JMIR Retracts 2020 Diabetes AI Diagnostic Paper

According to the retraction notice posted on the JMIR Medical Informatics website, JMIR Publications Editorial Office is retracting the article "Artificial Intelligence 6Based Neural Network for the Diagnosis of Diabetes: Model Development" by Yue Liu, originally published May 27, 2020 in JMIR Med Inform. The notice cites "identified concerns regarding potential manipulation of the submission process and the integrity of the peer review process," and says the concerns were identified during an investigation of a series of related articles. The notice further states that the corresponding author, Yue Liu, was contacted and offered an opportunity to comment but did not respond to attempts at communication.

The retraction notice itself does not present technical audit details of the original model, data, or code. The published statement focuses on procedural concerns affecting the submission and peer review records rather than on a documented error in model performance or dataset provenance.

Investigations that flag potential manipulation of submission or peer review processes commonly affect multiple papers within a cluster, creating uncertainty about the reliability of the affected literature. For researchers and ML engineers working from medical-AI publications, this pattern increases the value of independent reproduction, access to code and datasets, and conservatively weighting single-study results when informing models or clinical decisions.

Teams that cited or operationalized results from the retracted paper should identify dependent analyses and, where feasible, re-run experiments or substitute validated alternatives. Library managers, systematic reviewers, and regulatory reviewers tracking medical-AI evidence should log the retraction and update citation databases and evidence tables accordingly.

Observers should watch for follow-up detail from the journal about the scope of the investigation, any additional retractions in the cluster, and whether independent audits of the underlying code or data surface reproducibility issues.