How clinical information actually works, and where it's straining

From single answers to a living model

In most of medicine today, a question gets asked, a test gets run, and an answer gets filed away. That model works well when care happens in isolated episodes. Once monitoring never really stops, the same model starts to strain. Here's the shift, in four parts.

01 Today's default

The episodic diagnostic workflow

A test is ordered to answer one specific question. A specialist reviews it, reaches a conclusion, and that conclusion, not the underlying data, is what enters the record. Once the question is answered, the episode is considered closed.

ConcretelyA cardiac MRI can generate gigabytes of imaging data across hundreds of slices. What actually reaches the patient's record is a single line: healed normally. Once the question that prompted the scan is answered, the rest of that data effectively disappears from the ongoing story: archived, technically available, but for practical purposes inert.

02 What actually feeds the chart

An ecosystem of two very different kinds of data

Both kinds of information describe the same patient, but they behave completely differently. Traditional sources answer one specific question, then go quiet. Continuous sources keep producing data whether or not anyone asked.

Traditional and episodic

Feeds the System of Record: captured at a visit, reviewed once, then settled.

History and demographics

Intake facts: allergies, family and social history, past surgeries.
Encounters

Scheduled visits, ER trips, telehealth calls, logged as discrete events.
Diagnoses

The coded problem list a physician has formally assigned.
Medications

What's prescribed, via e‑prescribing and pharmacy fill records.
Vitals (in clinic)

BP, HR, weight: a single snapshot taken at intake.
Lab results

Finished panels with reference ranges, from the lab system.
Imaging reports

The radiologist's signed interpretation, not the raw scan.
Clinical notes

Subjective, objective, assessment, plan, per visit.

Continuous and always on

Generated between visits: this is what continuous medicine adds on top.

Wearables

HRV pattern, sleep trend, activity and step decline, home BP trend.
Remote monitoring

Continuous glucose monitors, home telemetry platforms.
Implantables and telemetry

Pacemaker or defibrillator data streaming between visits.
Closed-loop therapies

Neuromodulation and adaptive devices that adjust themselves live.
Patient-reported input

"More tired climbing stairs", logged through an app, not a chart.
Caregiver-reported input

"Less active this week", observed by someone who isn't a clinician.
Ambient sensing

Home-based sensors picking up routine and behaviour changes.
Model-flagged anomalies

Patterns an algorithm notices that no one specifically asked for.

03 What breaks when observation never stops

Five ways continuous medicine changes the rules

Episodic medicine treats a conclusion as an endpoint: the question is answered, the case is closed. Once observation is continuous, conclusions start behaving less like completed answers and more like working models that stay open.

New measurements arrive continuously

Remote monitors and wearables produce readings every minute of every day, not just at a scheduled visit. The old rhythm of "measure, then wait" no longer applies.

One reading per visit A constant stream

Treatments generate new observations while being administered

Closed-loop and adaptive devices produce data while they're actively treating the patient, so treatment and observation are no longer separate steps in sequence.

Treat, then re-test later Treating and observing at once

Emerging patterns create new questions

A trend that only becomes visible over weeks, a gradual decline in activity or a drifting sleep pattern, raises questions that no single visit would ever have prompted.

No visit, no question The pattern asks the question

Prior observations acquire new significance

A scan performed years ago can suddenly matter again because of something a clinician notices today, but only if that old data is still reachable and connected to the present.

Old data stays archived Old data can resurface

Previously irrelevant information becomes important

An occasional palpitation, or an anomaly an algorithm flagged and nobody acted on, dismissed as noise at the time, can turn out to be the missing piece once a new pattern emerges.

Filed away as noise A clue waiting to be reconnected

04 The missing piece

The Context Layer: a living model, not a filing cabinet

The Context Layer isn't a bigger version of the record. It's a continuously updated model of patient state that holds the relationships between observations, treatments, interpretations, and trajectories, and keeps them alive instead of letting them collapse into a one-time report.

What exists today

System of Record

The EHR and its archives

What's missing

Context Layer

IMI, beside the record

Its job

System of Record Preserve decisions, actions and conclusions.

Context Layer Hold what everything currently known means for this patient now.

Question it answers

System of Record “What was decided, and when?”

Context Layer “What does all of this mean for this patient today?”

What it stores

System of Record Signed reports, coded diagnoses, orders, medication lists and visit notes.

Context Layer Every observation plus the links between them: trends, open questions, clinician concerns and constraints.

Unit of information

System of Record A document or a coded entry, filed on its own.

Context Layer A relationship: this reading, understood against that history.

Treats a conclusion as

System of Record An endpoint. The case is answered and closed.

Context Layer A working model, open to revision as new data arrives.

Uncertainty and nuance

System of Record Mostly discarded once a report is filed.

Context Layer Kept on purpose. Weak signals and “watch for” notes stay attached.

Update rhythm

System of Record Per visit or per order.

Context Layer Continuously, as each observation arrives.

Relationship to other systems

System of Record Arm's length. It receives finished artifacts only.

Context Layer Actively connected to devices, labs, the record and patient-reported input at once.

How insight reaches a clinician

System of Record Someone has to remember to search for it.

Context Layer It surfaces what matters with its reasoning; a clinician confirms or dismisses.

Worked example The cardiac MRI, one year later

System of Record The chart holds one line: “healed normally.” The scan is archived. When palpitations begin months later, nobody thinks to reopen it.

Context Layer The scan stays linked to the patient's picture. When palpitations and a drifting heart rate appear, the prior MRI resurfaces for review with the reason it matters now.

Where the patient and the clinician meet

Everything from sections 1 to 3 flows through one synthesizing core. Settled facts from the System of Record and continuous signals from the patient's life go in; a working, reviewable picture comes out for the clinician.

System of Record: settled facts Continuous medicine: always-on signal Surfaced by IMI for clinician review Clinician decision

Discovery asks what can be learned from data. Context asks what everything currently known means for this patient now.

None of this replaces the System of Record. The Context Layer draws on it constantly, and feeds insight back into it through alerts and decision support. But it can only work if clinicians have a privileged role in how it's assembled and interpreted, and building it requires AI: the semantic integration across records, devices, patients and caregivers is too dynamic to do any other way. AI without context can automate the record. Clinical reasoning requires context.