App activity can support a learner's own decisions, but it can also become a misleading measure of someone else's performance. A clinician may use a tool less because the task has become easier, because they prefer another format or because access is difficult. A login count cannot distinguish those explanations on its own.
This article examines a proposed monitoring risk, not an allegation about iatroX or another provider. The central question is what a metric can legitimately establish and whether its collection and use match the purpose agreed with the people concerned.
Distinguish self-monitoring from employer monitoring
A learner choosing to review their own question history is doing something different from an employer inspecting that history to judge diligence or competence. The information may look similar, but the purpose, relationship and consequences differ.
The ICO's monitoring-workers guidance, checked on 3 October 2026, addresses the employment context. The ICO states that this guidance is under review following the Data (Use and Access) Act, so organisations should check its current status when designing a real deployment.
A learning platform should not be assessed only by what its dashboard can display. Ask who can see the data, what decisions it may influence and whether the learner understands those arrangements.
A tool introduced to support an individual should not quietly acquire a different purpose simply because activity data are available.
What the common metrics can and cannot answer
The table below is an analytical framework, not a set of observed iatroX results.
| Measure | What it may help describe | What it does not establish alone |
|---|---|---|
| Logins | Recorded access to an account | Motivation, learning quality or clinical competence |
| Time in the application | Time recorded within the interface | Attention, effort or whether a longer session was useful |
| Questions attempted | Volume of recorded practice | Understanding, breadth or the quality of review |
| Correct-answer rate | Performance on the questions attempted | Independent performance on a different assessment or in practice |
| Records exported | Completion of an export action | Truthfulness, accreditation or benefit to patients |
| Return after an error | A pattern of further recorded activity | Whether the misconception was actually resolved |
These limits do not make the metrics useless. They make context necessary. A question history may help a learner choose what to revisit without becoming a valid basis for ranking clinicians.
The unit should remain visible. Minutes, questions, sessions and completed records should not be merged into a single score that appears to measure professional capability without evidence that it does so.
A fictional example of a misleading dashboard
Imagine two fictional clinicians completing the same learning objective. One reads an explanation online and discusses it with a colleague. The other spends longer in the application, repeatedly opening the same material because the navigation is unclear.
A dashboard that rewards time in the app could rank the second clinician higher while missing both the first clinician's discussion and the second clinician's access difficulty. Neither conclusion would follow from the data alone.
For a neurodivergent learner using a different format or an agreed support arrangement, this mismatch may be particularly important. The appropriate response is to understand the activity and barrier, not infer effort from one interface's records.
This scenario is hypothetical. It shows why an evaluation should be designed around the actual learning question before selecting the easiest available metric.
Consent and power need careful treatment
The ICO's detailed monitoring guidance, checked on 3 October 2026, explains that consent is not usually an appropriate basis in employment because of the power imbalance. It also addresses transparency, purpose and assessment of high-risk processing.
An employee agreeing to use a helpful tool should therefore not be treated as resolving every question about monitoring. The organisation must establish the appropriate lawful arrangements and explain the actual use of the information.
For a proposed support programme, ask what happens if the clinician prefers a different tool, uses an accessible alternative or does not want optional analytics shared. Those questions should be addressed before the programme is presented as voluntary.
A meaningful support discussion should allow the person to describe what helps without requiring unnecessary disclosure to everyone who can access a dashboard.
Measure the support goal directly
If the purpose is to make learning easier to resume after interruption, evaluate whether the learner can identify the next action and continue without reconstructing the whole session. If the purpose is clearer explanations, assess understanding of the relevant concept.
If the purpose is professional evidence, check whether the record accurately describes a real activity and can be retained appropriately. Do not substitute the number of generated documents for their usefulness.
These are proposed evaluation questions. They can be explored through a bounded task, the learner's account and appropriate educational assessment, rather than continuous surveillance of every click.
A support review should also identify unintended burden. Requiring detailed activity tracking may add exactly the administrative work the intervention was meant to reduce.
Apply the same standard to iatroX
According to iatroX's September 2026 product information, its learning tools include question practice, guided Tutor discussions and reviewed CPD records. These activities may generate information useful to the learner, but their presence does not establish a validated measure of clinical performance.
A learner can use an error pattern to choose a topic for review. That is different from an employer concluding that a person who completes fewer questions is less competent. A CPD export can preserve evidence without proving that the activity changed patient outcomes.
Any institutional use should therefore specify the data, access arrangements and intended decisions rather than assuming that more visibility is necessarily beneficial. This principle applies to the publisher's product as much as to any alternative.
Keep learning support and performance assessment distinguishable
Professional performance may legitimately need assessment through appropriate processes. The issue is not whether assessment should exist, but whether the chosen evidence supports the conclusion being drawn.
A formative learning environment should make it possible to expose uncertainty and practise mistakes. If every exploratory error is treated as a performance signal, users may reasonably change how they engage. That possibility should be investigated rather than assumed away.
Agree what the data are for, what they are not for and how a person can challenge an inaccurate interpretation. Evaluate relevant outcomes and preserve context. A supportive system should help people learn, not merely make their digital activity easier to count.
Frequently asked questions
Are learning analytics inherently surveillance?
No. Their significance depends on what is collected, who uses it, the purpose and the consequences; voluntary personal review differs from employer monitoring.
Can question-bank activity measure clinical competence?
Not by itself. It describes a particular educational activity and requires appropriate evidence before being used to support broader conclusions.
Is employee agreement enough to justify any monitoring?
No. Organisations need to establish the relevant lawful arrangements and safeguards, rather than treating a general agreement to use software as permission for every subsequent use of its data.
Explore learning focused on reasoning rather than activity counts →