Europe spends close to €600 billion each year on mental health. That number should suggest a system functioning at scale, resourced and prepared to meet the challenge. Yet outcomes tell a very different story. Relapse rates for conditions like bipolar disorder can reach 90% over a lifetime, and rehospitalisation rates for schizophrenia exceed 70% within five years. With care still defined by long gaps between appointments, missed warning signs, and interventions that arrive too late.
The problem is a fundamental misalignment between the nature of mental illness and the design of the systems meant to treat it.
A system built for crisis, not prevention
Mental health conditions do not arrive suddenly. They evolve. Subtle shifts in behaviour, cognition, and mood can signal deterioration weeks or even months before a patient reaches crisis. Yet European healthcare systems remain structured around episodic assessments: a consultation every few weeks, a questionnaire in a waiting room, a clinician reconstructing a patient’s recent experience from memory and self-report alone.
The result is a system that absorbs enormous cost managing relapse rather than preventing it. The bulk of mental health expenditure is concentrated downstream, in hospitalisation, crisis response, and long-term disability, rather than in the early intervention that might have changed the trajectory. Clinicians are having to work under serious capacity constraints, without the tools to monitor patients consistently between consultations. This is absolutely not a criticism of clinicians. They are operating within a model that has not fundamentally changed in decades.
What AI can see that we cannot
Speech is one of the most information-rich and accessible indicators of mental health: approximately 30 seconds of recorded speech can yield clinically relevant signals across psychosis, mood disorders, and cognitive decline. By evaluating not just what is said, but the coherence and structure of language, its rhythm and rate, the way speech holds together or begins to fragment. You are able to pick up signals that are easy to miss in an infrequent clinical consultation.
By providing clinicians with objective and continuous data between appointments AI can introduce an entirely new layer of insight into care. Early warning signs can be flagged before a crisis emerges and interventions can be timed more precisely.
The same continuous signal also has significant implications for drug development: CNS trials have long been constrained by subjective, infrequent endpoints, and remotely captured speech data opens the possibility of more sensitive trials and faster treatment decisions.
Scaling responsibly
Building an AI system that works clinically is one challenge. Scaling it responsibly across European healthcare systems is another. Trust is the foundation: mental health data is among the most sensitive there is, and clinicians rightly require strong evidence before integrating AI into decision-making. The answer is not to shortcut that process, but to invest in genuine clinical validation and transparent design. Encouragingly, European healthcare is already ahead of most sectors in AI governance, with 36% of organisations having a formal AI strategy and significantly higher adoption of responsible AI frameworks than the cross-industry average.
Infrastructure is equally critical. Handling sensitive health data demands secure cloud environments that meet the compliance requirements of regulated healthcare across multiple jurisdictions. This is precisely why working with partners like AWS play a meaningful role, providing not just technical infrastructure, but guidance on data security, compliance standards, and the architectural decisions that determine whether an innovation can actually move from pilot to deployment at scale. For an early-stage company navigating European regulatory complexity, that kind of structural support is not peripheral to the work. It is the work.
The regulatory environment also needs to evolve. Europe has rightly prioritised responsible AI, and frameworks like the EU AI Act reflect genuine values. But when compliance consumes a disproportionate share of technology budgets, and when a startup operating across multiple markets must navigate fragmented national interpretations of shared frameworks, momentum stalls. Harmonising implementation so that responsible compliance in one country applies across the single market would make a material difference; not by lowering standards, but by removing unnecessary complexity.
The needed change
AI does not and should never replace clinicians. AI tools are not a diagnosis. But they do have capability that extends what a clinician can perceive and create the conditions for earlier, more informed decisions. The goal is to give clinicians the means to act before a crisis.
Achieving the shift from reactive treatment to continuous, preventative care requires more than good technology. It requires healthcare systems willing to invest upstream, policymakers prepared to enable responsible innovation, and a clinical community genuinely engaged with the evidence. It also requires the kind of ecosystem support that helps responsible innovations reach patients rather than stall in procurement. Programmes like the AWS Pioneers Project are a practical example of how the private sector can help close that gap. Europe is spending €600 billion on mental health. The question is not whether we can afford to change how that money works. It is whether we can afford not to.
