On 29 October 2026, Doximity launched DoseCheck, a medication dose-verification feature within Doximity Ask. It adds an automated comparison between dosing claims in an AI answer and a structured medicines database. Doximity says it is free for Ask users on mobile and desktop. Doximity launch announcement.
The interesting development is the division of labour. An AI system can assemble an explanation; a separate checking process can test a specific factual claim. For clinicians, the useful question becomes: which part of this medication answer has been checked, against which source, and for which patient circumstances?
That development is directly relevant to iatroX’s partnership with Datapharm, which brings emc medicines information into the iatroX clinical AI workflow. Both developments put maintained medicines data closer to the answer. Understanding how that data is used helps clinicians get more value from the tools.
What DoseCheck actually does
Doximity describes a process that identifies the medicine, dose, route, frequency and relevant indication in an answer, then compares the claim with its structured drug database. A matching dose receives a green verification badge and a link to its source. The company describes the comparison as deterministic: repeating the same comparison produces the same result. Doximity’s product explanation.
This changes what a reader can inspect. A citation at the end of a paragraph offers a document to read. A dose-check result makes a more specific claim about an extracted piece of that paragraph. The source link then provides a way to examine the surrounding instructions and qualifications.
Consider an illustrative answer that mentions three regimens: starting treatment, increasing treatment and switching from a previous formulation. A reader needs to know whether verification applies to each regimen individually. A badge next to one statement should be read in relation to that statement, with the rest of the answer assessed on its own terms. This is a useful reading habit for any clinical interface that combines narrative answers with factual verification.
For a closer look at the reported accuracy figures and the difference between claims and complete answers, read our DoseCheck evidence analysis.
Why a dose is more than a number
A numerical amount can be faithfully copied while the recommendation around it is wrong. The same medicine may have several indications, formulations, treatment phases and patient-specific adjustments. Verification becomes meaningful when those conditions travel with the number.
The FDA’s explanation of prescribing information places recommended dosage, route, titration, interaction-related changes and dosing in specific populations within the Dosage and Administration section. Its structure reflects the fact that these belong together. FDA prescribing-label FAQ.
A practical way to read a medication claim is to reconstruct the complete instruction:
- Product: which active ingredient, strength and formulation does the answer mean?
- Purpose: which indication and stage of treatment are being discussed?
- Administration: what are the amount, unit, route, interval and relevant duration?
- Patient conditions: which characteristics determine whether this regimen applies?
- Linked decisions: what interactions, monitoring or preparation requirements accompany it?
- Source: does the linked passage support this combination of details in the relevant jurisdiction?
These are reading questions for the clinician. They are not a list of capabilities independently demonstrated for DoseCheck. They help turn a verification signal into a useful check of the actual clinical statement.
A worked example: the condition that disappears
Imagine a teaching vignette about a medicine with separate instructions for two levels of kidney function. The question supplies the patient’s renal information. An AI answer correctly repeats the medicine name and a regimen from the product information, but selects the branch intended for a different patient group.
Someone checking only whether that regimen appears somewhere in the document could find a match. Someone checking the full conditional statement would ask why that branch was selected. The learning point is condition matching: the source must support the recommendation for the circumstances described.
Now change the vignette. The renal information is missing. The productive next step is to identify the missing information and how it affects the decision. A fluent answer that simply supplies a regimen has skipped that step. For an AI tool, a useful response would make the dependency visible; for a learner, recognising the dependency is part of the answer.
This example also suggests a better question to ask a clinical assistant: “Which patient characteristics determine the dosing branch, and where does the source explain them?” That prompt makes the decision structure easier to inspect. It also leaves a clear learning objective when a detail is unfamiliar.
How to use a verification badge well
Start with the sentence being verified. Check that it refers to the intended product, indication and patient group. Open the linked source at the relevant passage, then look for conditions that could change the interpretation. Finally, connect the dosing information with the rest of the treatment decision.
The FDA makes a related point about its own label summaries: Highlights provide a concise overview, but should not be used as a stand-alone document. That is a useful analogy for an AI interface. A compact signal is most valuable when the full information is easy to reach. FDA guide to using prescription drug labeling.
For teams considering a new medication tool, the demonstration should include an ordinary case, a case where one patient characteristic changes the answer, and a case where the necessary information is absent. Observe whether the interface makes those differences visible. This reveals more about day-to-day usefulness than repeatedly asking an uncomplicated lookup question.
A further test is to ask about two formulations of the same active ingredient. Does the answer keep their instructions separate? The aim is to see how clearly the product communicates the boundary of a check and helps a reader resolve uncertainty.
Where iatroX’s Datapharm partnership fits
Datapharm has publicly confirmed its partnership with iatroX, enabling access to emc medicines information within the clinical AI workflow. emc is Datapharm’s UK medicines-information resource, containing approved product information such as Summaries of Product Characteristics, or SmPCs. Datapharm partnership announcement, about emc.
For UK medicines questions, that connection matters. It gives the workflow access to product-specific source material covering the conditions that make a recommendation applicable. The existing iatroX–Datapharm announcement explains the integration; iatroX Medicines provides concise guides with source links for further scrutiny.
The relationship to DoseCheck is clear, while the mechanisms differ. Datapharm integration supplies authoritative UK medicines material for source-grounded answers. DoseCheck is an additional automated comparison of extracted dosing claims with structured reference data. The partnership should be assessed for the sources it makes available and how clearly an answer uses them; the checker should also be assessed for the claims it verifies and the cases it can handle.
This distinction helps clinicians choose the right workflow. A question about a UK product’s precautions calls for the relevant UK product information. A verification badge on a dosing claim answers a narrower question about the matched statement. Both can contribute to a more inspectable medication answer.
Our DoseCheck and iatroX–Datapharm comparison explores the connection through UK examples involving indication, formulation and brand identity.
The link to question banks and CPD
Medication checking becomes more valuable when the clinician can explain why a detail matters. A question bank can reveal whether a learner notices the formulation, interaction or patient characteristic that changes an answer. Reviewing the explanation can then turn a missed detail into a specific rule to apply next time.
On iatroX, the exam catalogue is the starting point for revision aligned with the intended assessment. The Socratic Tutor supports discussion of attempted questions. For practice-led learning, CPD provides a route from a topic to a targeted session and a reflection the clinician reviews and finalises.
A useful learning record would say, for example: “I noticed that I was checking the dose number before confirming the formulation. I practised separating those decisions and will check the product identity first in future reviews.” That records a change in reasoning that can be revisited through another question or discussion.
Our three medicines exercises for revision and CPD provide a practical starting point. DoseCheck’s launch gives this learning a timely prompt. The skill to practise is recognising which conditions make a medicines statement true, then finding the source that supports them.
Frequently asked questions
Is Doximity’s new medication checker called DoseCheck?
Yes. Doximity announced DoseCheck on 29 October 2026 as a free additional verification feature in Doximity Ask. The launch announcement describes availability on mobile and desktop.
Does a verified dose establish that the whole treatment plan is appropriate?
Read the verification against the specific claim and its source. The complete decision also involves the indication, product, patient circumstances, other medicines and monitoring requirements. The worked example above shows why matching those conditions matters.
Is iatroX’s Datapharm partnership the same feature?
It supports a related aim through a different mechanism. The partnership brings emc medicines information into iatroX’s clinical AI workflow. DoseCheck performs automated checks of extracted dosing claims against structured reference data.
Put the reasoning into practice
Start a medicines-focused iatroX CPD session, practise the decisions through questions and Tutor, then review and finalise a reflection on what you will apply next time. CPD is included in iatroX all-access.