Instap

Patient Viewer

Clinical monitoring data — live, stored, searchable

IPV is not a new source of data. It keeps the stream your bedside monitors already produce, at full resolution and for as long as you decide, and makes it reviewable from any authorised computer, tablet or phone.

  • Live vitals and waveforms in near real time, anywhere in the hospital
  • Full waveform history, notification log and beat-level replay — kept, not overwritten
  • On-premise, integrated with the CARESCAPE Gateway, live at WIM Warsaw since December 2025

The asset you already own

Your monitors produce the data. Almost none of it survives the shift.

A hospital that has bought continuous monitoring has already bought the most detailed data stream it will ever generate. Today that stream scrolls off the screen and is gone.

≈480,000
measurements every second
Twenty waveform channels per bed at 240 Hz, plus some 180 numeric parameters a minute, across a hundred monitored beds.
Hours
is all the monitor keeps
The bedside monitor holds a short rolling window. Once the curve scrolls past it is not archived or summarised — it is overwritten.
0.1 s
is what a full day of charting captures
Even at twenty parameters written down every hour for every bed, a day of charting is 48,000 values — what the monitors produce in about a tenth of a second.
€0
extra to produce the signal
The monitors, the interface gateway and the network are installed and budgeted. The data costs nothing more to generate — only to keep.

Figures illustrative, for a 100-bed critical-care estate.

The cost of letting it go

Four things your hospital cannot do today — all for the same reason

  1. 01

    The clinician cannot look back

    When a patient deteriorates, the minutes that explain why have already scrolled off the monitor. The review is done from memory and a nursing note.

  2. 02

    Quality review has nothing to review

    An adverse event cannot be reconstructed. There is no objective, timestamped record of what the physiology and the alarms actually did.

  3. 03

    Research has no dataset

    High-resolution physiological data is what modern clinical and AI research runs on. The hospital produces it every day and archives none of it.

  4. 04

    Analytics has no foundation

    Every predictive model needs history to learn from and a live stream to run on. Without stored monitoring data there is nothing to build on.

A night on the ward

03:12. The patient destabilises.

The same night, the same team — with and without a record of what happened.

Today

What the hospital can actually see

  1. 03:12The alarm soundsThe monitor alarms at the bedside. The nurse responds, silences it and starts working. Nothing about the episode is recorded beyond that response.
  2. 03:20The consultant is calledOn the phone they have numbers read aloud. They cannot see the waveform, the trend, or the sequence of alarms that led up to the call.
  3. 03:55The consultant arrivesBy the time anyone senior reaches the monitor the patient is stabilised and the screen shows the present. The decisive forty minutes are gone.
  4. 07:30The morning handoverThe episode is reconstructed from memory and two entries in the chart. Nobody in the room can look at what actually happened.

With IPV

With the record still there

  1. 03:12The alarm is capturedThe monitor alarms exactly as before. IPV stores the event, its priority, the triggering value and the waveform around it, and keeps them.
  2. 03:20The consultant looksFrom home, on a phone, they open the patient, scroll back through the last hour of curves and read the notification sequence before advising.
  3. 03:55Arrival with contextWhoever arrives has already reviewed the episode. The conversation begins from what happened, not from an attempt to reconstruct it.
  4. 07:30Handover on evidenceThe episode is replayed on screen, measured where the timings matter, and annotated so the record survives the shift change.

How it works

From the bedside to the clinician’s pocket

One web and mobile application for clinical staff, fed from the monitors the hospital already owns, over the interfaces those monitors already publish.

At the bedside
ECG · 12 leads
240 Hz continuous
SpO₂ · pleth
saturation + waveform
ART · invasive
sys / dia / mean
Temp, PI, ADT
numerics, once a minute
36.8 °C
PI 2.4
GE CARESCAPE Gateway · HL7 + HSDI
Instap Patient Viewer
Ingest, store, replay

Numeric values and full-rate waveforms written to one time-series store, on the hospital’s own servers.

Raw
≥ 7 days
Archive
≥ 1 year
Wherever the clinician is
Instap Patient ViewerLive
R. Majewska
ICU · bed 302B · 61 y · 74 kg · 165 cm
HR118
SpO₂89
ART96/54
T137.9
ECG II
ECG V1
ART
SpO₂
RESP
Notificationslast 60 min
  • 03:12:41SpO₂ low
  • 03:12:58Tachycardia
  • 03:14:06ART mean low
  • 03:19:22SpO₂ low
  • 03:41:10Lead off — V4
ECG waveform sampling
240 Hz
Configurable trend write interval
≤ 1 min
Raw data retention
≥ 7 days
Archive retention
≥ 1 year

In production today

IPV is live at WIM, Warsaw

Not a concept: a live reference site in Poland, in daily clinical use on GE CARESCAPE monitors — and you are welcome to visit it.

