BGPT: Author Review: Shorena Janelidze

Explore by Goal

Quick Explanation Copied

Scientific strength snapshot (Shorena Janelidze): Evidence from high-impact biomarker work suggests strong skills in diagnostic/prognostic validation using multimodal cohorts, plus statistical modeling (ROC/logistic/Cox) in neurodegeneration/psychiatry; key scientific risks are assay generalizability, confounding, and replication across independent clinical settings.

Long Explanation

Author Review — Shorena Janelidze

Skeptical, evidence-based critique focused on biological/scientific merit.

Citation/impact signals provided

h-index: 3 (from provided author snapshot) and also OpenAlex shows alternative profiles with much higher h-index values (data suggests name-matching/identity ambiguity risk).
Total citations: 129 (from provided snapshot), while OpenAlex matches include higher citation counts for an ORCID-matched profile.
Works count: 4 (provided list) vs OpenAlex showing many more works for a matched ORCID profile.

Note: The platform-level metrics may reflect author identity disambiguation. That’s a known failure mode when evaluating individuals.

Figures built from extracted quantitative results (from provided paper data)

Most informative example papers (why they matter scientifically)

1) Multimodal AD biomarker validation using plasma p-tau181

The Nature Medicine study evaluates plasma p-tau181 for (i) tracking relation to CSF tau, (ii) differentiation of AD dementia vs non-AD neurodegenerative disorders, and (iii) prediction of progression to AD dementia, using multiple cohorts plus imaging/neuropathology context. The authors report strong diagnostic separation (e.g., AUC values) and meaningful progression hazard ratios (Cox-type modeling), while also emphasizing assay/model limitations (e.g., research-grade assay not yet established for primary-care general use).

Quantitative extracted values used in the Plotly figures come from:

Strength: Multicohort evaluation with multimodal anchors (CSF/PET/neuropathology), enabling stronger causal-*direction* plausibility than single-modality studies (still observational/associational).
Risk: External validity depends on assay standardization and cohort representativeness; the provided extracted limitations explicitly flag research-grade assay status and replication in primary-care settings.

2) Confounding mitigation: CKD and phosphorylated-to-total tau ratios

The JAMA Neurology study addresses a major biomarker problem: comorbidity (kidney dysfunction) can shift circulating tau measurements and confound Alzheimer’s interpretation. It tests whether using phosphorylated-to-total tau ratios (notably pT217/T217) reduces CKD-related variability relative to individual phosphorylated measures, and evaluates associations across memory-clinic cohorts (including replication).

Strength: This is “biomarker epistemology” done well: it directly targets a plausible physiological confound (clearance/renal dysfunction effects on circulating proteins) rather than assuming specificity.
Risk: Even with covariate adjustment, residual confounding can remain; extracted limitations include secondary-clinic recruitment and limited CKD+ subgroup sizes.

3) Biomarker comparison and performance ranking: p-tau217 vs p-tau181

In Nature Communications, p-tau217 is compared to p-tau181 as an Alzheimer’s biomarker using cerebrospinal fluid performance criteria, with the study reporting that p-tau217 performs better than p-tau181 under their evaluation framework.

Strength: Performance comparisons are essential for biomarker “choice under uncertainty”; comparative tests reduce the risk of over-investing in a single marker.

4) Psychiatry-adjacent immuno-CSF signal: IL-6 in suicide attempters

The provided OpenAlex citation list includes an early Biological Psychiatry paper where IL-6 in CSF is reported elevated in suicide attempters and related to symptom severity. This indicates the author has worked in cytokine/CSF quantitative phenotyping rather than only AD imaging.

Risk to keep in mind: Cytokine biomarkers can be affected by many systemic factors (infection, stress, timing, medications). Without detailed designs and controls, effect sizes can be unstable across cohorts; this is a general vulnerability for immuno-psychiatry CSF work, so independent replication is crucial. (No additional quantitative limitations were provided in the prompt.)

Skeptical evaluation: where the evidence can break

Assay/platform generalizability: Even if correlations/ROC metrics look strong, they may not translate across labs/platforms without calibration and reference standards; the provided extracted limitations for plasma p-tau181 explicitly mention assay validation/generalizability needs.
Subgroup/sample-size instability: Diagnostic metrics can overestimate performance when strata are small or when the evaluation set is not representative; the provided extracted limitations mention small subgroups in some strata.
Confounding and causal direction: Plasma biomarker elevation is associative; “tracks central pathology” is biologically plausible but still requires stringent longitudinal design and controlling comorbid physiology. The CKD-ratio study is a direct attempt to address confounding.
Industry/COI risk: Biomarker work often intersects with pharma diagnostics and radiotracers. The prompt includes detailed competing interest statements for the Nature Medicine plasma p-tau181 paper, indicating consult/advisory relationships by several coauthors and industry funding/sponsorship lines. This does not prove bias, but it increases the need for independent replication and transparent methods.

Overall scientific appraisal (evidence-weighted)

What looks strong: The author’s visible work (from the provided examples) clusters around biomarker measurement tied to underlying biology (tau/amyloid dynamics) and around quantitative validation with diagnostic discrimination and longitudinal prediction. The p-tau181 work emphasizes discrimination and progression with modeling approaches and multimodal anchors.

What remains uncertain/fragile: Many biomarker headlines are sensitive to cohort selection, assay platform, and handling of missing data/exclusions; effect sizes can shrink in less-selected real-world populations. The CKD-ratio strategy is a good sign of methodological maturity for confounding, but the provided extracted limitations still indicate replication and generalizability work remains.

Confidence in this appraisal: moderate. Reason: this review is based on the subset of provided author papers and extracted numeric results; it may under- or over-estimate overall capability without inspecting the full publication record and methods section of each work.

What to verify next (high-yield skepticism)

Check whether the paper’s reported performance metrics hold when re-evaluated using alternative cutoffs and when the evaluation is restricted to pre-specified external cohorts (not same-study tuning).
Evaluate whether the CKD-ratio logic is stable across broader renal phenotypes and whether ratios remain robust under different total-tau measurement noise assumptions.

Feedback:

Updated: April 27, 2026