AI for Paperwork, Docs for Decisions – Interview with Hilaryano Ferreira on the Future of European Healthcare
- Jul 12
- 7 min read
Hilaryano Ferreira is a cardiologist and the founder of AETHERA Health, an EU-licensed telehealth platform delivering physician-led online care across Europe. In 2021, years before AI became a mainstream healthcare topic, he co-authored peer-reviewed research on using artificial intelligence to automate the remote follow-up of cardiac devices, published in EP Europace. In this interview, he explains where AI genuinely helps medicine, where it must never decide, and what a borderless Europe of healthcare could look like.
Hilaryano Ferreira, Cardiologist and Founder of AETHERA Health
What inspired you to turn your experience as a cardiologist into a platform like AETHERA Health?
Frustration, honestly, of a very specific kind. In cardiology, I watched superb clinicians spend their evenings on documentation instead of patients. The research literature confirmed what we all felt, for every hour with a patient, physicians spend nearly two more on records and desk work.
Medicine did not lack intelligence or dedication. It lacked time, and the time was being consumed by administration.
In 2021, I co-authored research showing that software could take over one of those administrative burdens, the routine reading of cardiac device reports, without touching a single clinical decision. That experiment changed how I saw my own profession. If automation could carry the paperwork in one narrow corner of cardiology, it could carry it across an entire care journey.
AETHERA Health is that idea taken seriously. We built a telehealth platform where technology absorbs the repetitive work around the consultation, so that when a patient meets their physician, the physician is actually present. The inspiration was never technology for its own sake. It was giving the consultation its time back.
Your research explored using AI to automate the follow-up of cardiac devices years before AI became a mainstream healthcare topic. What did that experience teach you about where AI creates the most value?
It taught me that AI creates the most value where the work is essential but does not require judgment. Patients with pacemakers and similar devices generate a constant stream of follow-up reports. Reviewing them matters enormously, yet for the routine majority, it is repetitive reading, not clinical reasoning. Our team combined optical character recognition, AI, and natural language processing to read those reports automatically, and in testing, the system processed thirty reports in under five minutes.
The revealing part was what the software did not do. It never made a diagnosis and never decided on a treatment. It extracted, organised and presented, and clinicians did the rest. That division held up under real cardiac data in a real hospital, and it convinced me that the fashionable question, can AI think like a doctor, is the wrong one. The productive question is how much non-clinical weight AI can lift so that doctors can think like doctors. Everything we have built since rests on that lesson.
You often say AI should support physicians rather than replace them. Why is that distinction so important to you?
Because the two paths lead to completely different systems, and only one of them is safe to scale. If you design AI to replace physicians, every error the model makes lands directly on a patient, and accountability dissolves into a supply chain of vendors. If you design AI to support physicians, errors are caught where they have always been caught, by a trained human whose name is on the decision.
There is also a quieter reason. Medicine is not only pattern recognition. It is responsibility, consent, context, and the judgment to know when the textbook answer is wrong for this particular person. Those things do not compress into a model, and pretending otherwise erodes the trust that makes care possible.
At AETHERA, the boundary is written into the product. Artificial intelligence structures each patient’s assessment, screens for contraindications and prepares a clinical summary. A licensed physician reviews and approves every case, every prescription, every plan. The machine organises. The clinician judges. I consider that distinction the foundation of responsible digital health, not a marketing line.
What do you think the healthcare industry gets wrong when it talks about AI transforming medicine?
It talks about replacement and prediction, and it forgets subtraction. Most conference talks concern AI diagnosing better than specialists or predicting disease years ahead. Those are worthy research directions, but they concern the hardest five percent of medicine, and they are precisely where AI’s failures are most costly and least forgivable.
Meanwhile, the industry steps over an enormous, boring opportunity, the administrative surplus. Documentation, referral letters, report reading, scheduling, coding, follow-up logistics. This is where clinician time actually disappears, where burnout actually starts, and where automation is both mature and safe, because a mistake in a draft summary gets corrected by the person reviewing it rather than harming a patient.
The second mistake is treating AI as a strategy in itself. AI is an amplifier. Applied to a broken process, it produces broken outcomes faster. The organisations getting real value start from a redesigned workflow and then automate the parts that never needed a human. Transformation, in my experience, looks less like a robot doctor and more like a physician who finally leaves work on time.
How can telehealth expand access to specialist care without compromising quality or patient trust?