In production since December 2025

Deployed at the Military Institute of Medicine — National Research Institute (WIM) in Warsaw, and developed further with the clinical team ever since.

100+ monitored beds across 7 departments

Intensive care, cardiology, cardiac surgery and emergency — one installation covering the hospital’s critical-care estate.

Running on GE CARESCAPE

Fed from CARESCAPE monitors through the CARESCAPE Gateway — numeric values over HL7, full-resolution waveforms over HSDI.

The application

Clinical workflow, from ward overview to single-beat analysis

Every module was designed alongside the clinical teams using it — from the nurses’ station to the intensive care bed.

The ward

Every bed on one screen, from anywhere in the hospital

The view a nurses’ station or a duty room keeps open all shift — and the same view on a laptop at home.

  • Patients grouped by clinical unit — ICU, cardiology, cardiac surgery, emergency.
  • Up to 16 patients at once, with live values and moving waveforms for all of them.
  • Resizable tiles: expand the patient who matters right now.
  • Filter by ward, keep a personal “My Patients” list, reorder curves by clinical priority.
ipv.hospital.local/ipv/unit/icu
Intensive care — live
16 monitored beds · 4 shown
Live

R. Majewska

61 y · 74 kg · 165 cm
HR118bpm
SpO₂89%
ART96/54mmHg
T137.9°C
+10 more waveforms · full view

J. Wiśniewski

48 y · 97 kg · 187 cm
HR92bpm
SpO₂96%
ART124/71mmHg
T136.8°C
+10 more waveforms · full view

K. Brzeziński

67 y · 81 kg · 176 cm
HR74bpm
SpO₂97%
ART131/78mmHg
T136.4°C
+10 more waveforms · full view

A. Lewandowska

55 y · 68 kg · 170 cm
HR81bpm
SpO₂95%
ART118/66mmHg
T136.9°C
+10 more waveforms · full view

Live values and waveforms in near real time, on any authorised computer, tablet or phone — with no client software to install.

The patient

Down to a single beat, hours after it happened

Stored waveforms replayed and measured, with the numbers and the trend behind them on the same timeline.

  • Scroll and replay stored waveforms; pan, speed and gain with gestures.
  • Caliper interval tools for P-Q-R-S-T measurement, saved with a comment.
  • Live vitals and trends: HR, SpO₂, ART (systolic/diastolic/mean), dual temperature, perfusion index.
  • Trend interval configurable down to one minute; archival sampling rate set per curve and per patient.
ipv.hospital.local/ipv/patient/302b/archive/waveforms
R. Majewska · ICU · bed 302B
61 y · 74 kg · 165 cm · replay 03:12:38 – 03:12:43
Archive
HR118bpm11050
SpO₂89%10092
ART96/54mmHg16090
T137.9°C37.536
ECG II
ECG V1
ART
SpO₂
Heart rate · last 24 hpeak 132 bpm · 03:13
06:0012:0018:0000:00
Notification loglast 60 min
  • 03:12:41SpO₂ low86 %
  • 03:12:58Tachycardia132 bpm
  • 03:14:06ART mean low58 mmHg
  • 03:19:22SpO₂ low89 %
  • 03:41:10Lead off — V4—
Selecting a notification scrolls the waveforms to the event — a ±15 s window around the triggering value.

The record that does not exist unless something kept it: the leads around an event, the trend that led to it, and the notifications that fired.

The notifications

Notifications as episodes, not as a night of beeping

Every notification stored with its timestamp, priority and triggering value, grouped into the events a clinician would actually describe.

  • Filterable log; selecting a notification scrolls the waveforms to a ±15 s window around it.
  • Consecutive readings out of range are grouped into one episode, with duration and extreme value.
  • Status radar: each parameter sits on one of four rings — in range, below, above, or no data.
  • Mark as seen and comment. Mobile notifications are informational only: no guarantee of delivery, and no replacement for the bedside alarm.
ipv.hospital.local/ipv/patient/302b/notifications
R. Majewska · notifications
last 24 h · 5 notification episodes · grouped from 295 readings out of range
24 h
  • Heart rate03:12 – 03:41
    29 min · 88 readings132 bpm
  • SpO₂03:12 – 03:26
    14 min · 42 readings84 %
  • Respiratory rate03:15 – 03:33
    18 min · 54 readings31 /min
  • ART mean03:14 – 03:22
    8 min · 21 readings52 mmHg
  • Temperature05:40 – 07:10
    90 min · 90 readings38.4 °C
Status radarnow
HRSpO₂ART sysART meanTempRRPIEtCO₂
  • In range — the reference ring
  • Outside limits — pulled in or pushed out
  • No data on that parameter

A patient who is where they should be draws a circle on the radar. Anything else is a dent or a spike you can read across a room.

