BetaNew · MedGemma 2.4

Clinical decisions,
faster and more accurate
— in seconds.

MedIntel is an AI copilot for emergency rooms and wards. Triage, differential diagnosis, suggested workups, and literature citations — all in the physician's flow.

LGPD-readyOn-premise availableFHIR integration
app.medintel.health/painel
Shift
Dashboard
Patients 42
Alerts 3
Team
On-call

Clinical dashboard

Night shift · 22:14 · 42 active
Critical
3
+1 12min
Waiting
14
+3
Triage
3.8m
−68%
Decision
15m
−61%
Patient
Risk
Hypothesis
SpO₂
M. F. S., 58
PAC-2741
Critical
STEMI · 0.91
A. R. C., 71
PAC-2740
High
Abd sepsis · 0.78
J. L. M., 34
PAC-2739
Stable
Tension HA · 0.62
P. S. O., 28
PAC-2738
Stable
Mech back pain · 0.55
Insights
M. F. S. · Consider serial ECG
Troponin 1.2 → 1.8 in 90min. Pattern suggests evolving MI.
91%
A. R. C. · qSOFA = 2
Start sepsis bundle: Abx + lactate + cultures within 1h.
84%
+3 pending suggestions
In use and evaluation at partner hospitals and institutions
Representative logos. Full list under NDA.
How it works

From first symptom to treatment plan.

Four steps integrated into your flow. No screen-switching, no rewriting notes.

1
Smart triage
Vitals, chief complaint and epidemiological triggers processed in real time. Risk and priority computed in <2s.
2
Contextual history
The model cross-references conditions, allergies, ongoing meds and recent labs — auto-extracted from the EHR.
3
Bayesian differential
Ranked list with post-test probability, discriminating questions, and the highest-yield labs.
4
Evidence-backed plan
Suggested management, dosing, contraindications and UpToDate / guideline citations — straight into the note.
Features

Built for the shift tempo.

Every interaction designed with ER physicians. Low latency, keyboard-first, clinical copy.

Risk computed before the patient enters the room.
Direct integration with bedside monitors. Abnormal vitals fire team alerts and (optionally) push to the responsible physician's phone.
SpO₂, HR, BP, RR, Temp, Pain
Pediatric Z-score by age
Sepsis detection (SOFA/qSOFA)
Auto re-evaluation every 15min
// Real-time triage
POST /v1/inference
model: "medgemma-2.4-clinical",
latency_p95: 1.7s,
audit_id: "sha256:e7f9...c4"
Diagnosis recommendation

Ranked hypotheses, with reasoning you can audit.

With each new data point, MedIntel updates the post-test probability and shows exactly why a hypothesis rose — with suggested workups, management, and citations.

Patients / Helena Dias Vilela L. / New consultation
TriageHistoryClinical DataQuestionsDiagnosis
Differential Diagnosis
6 hypotheses · AI-generated · pre-test → post-test
#1
Influenza A Pneumonia
J10.0critical
90%
Patient presents with fever, productive cough, and signs of respiratory effort, corroborated by a positive Influenza A test.
Probability90%
See details
#2
Viral Respiratory Infection
J06.9
60%
Fever, productive cough, and respiratory effort may be caused by other viral respiratory infections.
Probability60%
See details ›
#3
Bronchiolitis
J21.9caution
40%
The patient is an infant, and bronchiolitis is a common cause of cough and respiratory effort in infants.
Probability40%
See details ›
#4
Sinusitis
J32.9caution
30%
Productive cough can be a symptom of sinusitis, especially if there is nasal congestion.
Probability30%
See details ›
#5
Viral Pneumonia
J18.9
20%
Fever, cough, and respiratory effort. Absence of COVID-19 and normal CRP reduce the probability.
Probability20%
See details ›
#6
Bacterial Pneumonia
J15.9
10%
Absence of leukocytosis and presence of Influenza A make bacterial pneumonia less likely.
Probability10%
See details ›
Before vs After

What changes on shift with MedIntel.

Retrospective pilot data · n=4,218 visits · 4 hospitals · 6 months.

Without MedIntel
Manual triage, free-text notes, UpToDate in another tab.
Avg triage time: 14 min
Differentials documented: 1.8 per case
Redundant labs ordered: 23%
Time to therapeutic decision: 38 min
30-day readmission: 12.4%
38min
to decision
With MedIntel
Auto-risk, ranked differential, inline evidence, pre-drafted note.
Avg triage time: 4 min
Differentials documented: 5.2 per case
Redundant labs ordered: 9%
Time to therapeutic decision: 16 min
30-day readmission: 8.1%
16min
to decision
Specialties

Trained on real clinical scenarios, from trauma to pediatrics.

Specialty models per area, with local prevalence and protocols.

ER Cardiology
ACS, chest pain, arrhythmias, decomp HF
Pulmonology
PE, asthma, COPD exacerbation, pneumonia
Neurology
Stroke, headache, seizure, NIHSS
Sepsis / Infectious
SOFA, qSOFA, empirical Abx, shock
Pediatrics
Z-scores, weight-based dosing, FWS
Trauma & Ortho
ATLS, CT triage, fractures
Internal Medicine
Multimorbidity, polypharmacy, frailty
Imaging & Labs
CT findings, POCUS, blood gas
Clinical results

Numbers that matter on shift.

