Healthcare is the environment where consumer smart home thinking transfers least well, and it is worth being blunt about why before discussing anything else. A device that fails in an interesting way is annoying in a house, inconvenient in an office and unacceptable in a hospital. A system nobody has validated is a curiosity at home and a governance problem in a clinical setting.
That does not mean the technology has no place. It means the boundary matters more than the capability, and the useful work sits almost entirely on one side of it.
The line, stated plainly
Anything that touches patient diagnosis, treatment, monitoring or care pathways is a medical device or a clinical system. In the EU that brings it under medical device regulation, with conformity assessment, clinical evaluation, post-market surveillance and a manufacturer who carries legal responsibility. Similar frameworks exist in the UK, the US and elsewhere. Software that influences a clinical decision can itself be a regulated device.
Consumer smart home equipment is none of those things. It carries no such certification, its manufacturers make no such claims, and using it in a clinical role would place the hospital in the position of having deployed an unvalidated device in patient care.
Where it genuinely helps: the estate
A hospital is also simply a large, complex, expensive building operating continuously, and buildings have problems that sensors solve well. None of the following touches a patient pathway.
Plant and infrastructure
Boiler rooms, chiller plant, air handling units, pump rooms, lift machine rooms, electrical switchrooms, generator houses. Temperature, humidity, water detection and vibration monitoring in these spaces catches failures before they become incidents. A leak in a plant room above a ward matters a great deal, and it is a building problem with a building solution.
Non-clinical storage
Linen stores, consumables, catering stores, records archives, IT rooms. Temperature and humidity monitoring protects stock and equipment, and none of it is regulated storage.
Energy and water
Hospitals are among the largest energy consumers in the public estate, running twenty-four hours a day with substantial mechanical plant. Sub-metering by building, by department and by plant item is where energy management becomes possible rather than notional.
Non-clinical space comfort
Offices, meeting rooms, staff rooms, training rooms, admin blocks. These behave like any commercial building and can be treated as such.
Estates workflow
Door contacts on plant room doors, occupancy in rarely used spaces, environmental alerting to the estates helpdesk. Unglamorous and genuinely useful.
The grey areas, and how to treat them
Several things sit near the line and deserve individual judgement rather than a general rule.
Pharmaceutical and sample storage
Medicines fridges, vaccine fridges and specimen freezers require validated monitoring with calibrated sensors, traceable certificates and audit trails. Consumer sensors do not meet that standard and must not become the compliance record. They can occasionally be justified as a supplementary early-warning layer — an additional alert alongside the validated system — but that has to be agreed with pharmacy and quality colleagues, documented as supplementary, and never presented as the primary monitor. We cover this properly in cold chain and stock monitoring.
Ward environment
Temperature and air quality on a ward affect patients, which makes it a clinical environment even though the sensor is measuring a room. Treat it as clinical: involve infection prevention and control, and use equipment specified for healthcare settings.
Water systems
Hot and cold water temperature monitoring is both an estates duty and a patient safety matter because of legionella. It is a legitimate and valuable application, and it sits inside a formal water safety framework with named responsibilities. Automate the logging; do not invent the policy.
Staff areas and rest facilities
Non-clinical, and a legitimate place for ordinary building technology. Worth noting that staff privacy expectations apply just as they would in any workplace.
Why hospitals are technically difficult
Beyond regulation, several practical realities shape what is possible.
The building never empties. There is no holiday shutdown and no quiet weekend. Work happens around continuous operation, in short windows, with permits.
Networks are tightly controlled, correctly. Clinical networks carry patient data and connected medical devices. Estates devices do not belong on them, and any proposal will — and should — be met with a security review. Expect to use a separate segmented network, and expect that to be a condition rather than a suggestion.
Interference is a genuine concern. Radio-emitting devices near sensitive medical equipment require thought. Many hospitals have policies on this, and they exist for good reasons.
Infection control governs what may be installed. Surfaces must be cleanable, devices must not obstruct cleaning, and some materials and mountings are unacceptable in clinical areas. Consult infection prevention and control early, not at the end.
Estates and IT are separate organisations with different priorities, and a building sensor project sits awkwardly between them. Identifying who owns it is a prerequisite rather than a detail.
The honest value proposition
Set against those constraints, is it worth doing? For the estate, yes, and for three specific reasons.
Failure prevention. A hospital cannot tolerate plant failures the way an office can. Early warning on temperature, water and vibration in plant spaces has a direct and quantifiable value.
Compliance evidence. Water temperature logging, emergency lighting test records, environmental records for non-clinical storage — automated logging replaces clipboards and produces defensible records for inspection.
Energy at scale. Because hospitals are so large and run continuously, percentage improvements represent substantial absolute sums. Sub-metering is the prerequisite, and most estates have far less of it than they would like.
What we would not claim is that any of this improves patient care directly. It improves the reliability of the building that care happens in, which is a real but indirect benefit, and overstating it is how estates proposals lose credibility with clinical colleagues.
How to start without causing a governance problem
- Pick a plant room. Non-clinical, no patients, no personal data, clear estates ownership. Temperature, humidity and leak detection.
- Get IT security involved on day one and accept their network requirements. A segmented VLAN is the normal answer.
- Write down the boundary in your own proposal: this system does not touch clinical systems, patient data, or regulated storage. Being explicit about what you are not doing prevents most objections.
- Route alerts to the estates helpdesk, not to individuals, with the same assign-acknowledge-escalate discipline used anywhere else.
- Demonstrate one prevented failure, then expand into the rest of the non-clinical estate.
- Approach the grey areas separately, each with the relevant professional colleagues, and never by extending an existing estates deployment into them quietly.
The discipline that makes this work is the willingness to say no to the interesting applications. A well-run estates sensing programme that has never gone near a patient is far more likely to survive and expand than an ambitious one that gets stopped by clinical governance in month three.