Access

Who can see what, answerable in one sentence

Permissions scoped per role, per department and per patient, through the hospital’s own directory.

  • Active Directory sign-in and smart-card authentication, with saved user settings.
  • Access granted per department and per patient by the administrator; every access written to the audit log.
  • De-identified accounts for teaching and research — archive without identity, and no live view.
  • Reports generated from numeric data and waveforms for the record.
ipv.hospital.local/ipv/settings/access
Access & roles
4 roles · 38 accounts · signed in through the hospital directory
Admin

Sees live waveforms and a year of archive on the intensive care unit, can measure and annotate, and can read — but not export — the two cardiac wards.

DepartmentLiveArchiveMeasureExport
Intensive careFull accessFull accessFull accessFull access
CardiologyRead onlyRead onlyNo accessNo access
Cardiac surgeryRead onlyRead onlyNo accessNo access
EmergencyNo accessNo accessNo accessNo access
Patient scope
Every bed in their own unit, plus patients referred to them
Data visibility
Identifiedevery access written to the audit log

Pick a role to see what it reaches. Researchers get de-identified archive and no live access; biomedical engineers get bed and device status and never a patient name.

The data

Everything the monitors said, on request

A documented HTTP API over the same store the screens read from. Ask for a bed, a signal, a range and a resolution; the store does the folding and sends back the answer.

  • Numeric trends for any bed and signal: mean, min, max, median or a percentile over any window, from a day to the whole archive.
  • Raw waveforms at the monitor’s own 240 Hz, any lead, any 30 seconds of the retained week. The samples, not a screenshot.
  • Notification episodes with start, duration, limit and extreme value; whole cohorts exported de-identified as CSV or Parquet.
  • Tokens carry a role. An integration reaches exactly what that role reaches on the access screen, and every call lands in the same audit log.
ipv.hospital.local/ipv/data/explorer
Data API · query explorer
REST over the same store the screens read · numerics, waveforms, episodes, exports
API v1
Signal
Range
Every1 min is the stored interval
Function
GET /api/v1/series?bed=ICU-04&signal=hr&from=-24h&every=15m&fn=mean
200 OK · 96 points · 2 ms · 3.4 KB · application/json1,440 stored values folded into 96 inside the store
HR · mean · every 15 minICU-04 · last 24 h
Response
{
  "bed": "ICU-04",
  "signal": "hr",
  "unit": "bpm",
  "from": "2026-03-14T07:30:00Z",
  "to": "2026-03-15T07:30:00Z",
  "every": "15m",
  "fn": "mean",
  "points": 96,
  "values": [
    ["2026-03-14T07:30:00Z", 76.4],
    ["2026-03-14T07:45:00Z", 75.9],
    ["2026-03-14T08:00:00Z", 74.5],
    ["2026-03-14T08:15:00Z", 73.7],
    … 92 more
  ]
}
Also on /api/v1
/waveform

Any lead, any 30 seconds of the retained week, at the monitor’s own 240 Hz. The samples, not a picture of them.

/episodes

Notification episodes with start, duration, limit and extreme value. Per bed or per unit, for a quality review or a study.

/exports

A defined cohort over months, de-identified, as CSV or Parquet. Runs in the background and calls back when the file is ready.

/stream

Live values as each one arrives, as server-sent events. Same token, same scope as everything above.

Authorization:Bearer ipv_k7…e3frole: consultantreaches what the role reaches · every call written to the audit log

Change the window or the function and watch the request rewrite itself. The aggregation runs inside the time-series store; what leaves it is the answer, at the size you asked for, whether that is a day at one minute or a month at one hour.

The intelligence layer

What the stored data makes possible — today, and next

Shipping today: the IPV assistant

Clinicians can ask questions about a patient in plain language, with answers based only on that patient’s stored data and events.

Natural-language queriesAsk about a patient in plain language — for example, to summarise overnight haemodynamic instability.
Grounded in real dataAnswers draw on that patient’s own trends, waveforms and notifications. Nothing is invented.
On-premiseRuns inside the hospital’s infrastructure. Patient data does not leave the institution.

In development: predictive analytics

The four modules below are in development and not part of the current release. The stored data stream is what makes them possible.

Vital-sign forecastingForecasting how vital signs will develop over the coming minutes to hours, from many signals at once.
Early-warning scoringCombined risk scores to flag at-risk patients before instability becomes visible.
Arrhythmia and anomaly detectionPattern recognition on high-rate ECG to find abnormal beat shapes and rhythms.
Automatic ECG landmarksAutomatic P-Q-R-S-T detection, to make caliper measurement faster and more consistent.

Beyond the ward

What the archive becomes once it exists

Stored monitoring data is not only a clinical tool. For a teaching or research hospital it is an institutional asset — with its own value, and often its own budget.

