JAMA Network Open Corrects Disclosures in Amyloid Imaging AI Study
JAMA Network Open has issued a correction to an Original Investigation on artificial intelligence software for automated detection and quantification of amyloid-related imaging abnormalities. The update addresses two separate problems in the published article: missing conflict of interest disclosures and a wording error in the Methods section. Both corrections matter because the paper concerns a commercially developed tool and named investigators with financial ties to the companies involved.
The disclosure section had been inadvertently deleted during production and has now been restored. The corrected language states that Dr Hughes reported personal fees and employment from Biogen during the conduct of the study and personal fees from Biogen outside the submitted work. Dr Huijbers reported personal fees from Biogen during the conduct of the study and personal fees from Philips Research outside the submitted work. Dr Curiale reported being an employee and shareholder of Biogen during the conduct of the study. Dr Belachew reported being an employee and shareholder of Biogen during the conduct of the study. Dr Van Hecke reported being founder of icometrix and shareholder outside the submitted work. Dr Smeets reported being an employee and shareholder of icometrix, which is commercializing the software that is the scope of the publication. No other disclosures were reported.
The Methods section was also revised to make clear that icometrix is the maker of the icobrain aria tool.
Why the Disclosure Record Matters in a Commercial AI Study
In research on software intended to detect and quantify radiologic abnormalities, the conflict of interest record is not a routine administrative detail. It is part of the scientific substrate. Readers need to know whether investigators had employment, equity, or fee relationships with the manufacturer of the software under evaluation, because those ties can shape study design choices, software development priorities, interpretation of performance, and the boundaries of outside validation.
The corrected disclosures show multiple relationships relevant to the study’s commercial ecosystem. Several investigators reported ties to Biogen, while Dr Van Hecke and Dr Smeets reported ties to icometrix. In a manuscript centered on an imaging algorithm with a defined commercial pathway, those facts should be visible from the first reading. Their absence during publication created a gap in the transparency record even though the correction now restores it.
A single missing paragraph can alter how a paper is read.
That is especially true when the article evaluates an AI-assisted medical tool rather than a purely academic algorithm with no commercial sponsor footprint. Software used for detection and quantification in neuroimaging often sits close to clinical deployment, regulatory scrutiny, and downstream integration into trial workflows. For that reason, full disclosure is not decorative. It is part of how clinicians, radiologists, trialists, and journal readers judge the independence of the evidence.
Clarifying Who Built the Software
The Methods correction is shorter but equally important. The revised text clarifies that icometrix is the maker of the icobrain aria tool. In technical studies, manufacturer attribution helps define provenance. It tells readers who owns the algorithm, who controls the product, and whose engineering pipeline underlies the software being tested.
That clarification has practical consequences. If a software tool is being studied for automated detection of amyloid-related imaging abnormalities, the identity of the maker informs questions about reproducibility, external validation, and future clinical uptake. It also helps separate a commercially produced platform from a generic research prototype. When the manufacturer is explicitly named, readers can better interpret whether the work is an independent assessment, a company-linked evaluation, or a hybrid arrangement involving collaborators with direct financial interests.
The corrected wording does not change the underlying clinical subject of the study. It does change the paper’s transparency profile, which is central to how biomedical research is weighed in practice. In a field where imaging algorithms are increasingly used to support trial monitoring and potential clinical decision-making, even a small textual omission can leave an outsized imprint on credibility.
Transparent Reporting Shapes Interpretation of Imaging AI Evidence
Amyloid-related imaging abnormalities are clinically important because they can arise in the setting of therapies and studies involving brain imaging surveillance. Tools designed to detect and quantify these findings aim to improve consistency, speed, and scalability across imaging review. When those tools are commercial products, journals and readers need a clean account of who developed them and who benefited financially from their use or evaluation.
The corrected article now aligns its Methods and disclosure language with that expectation. The publication’s scientific content may remain intact, but the corrected transparency record changes how the work should be read. A reader assessing the validity of an AI-assisted imaging study must be able to trace both the software lineage and the investigators’ financial relationships. Without that information, the evidence base can appear cleaner than it actually is.
The correction also reinforces a broader fact about modern biomedical publishing: rigor extends beyond algorithms, images, and outcome metrics. It includes the provenance of the software, the ownership structure around the product, and the possibility that commercial relationships may intersect with the work. In a study of an imaging tool with clinical applications, those details belong in the main article, not in the background notes.
This correction now places them where they should have been from the start. The article has been amended, and the journal has restored the missing disclosure text while clarifying the manufacturer relationship in the Methods section. Readers evaluating the study should now do so with the corrected record in view, especially because the paper concerns a tool marketed by a company with direct ties to several named investigators.