Reflections from a clinician working in Healthtech & AI.
New technology entering clinical practice is met with scrutiny. We ask: Is it safe? Is it effective? Does it reduce harm? Is it cost-effective? Healthcare does not resist innovation because it is hostile to progress. It resists innovation because it is conscious of the risks. Healthcare is a high stakes domain and clinical evidence and patient safety are paramount.
However, there is a structural asymmetry in our perception of risk that should be addressed: We demand quantified, rigorous evidence of improvement from new technologies, while often failing to measure- or even fully understand- the safety, effectiveness and cost of the systems they aim to replace.
I call this the Mirror Effect.
Innovation is required to prove itself. The status quo rarely is.
When it comes to clinical efficacy and safety, new interventions do not need to be perfect. They need to be better than current practice.
This raises a deceptively simple question: what exactly is current practice, and how well does it perform?
In many areas of healthcare, baseline performance is surprisingly opaque. Take triage (our domain at Visiba): we often do not have reliable data on incorrect triage rates, time to appropriate care, duplication of assessment, variation between sites or clinicians, or the cost of delay for current healthcare navigation pathways. Indeed, as a healthcare system we don’t really understand that navigation journeys patients take at all.
When a novel digital triage tool or navigation platform is presented, it needs to demonstrate improvements safety, clinical performance and economic value. Yet the performance of the pathway it seeks to improve remains unquantified.
This is uncomfortable but important to acknowledge. Because improvement can only be meaningfully assessed against a benchmark that is appropriately defined.
Healthcare’s caution has protected patients from premature adoption. History offers examples of technologies introduced with confidence and later withdrawn after harm emerged. Regulation exists precisely because enthusiasm can outrun evidence.
The demand for robust evaluation is not irrational conservatism; it is learned responsibility.
However, the answer to past overreach cannot be unexamined inertia. If we only subject innovation to rigorous evaluation while leaving legacy systems largely unmeasured, we risk protecting tradition rather than patients.
The correct response is not to lower the evidentiary bar for technology. It is to raise the transparency of baseline performance.
Part of this phenomenon is structural. Part of it is psychological.
Healthcare systems, like all human systems, exhibit status quo bias. Risk that is familiar feels less threatening than risk that is new. Harm associated with change feels active and attributable; harm associated with existing complexity feels diffuse and inevitable.
In behavioural terms, we are more sensitive to errors of commission than errors of omission.
When a new technology is introduced, any failure is visible and attributable. When the existing system produces delay, duplication or inconsistency, it is often interpreted as unfortunate but accepted.
Yet inaction carries its own risks. Fragmented navigation confuses patients. Variable triage decisions alter clinical trajectories. Repeated assessments consume workforce capacity. Overtriage brings emergency department overwhelm. Delays compound morbidity. These harms are rarely headline events, but they are real.
The absence of measurement does not equate to the absence of risk.
The Mirror Effect is not an argument that new technology is inherently superior. It is an argument that intellectual honesty requires us to apply equal scrutiny to what already exists.
The Mirror Problem becomes more complex when we consider how digital health creates value.
Technology rarely works as a plug-in replacement for an otherwise unchanged system. Its impact often emerges through pathway redesign. The tool enables different flows of information, different allocation of workforce and different patterns of decision-making.
Consider triage and navigation. Across organisations, triage is conducted through a mixture of receptionist prioritisation, clinician call-back, telephone decision-tree scripts, self-referral processes and informal advice. Navigation relies on patients interpreting fragmented front doors, non-clinical gatekeeping and variable referral criteria.
There is no single, stable baseline. There are multiple local micro-systems.
When digital triage is introduced, improvement does not arise solely from algorithmic accuracy. It may arise from system transformation and new ways of working.
But this creates an evaluation paradox.
You cannot fully demonstrate pathway-level benefit without implementing pathway-level change. Yet traditional evidence hierarchies are designed to isolate and evaluate discrete interventions under controlled conditions. Evaluating the tool in isolation misses the systemic effect; evaluating the whole pathway introduces contextual complexity that challenges classical comparators.
Regulatory frameworks are evolving to accommodate real-world evidence and adaptive evaluation, but our commissioning culture still tends to conceptualise technology as a bounded product rather than a catalyst for redesign.
The Mirror Effect intensifies when the intervention and the system change become inseparable.
Another layer of asymmetry lies in data architecture.
Baseline performance in access and navigation is often distributed across primary care records, secondary care outcomes, call centre logs, urgent care activity and patient self-referral patterns. No single dataset captures the full patient journey.
For an innovator attempting to demonstrate improvement, access to comprehensive baseline data can be limited or impossible. Yet without that baseline, the magnitude of change is difficult to quantify convincingly.
We therefore default to scrutinising the measurable imperfections of the new intervention rather than the unmeasured inefficiencies of the old.
There is opportunity embedded within this challenge.
Regulatory science is increasingly recognising the value of real-world data, continuous monitoring and iterative improvement. Health systems are developing analytics capabilities that allow longitudinal pathway analysis rather than narrow episode-based comparison.
If we take the Mirror Effect seriously, we will invest not only in evaluating new technologies but also in systematically measuring baseline pathway performance before transformation.
Such an approach would improve two things simultaneously. It would create a fair comparator for innovation, and it would illuminate improvement opportunities even where technology is not the answer.
Sometimes holding up the mirror may reveal that change is required regardless of whether a particular product is adopted.
Ultimately, the Mirror Effect is an ethical challenge.
When presented with a novel intervention, it is appropriate to ask whether it is safe enough, accurate enough and cost-effective enough. But ethical responsibility does not stop there.
We must also ask: how safe is our current process? How consistent are our decisions? What harm accumulates through delay and duplication? What variability do we tolerate because it has become normal?
The goal is not to champion technology for its own sake. Nor is it to diminish the importance of evidence. It is to ensure that our evidentiary standards are symmetrical.
Innovation should be tested rigorously. So should the status quo.
In a resource-constrained system facing rising demand and workforce pressure, the greater risk may not always be premature adoption. It may sometimes be unexamined persistence.
Progress in healthcare rarely comes from disruption alone. It comes from constant curiosity- the willingness to question not only what is new, but also what is familiar.
The Mirror Effect asks us to turn our analytical instinct inward as well as outward. Not to defend technology or tradition but to defend patients.
And that requires the courage to look in the mirror.