.
AI Accelerate is an EU-funded project with partners from eight countries financing it, including Helsinki University Hospital, Bambino Gesù in Rome, a children's hospital in Barcelona, and Erasmus Rotterdam.
Symptoma, an Austrian company, is a partner in AI Accelerate; it is an AI-supported symptom checker and diagnosis platform.
AI Accelerate provides a toolbox that hospitals can use to quickly onboard patients, including a hospital care pathway engine with modules for patient flow optimization, bed occupancy, resource planning, emergency admission, and triage.
Dr. Tozzi conducted a pilot study in the pediatric emergency department at Bambino Gesù, and the system reduced onboarding time from admission to treatment by one hour.
Focal cortical dysplasia is a congenital malformation in the brain that is very difficult to diagnose and causes severe epilepsy in children, often requiring surgery and sometimes resulting in lifelong disability.
There are radiological markers for focal cortical dysplasia, such as a specific angle in a vertebra, that a neurologist must recognize; specialized expertise is required to identify these markers.
This specialized expertise for diagnosing focal cortical dysplasia does not exist in rural hospitals, most city hospitals, or even many normal city clinics.
An algorithm can analyze MRI images in milliseconds, and if the algorithm is validated and transparent, a physician in Tanzania or a rural hospital in Bregenz can use AI to democratize the best diagnosis in the world to places that do not have the expertise.
There are only five centers worldwide that can diagnose focal cortical dysplasia.
The Pediatric Moonshot vision, inspired by Timothy Chu (a former Oracle executive and Stanford professor), asks: What if we could connect four million medical records from 500 different children's hospitals worldwide in a data-protection-compliant and naturally AI-enabled way?
Pediatric hospitals have six exabytes of data—ultrasound data, imaging data, genetic data—but this data sits in silos on local servers behind firewalls.
Six exabytes is 120,000 times larger than the entire imaging database of the US National Institutes of Health.
Federated learning works differently from centralized AI: instead of sending data to the cloud, the algorithm is sent to the local device (like Siri on an iPhone), and the algorithm runs on the local device.
In federated learning, instead of sending data to the cloud where there are data protection concerns, the algorithm is sent to the hospital, where it learns from the data (e.g., ultrasound images, patient records) without the data leaving the hospital.
Only the weights—the mathematical parameters, the abstract weightings that the model learned—are sent back in encrypted form; this is the learning that happens, similar to what Apple does with Siri.
Project Gemini, named like Google Gemini, involves 100 edge computing servers with 100 ultrasound devices across six hospitals on two continents, connected for pediatric cardiology applications such as ejection fraction measurement and diagnosis of congenital heart disease.
Bambino Gesù is implementing edge infrastructure in Europe for Project Gemini, not in Zurich, Vienna, or Frankfurt, but at the Vatican's hospital under Dr. Tozzi's leadership.
The Pontifical Academy for Life works with Microsoft and IBM on partnerships with the hospital, for example, AI against childhood cancer or AI in cardiology.
The Vatican has established ethical principles for AI.
In 2020, the Vatican established an AI ethics committee that brought together eleven religions—Muslims, Anglicans, Buddhists, Hindus, and others—in Hiroshima (the site of the atomic bomb) to develop AI ethics principles.
The Vatican's hospital in Barcelona (likely referring to a partner hospital) and in Italy has addressed these concerns through transparency, ethics, federated learning, and asking not whether but who will lead AI adoption.