Detection Report on "Longitudinal Changes in Functional Connectivity of the Caudate Is Associated With Recovery From Bell's Palsy" (Frontiers in Aging Neuroscience, 2019)

Academic fraud report · Geng Detector

Summary

This report assesses four potential integrity concerns in the 2019 paper by Hu et al. (DOI: 10.3389/fnagi.2019.00295) on functional connectivity changes during Bell's palsy recovery. The verdict is 'Questionable' (yellow). Key findings include: (1) a clear copy-paste error in the Table 2 abbreviation list, where both laterality columns read 'left' instead of 'left/right'; (2) a suspiciously identical mean AMEPs value (1.94) for the Mandibular branch at two different clinical timepoints (pre-treatment and mid-treatment) with different standard deviations, which is statistically uncommon for independent physiological measurements; (3) a corroborating internal consistency between reported correlation coefficients and p-values at n=21 (df=19), suggesting genuine computation rather than fabricated statistics; (4) insufficient image-level data to assess potential manipulation in the fMRI figures. Confidence is high for textual/typographical issues and moderate for the data coincidence claim. No definitive fraud determination is made; raw patient-level data are recommended for verification.

Verdict

Questionable (🟡) — The paper exhibits a clear low-level typographical error and a coincidental data pattern that warrant author clarification, but no conclusive evidence of intentional fraud has been established. The statistical computations appear internally consistent.

Key findings

Low-level typographical error in Table 2 abbreviation list — Both laterality abbreviations are listed as "L, left" and "R, left" instead of "L, left; R, right", indicating careless proofreading.
Identical mean AMEPs across two timepoints — Mandibular branch motor evoked potential amplitude is reported as $1.94 \pm 0.88$ (pre-treatment, PT) and $1.94 \pm 0.64$ (mid-treatment, MT). Identical means at two independent clinical assessments with differing standard deviations is uncommon and raises questions.
Internal statistical consistency — Reported correlations (e.g., $r=0.64, p=0.0018$; $r=0.46, p=0.0367$; $r=0.056, p=0.808$) are mathematically consistent with $df=19$ ($n=21$), supporting that genuine statistical analysis was performed.
Image analysis not feasible — The review text contained no pixel-level image data, so potential manipulation of fMRI connectivity maps and bar charts (Figures 1–6) could not be evaluated.

Evidence highlights

Table 2 abbreviation note: "Abbreviation: MI, precentral gyrus; SI, postcentral gyrus; STG, superior temporal gyrus; TH, thalamus; MFG, medial frontal gyrus; L, left; R, left." — verbatim from the PDF.
Table 1 (electroneurography comparison): Mandibular branch AMEPs values: PT = $1.94 \pm 0.88$; MT = $1.94 \pm 0.64$.
Correlation examples cited in Discussion/Figure 6 context: $r=0.64, p=0.0018$; $r=0.46, p=0.0367$; $r=0.056, p=0.808$, all internally consistent with two-tailed tests at $n=21$ ($df=19$).
DOI: 10.3389/fnagi.2019.00295.

Notes

The "R, left" error is most plausibly a proofreading failure rather than deliberate fabrication, but its severity is amplified by the AMEPs coincidence, raising concerns about overall manuscript quality control.
The AMEPs coincidence could alternatively reflect rounding to two decimals; raw per-patient values are needed to adjudicate.
Image manipulation cannot be ruled in or out without access to original Figure files; fMRI analyses additionally depend on preprocessing pipelines that should be examined.
Recommended actions: request raw patient-level AMEPs data from authors; file a PubPeer comment requesting an erratum for the laterality typo; consider notifying the journal editorial office if the data coincidence is not satisfactorily explained.