Six risks, not one
Deterioration, sepsis, delayed cerebral ischemia, lung injury, kidney injury and delirium — side by side. One score tells you a patient is sick. Six tell you how.
ICU IntelligenceEvery risk
Every risk
on one screen.
With the reason.
Intensive care already measures everything, every second. Medikan reads that stream continuously and turns it into predictions a clinician can act on — and shows what drove each one.
Research prototype · not a medical device · not for clinical use
Intensive Care Unit · 15 bedslive
Bed 01F · 71
UrgentHR
128
BP
82/54
SpO₂
86
Bed 09M · 54
StableHR
74
BP
118/71
SpO₂
99
Deterior.
0.91
DCI
0.87
Sepsis
0.62
ALI
0.41
AKI
0.28
Delirium
0.19
Medikan ICU
Live risk cockpit with explainable scores
What it does
The signal is already there. It just never reaches anyone in time.
On a night shift, two clinicians look after fifteen patients. Every one of them generates thousands of measurements an hour, spread across four separate systems. Medikan brings it together and points at the bed that needs attention first.
Multimodal and live
Waveforms, brain monitoring, ventilator settings, labs, fluids and drugs — read continuously, not sampled every fifteen minutes.
Never a number alone
Every score opens into the values behind it. A number gets ignored. A number with a reason gets acted on.
Explainability
A score you can argue with.
Closed models get switched off. Ours opens: hover any prediction and it shows the five values driving it right now, which ones raise the risk, and which ones are holding it down.
Signed contributions. Risk factors point right, protective factors point left.
Live, not retrospective. Drivers recompute as the patient changes.
Named inputs. Every model declares which data it used.
Clinical language. CPP in mmHg, lactate in mmol/L — not feature indices.
Model inputs
Continuous EEGalpha/delta ratio, burst suppression
Arterial pressurebeat to beat, CPP derived
Intracranial pressureexternal drain
Transcranial Dopplervasospasm surveillance
Labs and medicationblood gas, electrolytes, nimodipine
Every prediction names the data behind it. Nothing arrives as a number without a source.
One patient, one screen
Everything you need to act on the score.
A prediction is only useful next to the context that explains it. Open a bed and the whole patient is there — no second system, no second login.
Live ECGArterial blood pressureContinuous EEGIntracranial pressureSpO₂ & respiratory rateBlood gasMedications givenContinuous infusionsFluid balanceVentilator settingsLines and tubesNursing checksAdmission sourceSOFA · GCS · RASS · NEWS2
Sits on top, not instead
The hospital keeps its record system. Medikan reads from it and from the bedside devices, and gives back the one view neither of them provides.
Stays in the hospital
Designed to run inside hospital infrastructure. Patient data does not need to leave the building to be useful.
Built with clinicians
Designed at the bedside with the people who work the night shift, not adapted from a general-purpose dashboard.
Team
Clinicians and engineers, in the same room.
We build where the data is made — with the people who use it.
Jenestin
Clinical AI & product
PhD researcher in multimodal AI for clinical prediction. Builds the models and the interface they live in.
Team member two
Intensive care medicine
Bedside clinician. Makes sure every number on the screen answers a question a doctor actually asks on the round.
Team member three
Data & signal processing
Turns continuous waveform data into features a model can learn from, without losing what the signal means.
Team member four
Engineering
Builds the platform that carries live data from the bedside to the cockpit, safely and inside the hospital.
Get in touch
See it running on a live unit.
We work with intensive care teams and researchers. For a walkthrough or a conversation about a pilot, write to us.