Aggregate data from 2025/26 institutional pilots.

0%
Reduction in time to therapeutic decision in the ER
median, n=4,218
0.0×
More differentials documented per case
vs standard notes
0%
Fewer redundant labs ordered
retrospective audit
0.0/5
Average NPS among on-call physicians
survey · n=312
Security & compliance

Built for sensitive clinical data.

MedIntel was designed from day one to operate within Brazilian and international data requirements. Per-inference audit, granular consent, and on-premise deployment for hospitals where data cannot leave the building.

Pseudonymization by defaultNames and identifiers are stripped before inference. Re-identification only with the hospital's key.
Immutable logs (WORM)Every prompt, response and citation is logged with SHA-256 hash — auditable for 5 years.
Models never train on customer chartsContracts guarantee patient data does not leave the tenant or feed training.
On-premise via DockerFor hospitals with strict data residency. Same stack, no external API.
LGPD
Brazilian data-protection law · dedicated DPO
Compliant
HIPAA
Health Insurance Portability and Accountability Act
3rd-party attested
ANVISA SaMD
Software as a medical device · Class II
In registration
ISO 27001
Information security management
2026 audit
Who uses it

Physicians who trust it on shift.

In sepsis, every minute matters. MedIntel flags qSOFA before I even look at the chart. It already changed our protocol.
MA
Dr. Marina Albuquerque
ER Lead · HC-FMUSP
It's not just a chatbot. It's a senior resident that never tires, with all the literature in their head — and cites the source.
RT
Dr. Ricardo Tavares
On-call Cardiologist · Sírio-Libanês
I cut my note-writing time in half. The Bayesian differential forces me to think about hypotheses I'd otherwise miss.
CY
Dr. Camila Yano
Internal Medicine · Albert Einstein
Weight-based dosing errors in pediatric ICU can be fatal. MedIntel auto-calculates by weight and flags interactions instantly. That peace of mind is invaluable.
FR
Dr. Felipe Rodrigues
Pediatric Intensivist · Hospital das Clínicas de Recife
I work at a regional referral center. Cases I'd spend hours researching before — MedIntel surfaces the evidence in seconds, with the citation.
JF
Dr. Juliana Ferreira
Emergency Physician · Regional Hospital of Uberlândia
The ranked differential with post-test probabilities is exactly the reasoning we teach in residency — just in real time, without opening a textbook.
AC
Dr. André Castelo
Internal Medicine Preceptor · UNIFESP
I see a lot of elderly patients on polypharmacy. The drug interaction check caught two serious errors in my very first month using it.
PM
Dr. Patrícia Moura
Geriatrician · Hospital Santa Marcelina
My team was skeptical at first. After the first shift with MedIntel, nobody wanted to go back to the old way.
TL
Dr. Thiago Lessa
Emergency Chief · UPA-24h Goiânia
In sepsis, every minute matters. MedIntel flags qSOFA before I even look at the chart. It already changed our protocol.
MA
Dr. Marina Albuquerque
ER Lead · HC-FMUSP
It's not just a chatbot. It's a senior resident that never tires, with all the literature in their head — and cites the source.
RT
Dr. Ricardo Tavares
On-call Cardiologist · Sírio-Libanês
I cut my note-writing time in half. The Bayesian differential forces me to think about hypotheses I'd otherwise miss.
CY
Dr. Camila Yano
Internal Medicine · Albert Einstein
Weight-based dosing errors in pediatric ICU can be fatal. MedIntel auto-calculates by weight and flags interactions instantly. That peace of mind is invaluable.
FR
Dr. Felipe Rodrigues
Pediatric Intensivist · Hospital das Clínicas de Recife
I work at a regional referral center. Cases I'd spend hours researching before — MedIntel surfaces the evidence in seconds, with the citation.
JF
Dr. Juliana Ferreira
Emergency Physician · Regional Hospital of Uberlândia
The ranked differential with post-test probabilities is exactly the reasoning we teach in residency — just in real time, without opening a textbook.
AC
Dr. André Castelo
Internal Medicine Preceptor · UNIFESP
I see a lot of elderly patients on polypharmacy. The drug interaction check caught two serious errors in my very first month using it.
PM
Dr. Patrícia Moura
Geriatrician · Hospital Santa Marcelina
My team was skeptical at first. After the first shift with MedIntel, nobody wanted to go back to the old way.
TL
Dr. Thiago Lessa
Emergency Chief · UPA-24h Goiânia
Plans

Per physician, per month. No surprises.

Free 60-day institutional pilots. Cancel anytime.

On-call
R$189per physician/mo
For individuals and small clinics.
Triage + differential
Cited evidence
500 queries/mo
Chat support
Most popular
Hospital
R$149per physician/mo
For ERs, ICUs and wards.
Everything in On-call
FHIR/HL7 integration
Multi-team dashboard
Unlimited queries
Institutional SSO
Dedicated 24/7 support
Enterprise
Custompricing
Hospital networks and payers.
Everything in Hospital
On-premise / air-gapped
Custom models
Dedicated EHR connector
Signed BAA / DPA
Dedicated CSM

Ready to put AI on shift?

30-minute demo. Institutional pilot in 2 weeks. No credit card.