Digitalisation in Policing and Emergency Medical Services

That is worth saying out loud, because most commentary on public safety IT is either breathless or resigned. The more interesting question is what happens in the twenty minutes after that call: which systems talk to each other, which do not, and why the remaining gaps are rarely technical.

I have worked exclusively in the police domain since 2014, and I cover emergency services topics in a voluntary capacity alongside that. I am on firmer ground on the policing side than on the medical one, and I tend to look at the seams between organisations rather than at any single one of them. The view is Swiss, with sideways glances at Germany and Austria, where the same problems are being solved on different schedules and with different assumptions about who is in charge.

It starts in the dispatch centre

The dispatch centre is where digitalisation either pays off or quietly fails. Everything downstream depends on the quality of the decision made there in the first ninety seconds.

Switzerland has made this harder for itself than it needed to. Policing and health care are cantonal responsibilities, which means 26 structures, several separate emergency numbers, and dispatch centres built at different times by different organisations with different suppliers. Some cantons share a 144 centre, others operate their own. Police dispatch is organised separately again. A caller who is not sure whether they need the police or an ambulance has a decision to make that the system should be making for them.

The counter-example is worth studying. Notruf Niederösterreich in Austria runs the medical emergency number and the health advice line for an entire province from one organisation, and the effect shows up in the details: chat-based emergency access for callers who cannot speak on the phone, documentation shared between paramedics and physicians, live ECG transmission from the scene to the receiving hospital. None of this is exotic technology. These are ordinary integrations that become affordable once you only have to build them once.

Germany is trying to fix its version of the same problem by legislation. The emergency care reform approved by the federal cabinet in April 2026 requires digitally supported standardised triage at every entry point and, more importantly, that the 112 dispatch centres and the 116117 service coordinate digitally rather than by telephone handover. Around 225 million euros is earmarked for dispatch centre digitalisation and networking. Spread across sixteen federal states, that is less impressive than the headline suggests, but the direction is unambiguous and it is backed by law.

The Swiss equivalent debate happens mostly at conferences. There is no federal instrument comparable to the German reform and limited political appetite for one. What we get instead is cooperation between cantons where the operational pain is high enough to overcome the institutional friction.

The connectivity problem nobody wants to own

Polycom, the national security radio network, completed its renewal programme in mid-2026. It converted the network to IP technology, ran broadly to plan, and will keep delivering reliable voice with good coverage and independent power supply for years.

It also solves a problem from the 1990s. Polycom is narrowband. It does not carry a live video feed from a body-worn camera, a building plan, or a patient monitor stream.

Its designated successor is a broadband system built on commercial mobile technology, intended to replace Polycom gradually by 2035. As of mid-2026 it sits in the initialisation phase of a preliminary project, with the financing of that preliminary project still to be clarified. Read that again and then look at the date. We are discussing a nationwide broadband capability for blue-light organisations that arrives, at the earliest, towards the end of the next decade.

Germany is in the same position and argues about it more publicly. The federal agency responsible has stated plainly that broadband for emergency services currently exists only as isolated solutions for individual organisations, and that a next-generation network only makes sense if it is built once, nationally, rather than sixteen times.

The consequence in both countries is predictable. Organisations that need broadband now buy it now, from commercial operators, with commercial service levels. That is entirely rational for a single ambulance service or police corps. Collectively it produces exactly the fragmentation the national programmes exist to prevent, and an installed base that will have to be migrated later at someone’s expense. There is no clean answer here. Telling operational units to wait ten years for a capability they need this year is not a policy.

What the crew actually carries

The vehicle is where digitalisation is most visible and where the gap between the slide deck and the shift is widest.

On the ambulance side, the electronic patient care record has largely replaced paper, and the second-order benefits are real: legible documentation, structured data for quality management, pre-notification of the receiving hospital before the vehicle arrives. What matters operationally is how much the crew does not have to type. Current systems take vital signs, the 12-lead ECG and ventilation data directly from the connected monitor, defibrillator or ventilator, then pass the completed record to the dispatch and billing systems without a second round of data entry. Getting that ECG in front of the cardiology team before the ambulance arrives shortens door-to-balloon time, and that is measurable in patient outcomes. It is the strongest single argument for the whole exercise and it gets less attention than the flashier applications.

On the police side, mobile access to the systems that matter has been the goal of the harmonisation effort for over a decade. The cantonal corps and federal agencies now maintain a shared portfolio through a joint organisation, covering case management, identity and access management, encrypted messaging and mobile operations support. It is run largely as a militia structure with a small professional core, which tells you something about both its cost efficiency and its speed.

Three problems recur regardless of organisation. Double entry: if the tablet does not write into the system of record, the crew fills in the form twice, and after a few weeks they optimise by filling in the tablet badly. Connectivity: a system that assumes coverage behaves unpredictably in an underground garage, an Alpine valley, or a stadium with 30,000 phones on the same cell, and offline-first design is still not universal. And training: introducing an application to a workforce that works irregular shifts under pressure, with a high proportion of volunteers in some organisations, is a change management problem before it is an IT problem. Budgets rarely reflect that ratio.

Handover, or the interoperability question

The moment a patient is handed over at the hospital door is where the digital chain in Switzerland currently breaks.

The electronic patient record was supposed to be the shared substrate. By spring 2026, roughly 136,000 records had been opened in a country of nine million people. Adoption has been slow enough that the federal government has chosen to replace the underlying law rather than patch it, with a realistic operational date for the successor around 2030.

For an ambulance crew this has a simple meaning. You cannot build your handover process around a record most of your patients do not have. Pre-notification therefore still runs on regional agreements, direct interfaces between dispatch and specific hospitals, telephone, and in the better cases a shared capacity registry showing which emergency department can actually take the patient. These regional solutions work well. They also stop at the cantonal border, which is a problem in a country where the nearest suitable hospital is frequently in the next canton.

The police equivalent is the same story with different data. Harmonising across 26 cantonal systems is a governance exercise with an IT component, not the other way around. The technical means to share a query across corps have existed for years. What takes the time is agreeing on data protection responsibilities, retention rules, who pays for the shared platform, and which corps gives up a system it is fond of.

Artificial intelligence, minus the marketing

There is one application here where the evidence is genuinely good. Software that listens to the emergency call and flags a probable cardiac arrest, based on the caller’s description and background audio, catches a meaningful share of cases the human call-taker misses, and catches them faster. Cardiac arrest is the ideal test case: the decision is binary, the time window is short, and the cost of a false negative is a death.

Beyond that, the picture is mixed. Live transcription, automatic translation for callers with no local language, structured extraction of address and symptom data, prioritisation support for simultaneous incidents: all reasonable, several in trials, none yet transformative.

The constraints deserve equal billing. A dispatcher’s screen is already full, and an assistance system that adds a fifth window will be ignored under load, which is precisely when it was meant to help. False positives are paid for in resources not available elsewhere. And the regulatory position is now explicit: the EU AI Act classifies systems that triage emergency calls and prioritise dispatch as high-risk, with the documentation, oversight and testing obligations that follow. Switzerland is not bound by that regulation, but the vendors selling into the Swiss market are, and there is no realistic scenario in which Swiss procurement ignores the standard.

My own view, held loosely: AI in the dispatch centre will earn its place through narrow, well-evaluated functions that reduce the cognitive load on the call-taker, not through a system that proposes to make the disposition decision. The judgement of an experienced dispatcher remains the most underrated asset in the chain.

What would actually help

Four things, none of them requiring new technology.

The direction of travel is right and several of the individual achievements are good. The remaining problems are mostly organisational, which is a less satisfying conclusion than a technological one but a more useful place to spend the next few years of effort.