From Health 201FailSystems

how automated healthcare fails, how you'd know, and what to do at each tier — every claim sourced, reviewed continuously


Healthcare is becoming automated. What happens when the system fails?

Hospitals are turning into cognitive composite systems: humans, AI models, agents, devices and infrastructure working as one. When the power goes out, the network drops or the model is wrong, lives depend on what happens next. FailSystems is a sourced, continuously updated reference for understanding, classifying and designing against that failure.

The failure layers

Layer 1

Power & infrastructure

Electricity, fuel, water, heat and cooling, including the physical data centres everything clinical now runs on; when they fail, the decision support goes too.

Involved in 7 of 39 incidents (main layer in 7)
Draft score 45/125 v0.1 draft
Layer 2

Connectivity & data

The networks, interfaces, vendors and records that carry clinical data, and what happens when data is missing, late or wrong.

Involved in 15 of 39 incidents (main layer in 10)
Draft score 75/125 v0.1 draft
Layer 3

Devices & electronics

The monitors, pumps, ventilators, sensors and analyzers that measure and act on patients, and the software and updates that run them.

Involved in 13 of 39 incidents (main layer in 7)
Draft score 80/125 v0.1 draft
Layer 4

Models & agents

Predictive models, LLMs and agents that stay up while giving wrong answers, so care continues on bad output without anyone switching to a fallback.

Involved in 12 of 39 incidents (main layer in 9)
Draft score 60/125 v0.1 draft
Layer 5

The human handoff

The moment automation hands work back to a person who has stopped watching, stopped practising, or stopped hearing the alarms.

Involved in 26 of 39 incidents (main layer in 6)
Draft score 48/125 v0.1 draft
Pattern

Cascades

Not a layer: the way one failure crosses power, connectivity, devices, models and handoff, and spreads from one organisation to many.

Involved in 16 of 39 incidents
Draft score 80/125 v0.1 draft

An incident usually involves several layers, so the counts overlap; there are 39 incidents in all. Scores are first-draft FailSystems judgements, not measurements: how they are made.

Graceful degradation, planned in advance

A safe automated system knows how to become a less automated one. See the full layer × tier matrix.

0

Full automation

Normal digital operation: software acquires, routes and checks data, recommends, and in places acts; clinicians supervise and decide.

1

Assisted operation

Automation degraded or distrusted — humans decide, machines advise.

2

Manual operation

Electronics up, cognition down. Staff run the workflow on devices alone.

3

Analog fallback

Power or network gone. Paper, batteries, hand calculation, judgement — and a rehearsed plan.

Recently updated

27 September 2026

Accuracy review: classifications, tiers and sources corrected

27 September 2026

New incident: Patient dies while a heart monitor's crisis alarm is switched off

26 September 2026

FailSystems becomes a sourced reference

26 September 2026

Degradation tiers: the layer × tier matrix

26 September 2026

Nightly evidence pipeline switched on

39 incidents · 216 sources · full changelog · how this stays current

Information only, not advice. FailSystems is an aggregation and synthesis of published sources. It is not consulting, engineering, legal, regulatory or medical advice, and using it creates no professional relationship. Health systems are complex and no approach fits every organisation: anything you adopt is your own decision, at your own risk, and should be checked against the current official sources and by qualified people who know your setting. Full disclaimer.

Dealing with an incident right now? This site is a reference, not an incident-response service. Activate your organisation's emergency operations plan and incident command, and: