Health, Medicine, Personal Care Services

No service weighs more heavily on the quality of life of a territory than health care, and none is more exposed, over the next ten years, to the combined pressure of an aging population and the concentration of hospital services. That is why health care in the mountains deserves a chapter of its own, and why how it fares in the age of AI is one of the grounds on which this essay's thesis will either hold up or fall apart.

Health issues in Valais are regularly discussed: hospital organization, the shortage of doctors, an aging population, the cost of insurance premiums. Rarely, however, at the intersection of technological change and cantonal geography. Yet it is precisely at that intersection that one of the most consequential questions for Valais over the next fifteen years will be decided: can a dignified health service still be offered in valleys that are both emptying out and growing older, or must we accept a gradual urban concentration that would eventually transform the very nature of settlement in Valais?

I believe we can. And I believe AI plays a decisive part in this, not as a single solution, but as what makes possible a model that, without it, would slowly become untenable.

The equation nobody likes to state

Let's start with the equation as it actually stands, without euphemism. Valais has a little over a hundred and twenty communes, most of them home to fewer than a thousand permanent residents, scattered across valleys sometimes connected to the plain by a single road — one whose access can be cut off for several days a year by snow, avalanche, or rockfall. This geography is not a detail: it is the condition under which any health policy must operate.

To this geography is added a demographic reality that offers no favors. The canton is aging, as we saw in chapter 3, and more than one Valaisan in ten will have turned eighty by 2035. Health care consumption rises mechanically with age, and it rises faster still when geographic isolation makes preventive care harder to access. An elderly patient facing a two-hour trip for a specialist consultation forgoes it more often than one who faces a ten-minute drive. This inequality of access, long accepted as the price of living in the mountains, becomes politically harder to sustain as the population it affects keeps growing.

To geography and demographics is added a third, blunter fact: the shortage of doctors. Family medicine in Valais, like in nearly every rural region of French-speaking Switzerland, is aging faster than the population it treats. Doctors settled in the valleys retire without always finding a successor. Young doctors trained at Swiss medical schools overwhelmingly choose urban or suburban practice, where working conditions are simpler, on-call schedules better organized, and professional life less solitary. The phenomenon also affects the Jura, central Switzerland, and certain valleys in Vaud or Ticino, but it takes on a particular intensity here because of the sheer geographic dispersion.

Hospital organization, for its part, is concentrating. The Valais Health Network, which employs around six thousand two hundred staff³⁸, has for some years been carrying out a reorganization approved by the Council of State: surgical activity is being consolidated in Sion for French-speaking Valais and in Brig for Upper Valais, while the sites in Sierre and Martigny retain basic care and geriatrics. This reorganization has its logic — quality of technical platforms, critical mass of expertise, economic efficiency. It also shifts, structurally, the hospital system's center of gravity toward the urban hubs. The valleys are mechanically pushed one notch further away.

All these factors converge on a simple observation: if nothing changes, access to care in the valleys will deteriorate over the next ten to fifteen years. Not in dramatic terms, not as outright deprivation — Switzerland remains Switzerland. But in longer waiting times, specialists who must be reached at greater distance, family doctors who are no longer replaced, regional hospitals that scale back their remit. This is the likely trajectory, and it is neither a scandal nor an inevitability: it is simply what the combination of constraints just described produces for this generation.

This likely trajectory can, I believe, be bent.

What AI can change, and under what conditions

In the field of health, artificial intelligence is not an abstract promise. As the models improve, it is concretely transforming several dimensions of care. Three of them concern a scattered canton like Valais directly.

Diagnostic support is the most immediately measurable contribution. Medical imaging models now match or exceed human reading on certain tasks: detecting lesions on X-rays, MRIs, retinal scans, biopsies. This does not replace the radiologist, the ophthalmologist, or the dermatologist; it multiplies their reach. A valley doctor equipped with reading-assistance tools can perform an initial screening on site and refer to specialist consultation only the cases that warrant it. At the scale of a canton, the effect is to massively cut unnecessary travel, shorten referral times, and put specialists' time — by nature the scarcest resource — to better use. Under serious conditions (data governance, tool certification, practitioner training), this contribution is already measurable in the places that have deployed it.

AI-assisted telemedicine is another field being transformed. The concept is old, and its real-world rollout long fell short of its promise: remote consultation cannot replace a clinical encounter for many conditions, and earlier tools amounted to little more than a souped-up videophone. What changes with recent models is the quality of the assistance built around the remote consultation: intelligent pre-intake, automatic transcription, suggested differential diagnoses, flagging of warning signs in the patient's account. Remote consultation becomes a serious clinical tool rather than a stopgap. For valleys where physical distance from the practitioner is a structural fact of life, this amounts to a major shift in the conditions of access to care.

