▤
0
Active studies
Editable evidence registry
RESEARCH INTELLIGENCE · TRANSPARENT SYNTHESIS
From scattered studies
to clearer decisions.
Organize research, expose evidence gaps and test how your conclusions change when screening criteria shift.
0Screened studies
Demonstration workspace: every preloaded study is fictitious. Scores reflect user-entered judgments, not verified scientific findings, effect sizes, evidence grading, or a formal meta-analysis.
◈
0
Weighted directional balance
Index from −100 to +100
▥
0%
Identifier coverage
DOI or source URL provided
⚑
0
Integrity flags
Record quality prompts
Evidence at a glance
Distribution and balance based on your active records.
Direction of findings
Share of weighted, user-assigned judgments
Weights = quality × relevance. Equal method treatment; no inferential statistical meaning.
Research integrity prompts
Potential gaps to inspect, not automatic validity judgments
Evidence map by research theme
Explore where themes show agreement, conflict or uncertainty.
Directional distribution across themes
Counts of studies classified as supportive, mixed or contradictory
SupportsMixedContradicts
Study evidence library
Curate, inspect, archive and restore evidence records.
Scores and direction labels are subjective annotations. An identifier does not independently verify a publication; check sources and conduct formal appraisal before drawing scientific conclusions.
Active records
Editable evidence inventory
Archived studies
Preserved locally, excluded from active calculations.
Sensitivity & screening lab
Test how the evidence balance changes under transparent inclusion rules.
The lab filters the same annotated records. It does not search publications, assess causal relationships or estimate effect sizes. Screening thresholds are exploratory, not methodological recommendations.
Screening criteria
Adjust thresholds for an illustrative sensitivity check
Screening results
Live comparison with all active studies
Included studies0 / 0
Screened directional index0
Change versus full library0
Weighted directional shares0 / 0 / 0
Threshold stress test
How stable is the directional index when minimum quality changes?
Quality-threshold sensitivity
Same year, relevance, method and theme settings
Included sample
Records currently passing all screening criteria
Research synthesis brief
Reproducible descriptive summary generated from active records.
Not a systematic review, meta-analysis, citation verification or automated literature search. Review the wording, full texts and methodological context before using any generated summary in academic work.
Evidence summary
Rule-based text, ready for human review and editing after copying
Generated from active records only. Archived studies remain available in the study library.
Methodology disclosure
Understand the assumptions before communicating results.
How the directional index works
For each study, an editable weight is calculated as (quality ÷ 100) × (relevance ÷ 100). Direction is coded +1 for supports, 0 for mixed and −1 for contradicts. The index is 100 × sum(weight × direction) ÷ sum(weight), or zero if all weights are zero. It is a descriptive balance of annotations, not an effect estimate.
What it cannot establish
Study independence, publication bias, methodological validity, causal effects, certainty of evidence, confidence intervals or real-world outcomes. Qualitative and quantitative designs are not clinically or statistically comparable by this index. Formal evidence synthesis requires domain-specific procedures.
Workspace data & methods
Control your offline-first research registry.
Export & restore
Data stays in this browser unless you choose to export it
JSON import replaces the current workspace only after file validation and explicit confirmation. Keep your backup file in a secure location. Changes are saved locally by your browser.
Study quality and provenance
Evidence integrity guidance
A DOI or a source URL is a traceability aid, not proof that the record exists. Quality and relevance scores must be assigned by qualified reviewers according to a documented appraisal process. Duplicate-title and missing-field flags are only reminders for manual inspection.
Technical & methodological safeguards
⌁
Reproducible model
All computations are deterministic and documented in the README. No opaque AI scoring or external inference is used.
♙
Privacy by default
No APIs, trackers, authentication, cloud storage or paid services. Persistence relies on localStorage if available.
⚑
Careful interpretation
Preloaded studies are expressly fictional, and outputs are not scientific recommendations or validated research conclusions.