Skip to main content

VitalSync — vitals charted without anyone typing

A small camera at the bedside reads the patient monitor you already own. Readings are recorded automatically every 30 seconds, trends build themselves, and the ward is warned when a patient starts to deteriorate — hours before a routine observation round would have caught it.

New monitors required
0New monitors required
Speaks your standards
FHIR + HL7Speaks your standards
Between readings
30 secBetween readings
BEDSIDE VITALS, CAPTUREDHR78SpO296BP138/88No wiringEvery30 secWard alertsBed 12 - BP risingBed 04 - stable

What VitalSync changes on the ward

Monitors you have to replace
0Monitors you have to replace
Charting interval, hands-free
30 secCharting interval, hands-free
Outbound transports, from MQTT to MLLP
7Outbound transports, from MQTT to MLLP
Confidence recorded before it can alarm
Per readingConfidence recorded before it can alarm

Why it matters

Why wards run VitalSync

Vitals are usually charted a few times a shift, by hand, onto paper. Everything that happens between those rounds is invisible.

Nobody transcribes numbers

Readings go from the monitor screen into the record directly. That is time returned to nursing, and one whole class of transcription error removed.

Keep the monitors you own

VitalSync reads existing bedside monitors with a camera and speaks to protocol-capable devices directly. No capital replacement programme.

Deterioration is caught earlier

A continuous record makes a trend visible. A scoring model watches those trends and raises a patient who is drifting the wrong way before the next round.

Alarms you can trust

Every reading is scored for confidence and cross-checked against the patient's other vitals. A number the system is unsure about is shown, but never allowed to raise a clinical alarm.

How it works

How capture works

From the monitor screen to the ward alert.

  1. BEDSIDE CAPTURE

    Step 1: Read

    A bedside camera watches the monitor you already own, or the device is read directly over its own protocol.

  2. VERIFIED

    Step 2: Verify

    Each reading gets a confidence score and is cross-checked against the patient's other vitals before it is trusted.

  3. CHARTED

    Step 3: Chart

    Verified readings are written to the patient's record every 30 seconds, building a trend nobody had to type.

  4. ESCALATION

    Step 4: Warn

    A scoring model watches the trend and raises the patients drifting the wrong way on the ward alert centre.

Capabilities

What VitalSync does

Camera capture at the bedside

A mini PC and camera per bed read the monitor display continuously and publish each reading to the ward server.

Direct device integration

Monitors that speak a serial or network protocol are read directly, with their data checksummed at the source and trusted by default.

Confidence gating

Low-confidence or self-contradicting readings are stored and displayed but marked unverified, so a misread number cannot trigger a clinical alarm.

Deterioration scoring

A trained model scores each patient's trend and raises the ones heading the wrong way, separating clinical concern from technical noise.

Separated alert centre

Clinical, technical and system alerts sit on their own tabs. A camera problem never appears as a patient problem.

Built for hospital networks

Edge devices reach the server over MQTT on a small, fixed set of ports. Databases and queues stay on loopback and are never exposed to the LAN.

Modules

Integrations and advanced capability

VitalSync is not a closed box bolted onto a ward. It speaks the standards your other systems already speak, writes back into your hospital system, and carries the alerting and clinical scoring depth that a real deployment needs.

FHIR AND HL7

Standards: FHIR and HL7 v2

  • FHIR read API with a published CapabilityStatement
  • Patient and Observation resources served as FHIR
  • HL7 v2 ORU result messages for downstream systems
  • MLLP transport for classic HL7 interfaces

What it changes

  • Your existing integration engine talks to VitalSync unchanged
  • No proprietary format for another vendor to reverse-engineer
  • Vitals reach the EMR as observations, not as a screenshot
HMS WRITE-BACK

Hospital system sync and write-back

  • Configurable connector: capability discovery and a connection tester
  • Patient resolution against your existing identifiers
  • Request templating and JSONPath transforms, no bespoke code per site
  • Watermarks, idempotency keys and backfill for safe replays

What it changes

  • A new hospital is configured rather than custom-built
  • A retry or a replay cannot double-write a reading
  • Historical gaps can be backfilled without hand-editing data
FAN-OUT TO DESTINATIONS