It is probably in the care of the elderly, finally, that the structural stakes weigh heaviest over the next decade. Remote monitoring of vital signs, early alerts on signs of decline, support for emerging cognitive impairment, help for family caregivers: AI opens up possibilities here that traditional services — nursing visits, home meal delivery, visits from relatives — could not deploy at the same scale. For a canton whose elderly population is both growing and dispersing, this contribution is probably the most consequential of all. It allows older people to stay at home longer and in better conditions, easing pressure on care homes while preserving a way of life that people in Valais, like residents of mountain regions everywhere, hold dear.

These three contributions come with three conditions, without which they remain theoretical.

The first is the sovereignty of medical data. Health data is, in law as in public awareness, among the most protected data there is. All the uses described above rely on the processing of personal data, and the question of where it is stored, who can access it, and under what legal regime it is governed is not a technical detail. It is a political question. The canton can decide that its medical data will be processed on infrastructure under Swiss law; or it can, by default and without ever making an explicit decision, let market solutions route it toward servers subject to foreign jurisdictions. Both paths exist and are being actively chosen in other cantons and other countries. For Valais, which carries on questions of sovereignty a legitimacy inherited from its institutional history, this is probably where the gap between doctrine and practice becomes most visible.

Next comes the training of practitioners. No AI tool, however capable, delivers clinical value if it is poorly used. This requires initial and continuing training that is not, at present, equal to the scale of deployment expected. The HES-SO Valais, the Faculty of Medicine in Lausanne, and the University of Bern, which together cover medical training for the canton's two linguistic sides, all have parts to play, as do the Valais Medical Society and its French-speaking counterparts, the professional associations, and the Hôpital du Valais. Without this investment in people, the tools will remain underused or misused.

What remains is the governance of these trajectories. Who decides which tools get deployed, under what conditions, against what evaluation criteria? This governance cannot be left to technology vendors alone, who have a commercial interest in maximizing sales. Nor can it be left to clinicians alone, not all of whom have the time or the training to scientifically evaluate the tools put in front of them. It requires a body that combines clinical expertise, technological expertise, scientific rigor, and representation of the public interest. The canton can build one itself, or join intercantonal and federal initiatives; the first option has the advantage of proximity, the second that of critical mass. Most likely, the two need to be combined.

The shift, seen from a valley practice

The preceding chapters described the competitive shift that AI produces in skilled services and in age-old trades. In the field of medicine in Valais, this shift takes a particular form, governed by a different logic.

In health care, the point is not primarily to win back market share lost to offshoring or generic SaaS. It is something else, probably more important for the canton: offsetting the demographic shortage of practitioners by multiplying what the practitioners who remain are able to do. The balance of power is not being fought out between Valais medicine and that of another country; it is being fought out between the medicine Valais offers today and what it could offer tomorrow, with the same number of staff.

Picture a family practice in a Valais valley: one senior doctor, sometimes two, a nurse, a receptionist, roughly two thousand patients on the books. Under the current model, this practice absorbs most of its useful time on tasks that AI can markedly speed up. Writing letters to specialists, coding consultations for billing, updating patient records, preparing insurance reports, issuing medical certificates, reviewing laboratory results as a whole. According to the surveys the FMH regularly publishes on the administrative burden carried by Swiss doctors³⁹, these tasks eat up nearly a third of available clinical time.

Generative AI, well integrated into this practice's workflow, can significantly reduce that burden. Automatic transcription of consultations, AI-assisted drafting of letters, pre-diagnostic document analysis, automated coding subject to physician validation: all of these functions are technically possible today, and their deployment is advancing across several European health systems. The senior doctor who steers these tools intelligently can, without adding staff, free up a substantial share of clinical time. One hour recovered each day amounts, at the usual pace of a family practice, to roughly a quarter of a medical position gained without a single new hire.

For a canton struggling to recruit family doctors, and where a practitioner's retirement often means the outright closure of a valley practice, this leverage is anything but marginal. It can be the difference between a valley that keeps its doctor and one that no longer has one. This is not a matter of industrial productivity — medicine cannot be measured in consultations on an assembly line — but of the capacity to sustain a service that demographic constraints would otherwise have made untenable.

By the same token, the referral of patients to hospital specialists can be considerably improved through decision-support tools. A valley doctor hesitating over an atypical case can now work through the reasoning with an assistant trained on international medical literature, test the initial hypothesis, identify the most relevant additional tests, and refer the patient to the right specialist with a precision that initial training alone cannot provide. The valley practice gains in triage capacity, the cantonal hospital sees its caseload better directed, and the patient avoids unnecessary consultations. This gain shows up in the time it takes to reach the right care, which is the most telling indicator of a health policy in a scattered territory.

The senior doctor's clinical judgment

In medicine, human oversight of these tools carries particular weight, because error here has a cost it does not have when drafting a tax contract or a tasting note. The senior doctor steering the tools possesses something no tool can replicate: clinical judgment.

A doctor who has known their patients for ten years, who has seen a thousand variations of the same condition in practice, who can recognize from the faintest cue the clinical sign that changes the course of treatment, is able to use AI as a multiplier of that judgment. They question the tool about what surprises them, check hypotheses they had already formed, and extend their reasoning to cases they have encountered less often. The tool does not decide — the doctor does; but the doctor thinks more broadly and more quickly than they would have alone.

