Directly informs planned trial arms
The prioritisation output has entered operational planning for AD-SMART, with atomoxetine and metformin named as the initial active arms.
↳ Discussion, Implications for the AD-SMART trial
Abstract
INTRODUCTION: There is an urgent need to identify effective, safe, and affordable treatments for Alzheimer's disease (AD). Funded by the UK National Institute for Health and Care Research (NIHR), we developed a systematic drug prioritization pipeline to identify repurposed drug candidates for inclusion in a planned platform trial for clinical AD.
METHODS: The wider AD community was invited to propose compounds using a standardized proposal template. Fourteen proposals were presented to an international expert panel, who independently ranked compounds based on biological plausibility, preclinical efficacy, safety, and trial feasibility. Extended drug summaries were compiled for shortlisted compounds, supported by ReLiSyR, a machine learning-supported systematic review tool.
RESULTS: Following review in a second panel meeting, independent re-ranking identified the following compounds: atomoxetine (1st), metformin (2nd), isosorbide mononitrate and levetiracetam (joint 3rd).
DISCUSSION: This robust drug prioritization pipeline will speed the identification and progression of promising candidates for therapeutic evaluation.
Key Claims
Primary findingWell-supported
Following review in a second panel meeting, independent re-ranking identified the following compounds: atomoxetine (1st), metformin (2nd), isosorbide mononitrate and levetiracetam (joint 3rd).
the reported ranking follows the stated scoring procedure and is consistent across the Abstract and Results
Abstract, Results and Table 1
SecondaryTentative
The four repurposed drugs identified as having the highest therapeutic potential were: atomoxetine, metformin, and isosorbide mononitrate and levetiracetam.
panel prioritisation with heterogeneous evidence does not establish comparative therapeutic potential or clinical benefit
Discussion, opening paragraph
Derived from the full evaluation — not a separate score.
Strengths · Limitations
Strengths
Traceable multi-stage selection process
The paper documents the progression from 14 proposals to 10 shortlisted compounds, eight fully considered compounds, and four highest-ranked candidates, alongside explicit ranking rules.
↳ Methods, Final decision-making process; Results
Multiple evidence and stakeholder inputs
Drug proposals, clinical and regulatory sources, ReLiSyR-supported review, manual searches, feasibility assessment, expert ranking, and PPIE are combined in the decision process.
↳ Methods, Evaluation of efficacy and safety data; Machine learning-assisted literature search; PPIE
Limitations
Pipeline performance remains unvalidated
No comparator, sensitivity analysis, reproducibility test, or downstream performance measure shows that the process is robust or faster than alternative prioritisation approaches.
↳ Abstract, Discussion sentence; Methods; Discussion, Strength of the drug selection process
Scalability claim conflicts with limitations
The Conclusion describes a scalable pipeline, while the limitations state that the process remains time-consuming and may not scale to more compounds or meetings.
↳ Discussion, Limitations and challenges; Conclusion
Ranking language exceeds demonstrated evidence
The process establishes an expert-panel ranking for a specific trial, not comparative therapeutic potential or clinical efficacy in AD.
↳ Discussion, opening paragraph; Methods, Final decision-making process
Nabu Commentary
The paper's strongest feature is its transparent account of a real selection exercise, including explicit thresholds, two ranking rounds, evidence-supported drug CVs, and a concrete link to AD-SMART arm selection. Its contribution is nevertheless an incremental integration of established Delphi-style, MAMS, ReLiSyR, and PPIE approaches. The main score constraint is the move from describing one ranking exercise to claiming a robust, reproducible, accelerating, and scalable pipeline without comparative validation. The contradiction between the scalability claim and the stated time burden further limits the contextual assessment.
Check another paper
Author Note / Field Feedback
Nabu’s assessment, alongside the field’s view.
Are you an author of this paper?
Dimension breakdown
Study Quality
Limited2.8
Confidence mediumContribution
3.1
The paper integrates established expert-ranking, living-review, feasibility, and PPIE elements for an AD-specific platform-trial decision. The advance is operational and locally useful but incremental rather than a demonstrated improvement over existing prioritisation approaches.
Background; Methods, Preliminary work and Final decision-making process
our pipeline differed in key aspects
Methodological Rigour
2.5
The process is bounded, uses multiple evidence streams, and provides reconstructable ranking rules. One application without benchmarking, sensitivity analysis, or reproducibility testing cannot establish the broader methodological performance claimed, so the serious overgeneralisation ceiling binds.
Methods, Machine learning-assisted literature search and Final decision-making process; Abstract, Discussion sentence
This robust drug prioritization pipeline will speed the identification and progression
Reporting
3.4
The progression from proposals through shortlisting and final ranking is easy to follow. Precision is reduced where a panel ranking becomes a claim about therapeutic potential and where the Abstract predicts acceleration without comparative evidence.
Results; Abstract, Discussion sentence; Discussion, opening paragraph
identified as having the highest therapeutic potential
Positioning
2.8
The paper situates the process among Delphi methods, MAMS trials, ReLiSyR, and related neurodegenerative initiatives, and it traces several limitations to likely effects. Its scalability conclusion nevertheless conflicts with its acknowledgment that the process is time-consuming and may not scale.
Background; Discussion, Limitations and challenges; Conclusion
remains time-consuming and may limit the scalability
Reliability
Caveats4 of 4 checks
Internal CoherenceNoted
The reported ranking process and numerical progression are internally consistent, but broad claims about robustness, acceleration, bias mitigation, and scalability exceed or conflict with the evidence from one application. This affects interpretation of the pipeline rather than the reported ranking itself.
Research ConductClean
Consent requirements and conflicts are explicitly addressed, and the reported process does not present participant-level outcomes or an intervention. No text-visible conduct concern was identified.
Flags: 1 declared / 5 total
Reference IntegrityClean
50 references in manuscript 43 of 43 checkable references found in an index 7 are books, websites or datasets — counted, but not index-checkable
Post-Publication RecordClean
No retraction notice found in Retraction Watch.
Last checked: 16 Oct 2026
Sources: Retraction Watch ✓
Readiness
Where this paper’s evidence sits on the path from initial observation to real-world use.
Validated
The pipeline has moved beyond a proposal by producing ranked candidates and informing operational trial planning. Readiness remains incomplete because neither the process benefit nor the selected drugs' efficacy in AD has been tested in AD-SMART.
Results; Discussion, Implications for the AD-SMART trial
the initial active trial arms will be atomoxetine and metformin
AI-generated, human-governed. Something look off? Contact us to request a review.