By being honest about what transfers well through a screen and disciplined about what does not. A large share of specialist work is conversation, history, interpretation of results, and adjustment of treatment. That transfers beautifully, and for a patient in a small town or an expat who does not speak the local language, telehealth can mean the difference between seeing a specialist and seeing nobody.
Quality survives when the clinical standard is identical to in-person care. At AETHERA, every assessment is reviewed by an EU-licensed physician, prescriptions are issued only where clinically appropriate, and anything requiring physical examination or urgent care is directed to the right setting rather than forced through a video call. Declining to treat remotely, when that is the right answer, is part of the product.
Trust follows transparency. Patients should know who their physician is, what the technology does with their data, and what it is allowed to decide, which in our case is nothing. Telehealth fails when it behaves like an app that dispenses medicine. It succeeds when it behaves like medicine that happens to arrive through an app.
What practical changes could hospitals make today to reduce administrative burden and give doctors more time with patients?
Start by measuring where clinician hours actually go, because most institutions genuinely do not know. Once you see the numbers, three changes pay for themselves quickly.
First, automate documentation drafting. Ambient tools that produce a draft note for the clinician to review and sign are mature technology, and they save time in every single consultation.
Second, take report triage seriously. Our 2021 experiment showed routine device reports could be read by software in minutes, and the same logic applies to many high-volume, low-variance report streams. Let machines flag the abnormal minority and let clinicians concentrate on them.
Third, remove the small frictions that compound, single sign-on instead of five systems, referral templates instead of free-text letters, scheduling that does not require a physician’s involvement.
None of this requires a moonshot budget. It requires leadership that treats clinician time as the scarcest resource in the building, and a rule I would offer any hospital board, before buying anything that promises intelligence, buy back an hour of your doctors’ attention.
When developing AETHERA Health, how did you decide which parts of the patient journey should be automated and which should always remain in the hands of a physician?
We used a test that has never failed us, does this step require clinical judgment, or does it require diligence? Judgment stays human, permanently. Diligence is a candidate for automation.
Intake is diligence. A structured assessment can gather history, medications, and goals more completely than a rushed conversation, so AI runs it and prepares a clinical summary. Screening for red flags is diligence with a safety twist. The system checks for contraindications and escalates, but it can only ever say no or ask a physician, never yes. Diagnosis, prescribing, and treatment planning are judgments, so they belong to a licensed physician in every single case, with no exceptions and no volume-based shortcuts. Follow-up logistics, reminders, and monitoring of routine signals are diligence again, with clinicians alerted the moment anything deviates.
The honest answer is that our 2021 research made this decision for us before the company existed. We had already tested where the boundary holds. Building AETHERA was a matter of refusing to move it when growth made moving it tempting.
What principle has guided your decisions as both a cardiologist and a healthcare entrepreneur?
Primum non nocere, first do no harm, applies to systems as well as patients. In clinical practice, the principle is familiar. As an entrepreneur, I found it just as demanding, because product decisions can harm quietly and at scale in ways an individual consultation never could.
In practice, it means a bias towards being conservative exactly where it costs us. We are ambitious about automation and conservative about clinical authority. We would rather lose a sale than issue a prescription that a physician has not genuinely reviewed. We treat regulatory compliance, GDPR, and licensing not as friction but as the codified experience of people who saw what happens without them.
The same principle sets the pace. Healthcare rewards patience, it punishes the move-fast-and-break-things culture, because here the things that break are people. So every feature we ship answers one question first, if this fails, who is protected? If the answer is the company rather than the patient, it does not ship. That rule has cost us speed occasionally. It has never cost us sleep.
If you could change one thing about the future of healthcare in Europe, what would you want it to be?
I would make European healthcare as free-moving as Europeans are. The single market moves goods, capital, and people across borders effortlessly, yet a patient who relocates from Lisbon to Berlin often starts from zero. New records, new referrals, repeated tests, months of waiting to re-establish care that already existed.
Concretely, I would want genuinely portable health records and the mutual recognition that lets an EU-licensed physician care for an EU resident across internal borders with clear, uniform rules.
Telehealth has made the technical part almost trivial. The remaining barriers are administrative, and administration is exactly the layer we now know how to automate responsibly.
The prize is enormous. Europe has world-class clinicians unevenly distributed and millions of mobile citizens, from expats to cross-border workers to Americans building lives here, who fall between systems. Connect the two, and you raise the standard of care for the whole continent without building a single new hospital. That is the Europe of health I want AETHERA to help construct, one where your physician travels with you, wherever you decide to live.
Read more from Hilaryano Ferreira