Outbound delivery to any destination

  • Transports: MQTT, AMQP, Azure Service Bus, TCP, HTTPS push, database sink
  • Signed webhooks for third-party consumers
  • Publish rules decide what leaves and where it goes
  • Transactional outbox with a delivery log and dead-letter queue

What it changes

  • Feeds a research database and a vendor API from the same event
  • A destination being down delays delivery, it does not lose it
  • Anything undeliverable lands in a dead-letter queue to inspect
BEDSIDE CAPTURE

Device capture and protocol onboarding

  • Camera OCR against the monitor you already own
  • Direct serial, RS232, TCP and HTTP device protocols
  • AI-assisted protocol onboarding: sample, analyse, review, promote
  • Heartbeats, offline detection and ingest health per device

What it changes

  • An unfamiliar monitor is onboarded without waiting for a driver release
  • A biomedical engineer promotes the parser, not an engineering sprint
  • A silent bedside device raises a technical alert, not silence
NEWS2 AND PEWS

Clinical scoring and early warning

  • NEWS2 and PEWS computed and stored on every reading
  • Derived indices: shock index, ROX and MAP
  • Layered thresholds: ward, patient type, per-patient override, with history
  • Per-patient vital baselines

What it changes

  • The trend is kept, not only the moment a score crossed the line
  • A paediatric bed and an ICU bed are not judged by the same numbers
  • Every threshold change is attributable and reversible
TRAINED ON YOUR DATA

Deterioration model, trained on your data

  • Model versions, training runs and promotion tracked in the system
  • Promotion re-runs the runtime validation before a model goes live
  • Risk feature export for analysis and retraining
  • An invalid model disables scoring and raises a system alert

What it changes

  • Scoring reflects your own patient population, not a generic cohort
  • A bad model cannot quietly reach the ward
  • You can see which model version produced any given score
TIMED ESCALATION

Compound rules and escalation

  • Multi-condition rules with AND/OR logic and a severity
  • Sustained-breach duration, so a momentary spike does not alarm
  • Ward escalation policy with timed levels, default 5, 10 and 20 minutes
  • Routing to the nurse actually on shift

What it changes

  • "Low SpO2 and rising respiratory rate" is one alert, not two
  • Fewer false alarms means the real ones still get attention
  • An unacknowledged alert climbs the ward hierarchy on its own
WHATSAPP, SMS, EMAIL

Alert delivery channels

  • WhatsApp, SMS, email and push notification handlers
  • Delivery logged per alert, with a dead-letter path
  • Contacts, contact roles and per-patient contact overrides
  • Separate clinical, technical and system alert streams

What it changes

  • The alert reaches a person on the channel they actually read
  • You can prove an alert was delivered and when
  • A camera fault never reaches a nurse as a patient problem
API AND AUDIT

API access, audit and administration

  • Developer portal with managed API clients and keys
  • Versioned REST API across wards, beds, patients, vitals and alerts
  • Full audit log of configuration and threshold changes
  • Integration health and ingest health endpoints

What it changes

  • Your own team can build on the data without a bespoke export
  • Configuration changes are attributable during an incident review
  • Integration failures are visible before someone notices missing data

Put VitalSync on one ward first

We deploy to a single ward against your existing monitors so you can judge capture accuracy and alert quality on your own patients before going wider.

Also from Nefronix

Pairs with

HOSPITAL PLATFORMNexHIS1Register2Consult3Lab4Pharmacy5BillingAI fills the notesVoice and scans in,records out. Staff confirm.Go live in1 daynot weeksFlagship platform

NexHIS

AI-native hospital management platform

  • Switch in one day, not one month
  • Notes drafted from voice, staff just confirm
  • Runs OPD, NexLab, pharmacy and billing
See the platform
PATIENT CALL SYSTEM12sresponseNurse stationBed 01SettledBed 07Needs helpBed 14SettledBed 22SettledHospital IoT

CareConnect

bedside call and response tracking

  • Bedside button straight to the nurse station
  • Every response timed and reported
  • Covers a whole building without new cabling
See CareConnect

See automatic vitals capture on your own monitors