A younger doctor, by contrast, or a less experienced practitioner who defers to the tool without questioning their own judgment, can produce clinical errors that the system itself has no way of catching. The profession knows this well: a diagnostic aid is only as good as the person able to question it. The spread of AI in local, community-based medicine must therefore happen first and foremost with and through experienced practitioners, who have the clinical authority and the experience needed to validate its uses before they become widespread.

This distinction matters for cantonal strategy. It means that investment in community medicine should not aim only at attracting young doctors to the valleys — necessary, but not sufficient — but also at retaining, for as long as possible, the experienced practitioners who hold the territory together. Arrangements for gradual handover, extended part-time work, and support through practice succession become strategically just as important as measures to attract new talent. Without seniors to orchestrate it, AI in community medicine will not deliver its full effect; with them, it can extend by a decade the viability of a service that demographics alone threatened to undo.

Community medicine and specialist medicine

AI's contributions do not carry the same weight at every level of care, and they need to be distinguished rather than treated as a single block.

For community medicine — family practice, home nursing care, medico-social services — AI is above all a tool of capacity. It allows a reduced number of practitioners to cover a scattered territory without degrading the quality of care. It cuts the administrative burden that today eats up a disproportionate share of family doctors' time. It makes initial triage easier, which routes complex cases to specialists faster. For Valais, whose community medicine is precisely the link most weakened by the demographic shortage, this is where the benefits are most immediately visible.

Specialist and hospital medicine operates on a different register. Here AI improves diagnostic accuracy, the quality of surgical planning, prescription safety, and the prevention of medication errors. It opens up prospects in personalized medicine — adapting treatments to genetic profiles, predicting therapeutic responses — that are gradually transforming the management of chronic conditions. At this level, the stakes are not specific to Valais; they concern the whole of modern medicine. But the canton has every interest in actively taking part in these changes rather than simply undergoing them, by partnering with university medical centers (the CHUV in Lausanne in particular) and investing in its Institut Central des Hôpitaux, which represents a valuable lever.

On elderly care, finally, the stakes are more political than technical, because AI touches human dimensions that technology can either strengthen or damage. Well-designed remote monitoring improves safety and preserves autonomy; poorly designed monitoring reduces an elderly person to a bundle of data and deprives them of human presence. The line between these two models is not technical, it is political. And this is precisely where Valais's community institutions — the bourgeoisies, those traditional civic bodies that manage a share of local common lands and collective assets, communal services, local associations — can play a role no centralized solution could replicate. They know their elderly residents. They have the scale and the legitimacy to weave technology and human presence together. What they need is the means to do it.

The valley doctor of tomorrow

The valley doctor of Valais, ten years from now, might look something like this.

They practice in a well-equipped office that combines traditional consultations with remote sessions alongside specialist colleagues at the central hospitals. They have access to reading-assistance tools for routine imaging exams, which lets them handle a markedly larger share of clinical situations on site. Their administrative burden has been cut in half thanks to automatic transcription of consultations and intelligent record management, and the time this frees up goes toward face-to-face clinical care, which remains, as ever, the heart of the job. Around them stands a coordinated team: an advanced-practice nurse who follows up on the valley's chronic patients, a communal medico-social service that handles home visits to the elderly. The whole system runs on infrastructure governed by Swiss law, which guarantees data confidentiality and traceability of decisions.

This practitioner does not exist yet. But they could exist in ten years if the choices that would let this model emerge are made now. These choices involve the canton, the Hôpital du Valais, training institutions, the communes, and probably the Confederation through its support programs for community medicine. None of these actors can bring this model into being alone. But if they converge — and Valais has, when it comes to coordination, a tradition it can draw on — the canton can build a valley medicine that becomes, in turn, a model for others.

A number of initiatives in Valais are already exploring this ground, at the Hôpital du Valais as well as in the labs of the HES-SO Valais and the Idiap Research Institute. They remain scattered, and their maturity varies. But they exist, and they can serve as a foothold for a more ambitious cantonal policy.

On the question of health care in scattered territory, Valais holds a paradoxical advantage. Its geography forces it to innovate faster than better-served cantons. What is, elsewhere, a matter of optimization becomes, here, a matter of continuity. This structural pressure is also a privileged testing ground, where the solutions developed for the canton will serve as an example to other European regions facing the same challenges. A canton that knows how to care for its scattered valleys in the age of AI will have developed skills that do not apply to every setting, but that are valuable anywhere the geographic dispersion of care is a structural constraint — which is to say, across a good part of alpine, Nordic, and mountainous Europe.

This chapter is not a public health program. It offers a strategic reading: health care in the mountains is, I believe, one of the areas where technological change aligns most closely with what Valais has to offer at its best — community institutions, roots in the land, a sense of the long term, a capacity for pragmatic innovation. If the canton knows how to seize it, health care could become, against all expectation, a domain of excellence for Valais in the decade ahead. Not through the large centers, which will remain what they are, but through the quality of community medicine that technology makes possible. It is a wager worth making.