Research-grade physiological datasets
De-identified, high-resolution vitals and waveforms with notification context — the input modern clinical and AI research is built on, drawn from your own patient population.
Teaching material from real cases
Genuine deteriorations, arrhythmias and interventions, replayable on screen, anonymised and safe to show in a lecture theatre or at the bedside.
A place in collaborative studies
Hospitals that can supply structured continuous monitoring data are the ones invited into multi-centre studies and industry partnerships.
A head start on every future AI project
Whatever model the hospital eventually wants to run, it will need years of stored signal to train on. Capture begins the day IPV is installed.

Low-risk by design

IPV changes nothing about the equipment or the routine on the ward

The usual concern about a monitoring platform is what it forces the ward to change. For IPV: nothing.

The monitor remains the primary device

IPV reads from the interface gateway. It does not configure monitors, does not suppress or replace alarms, and is not the device the bedside depends on.

No new hardware in the patient room

Nothing is added at the bedside, nothing is unplugged, and no cable inside the patient environment is touched.

No change to clinical workflow

Nobody is asked to enter data, tick boxes or work differently. IPV adds a view; it removes no step from what nurses and doctors already do.

Nothing to install on the client side

The application runs in a browser on the computers, tablets and phones the hospital already owns. Training a clinician takes minutes.

Platform & deployment

Where it runs, how the data is kept, and how long it takes

Everything IPV asks of the hospital’s infrastructure, in one place. From kickoff to a live ward takes two to ten working days, and most of that is the hospital’s own change-approval process rather than the installation itself.

Where it runs
Ubuntu Server LTS on a single virtual machine, 1× Gigabit Ethernet, on-premise or in a private cloud, virtualised (e.g. VMware). Sized to bed count and retention period.
How data is stored and kept
A time-series store built for continuous, high-rate patient data. Raw data kept ≥7 days, compressed after that and retained ≥1 year — configurable by the administrator.
Security, access and audit
Encryption in transit, Active Directory and smart-card sign-in, access scoped per department and per patient, full audit logging, and de-identified accounts for research and teaching.
Clients and languages
Thin HTML5 client, nothing to install: PCs, tablets and phones. Fully internationalised — the interface can be delivered in any language.

Data intake, step by step

Bedside monitors
ECG, SpO₂, ART, temperature and perfusion — continuous physiological output from the equipment already installed.
Interface gateway
GE CARESCAPE Gateway: numeric values and patient ADT context over HL7, high-rate waveforms and alarm events over HSDI.
Instap ingestion and store
HL7 values and HSDI waveforms written to one time-series store at the monitor’s full output resolution — ECG at 240 Hz.

Before you ask

The five questions every hospital asks us

Short answers, so that they are on the record before the conversation starts.

Is IPV a certified medical device?
Certification is under way — we are targeting November 2026. Until then IPV is supplied as a clinical data viewing and archiving system alongside the certified bedside monitors.
Who is responsible for a clinical decision?
The bedside monitor remains the primary monitoring and alarm device. IPV presents and stores data; it does not replace clinical judgement or the equipment the ward relies on.
Where does patient data live?
On your own servers. IPV is deployed on-premise, with no cloud component and no patient data leaving the institution — the AI assistant included.
Does it integrate with our HIS?
The interfaces are open and the integration is scoped per hospital — from reading patient context to handing data back. We agree what is needed and price it with the deployment.
And if Instap disappears?
IPV runs entirely inside your infrastructure, so it keeps running. Source-code escrow and data-export terms can be written into the contract.

Why now

Three reasons this is a decision for this year, not the next one

The data clock is already running

No predictive model can be trained on data that was never stored. Every month without capture is a month of the hospital’s own patient history permanently lost.

The system you have is running out of road

Older remote viewing servers are typically out of support, on operating systems that no longer receive security patches — and unpatched systems increasingly fail hospital audits outright.

Mobile is now the expectation

Clinicians carry phones and tablets. A monitoring system that cannot reach the device in their pocket is a system that is not consulted when it matters.

Next steps

Start with one ward, then decide about the rest

The decision does not have to be the whole hospital. A single department answers every question that actually matters.

  1. 01

    Scope: one department

    One intensive care, cardiology or cardiac surgery unit, on the monitors already installed there — everything the full deployment does, at a scale you can supervise.

  2. 02

    Setup measured in days

    Two to ten working days from kickoff to clinicians logging in, on a single virtual machine, with no change at the bedside.

  3. 03

    Decide on evidence, not on a demo

    Agree the success criteria in advance — who uses it, for what, and what they should be able to do that they cannot today — then review them against real use.

  4. 04

    Extend only when it has earned it

    If it works, the same installation grows department by department. If it does not, you have spent weeks rather than a capital programme.

To begin

Three things and we can start

The make and model of your monitors, whether an interface gateway is already installed, and one department willing to try it.

Talk to usLicensing models