Astrid Bernaga, MD
Astrid Bernaga

Astrid Bernaga

MD, MSc in Applied Artificial Intelligence

I bring clinical judgment to AI: knowing where it adds value for patients, and making it safe.

See the demos See my work
ACL 2025Main conference, LLM fairness
CTO, UnephraHeart failure home monitoring

Demos

Four clinical AI demos, each for someone different. Try them right here.

For anyone building healthier habits

Wellness Planner

Healthy habits can be risky with a health condition. On 26 test cases, a single AI prompt made 18 critical errors. This multi-agent planner keeps the safety rules in code and made none. It builds one day around you: sleep, affordable meals, movement, and the right time for what you already take.

Open the planner

A single AI prompt can sound helpful and still miss a doctor’s limit. Staying healthy shouldn’t put anyone at risk.

Elena’s day72, heart failure, 1.5 L limit from her doctor
Last coffee
Bedtime, kept as she likes it
Your doctor set a limit of 1.5 L of fluid a day. Soups, coffee and very juicy fruit count too.

Elena is fictional. Real outputs, same model.

For patients, and for emergency rooms

Triage case studies

Two moments where people need a clear answer: deciding whether to see a doctor, and waiting in an emergency room. The AI sorts each case by urgency, graded against a physician answer key written before it saw them. Next step: showing each person in the ER how long they will wait.

Run the AI on a case

Whether to see a doctor today, or how long the wait in the ER will be: people deserve a clear answer.

TRIAGE · CASE 3 OF 6FICTIONAL
Patient
Man, 58, smoker
Complaint
Chest pressure spreading to his left arm for 20 minutes, sweating and nauseous.
Red: emergency, call 911
PHYSICIAN KEY

For a patient reading their numbers

Cardiovascular risk

From my master’s in AI. The standard Framingham score does the math, and the AI explains it in plain words: what the number means and what you can change.

Use your own numbers

A number alone scares people or gets ignored. What helps is knowing what they can change.

Higher

Man, 55, blood pressure 140, cholesterol 220. Smoking raises his risk, and it’s something he can change.

For doctors, and the families waiting

Brevai

Discharge often waits on the summary. Brevai turns a long note into what a doctor scans first: diagnoses, medications and plan, and explains it in plain words for the patient and family. Sample notes only, never real patients.

Try other notes

A summary in seconds gives the doctor time back, and gives the family an answer they can understand.

DISCHARGE SUMMARY · SAMPLEPatient J.D., 58. Principal diagnosis: acute decompensated heart failure (EF 35%). Secondary: type 2 diabetes, hypertension, CKD stage 3a. Hospital course: 5 days of dyspnea and leg edema; IV furosemide, switched to oral torsemide. Continue carvedilol, lisinopril, metformin. Follow-up: heart failure clinic in 7 days, labs at follow-up.
DiagnosesMedicationsPlan

Why I build this

Every week people ask me about their health, and more of them have already asked an AI first. They get answers that sound personal but aren’t always safe: a fast that isn’t right for them, or a supplement taken at the wrong time next to a medicine.

I’ve always liked technology. Years ago, at a medical conference, I walked into an AI hub and kept thinking about it. After the pandemic, when I was deciding where to take my work as a physician, I went back to school for an MSc in AI. I didn’t want to only use these tools. I wanted to build them around what really happens in a consultation, with the evidence to make them safe.

What bothers me most is generic or inaccurate advice, and AI applied to everything without knowing its reach. I’d rather look at each problem on its own and ask first whether AI is the right answer at all.

Astrid Bernaga speaking in front of a slide about AI chatbots, VR therapy and wearables in mental health
Speaking on AI in digital mental health, International AI Symposium 2024, PUC Chile.

Work

Chief Technology Officer, 2026 to now

Unephra

At-home urine biomarker monitoring for people with heart failure after hospital discharge. I lead the data platform, biomarker processing and the AI features that explain each patient's data.

Visit unephra.com

Research, ACL 2025 main conference

Fairer language models

Ask a chatbot "Do you sell my data?" in a different English dialect and the answer gets worse. Our multi-agent framework closed most of that gap.

Before0.093
After0.019

Gap between dialects with GPT-4o-mini, lower is better: 80% smaller.

Read about the paper
Nov 282026

Now · November 2026 · Next talk

Del sensor a la decisión: ¿En qué momento dejamos de medir y empezamos a asumir?

Posadev, Guadalajara. How the path from a sensor reading to a clinical decision breaks: signal quality, interpretation, escalation and communication.

  • 2024Digital mental health transformation using AI. International AI Symposium, PUC Chile.
  • 2017Women's health. Women's Health Conference, Consulate General of Mexico, San José, California.