Develop evidence-linked systematic review reports
This guide explains how AccuScripter turns selected full-text evidence into a reviewable report workflow: protocol context, PDF analysis, study linkage, report structure, evidence tables, writing, versions, report updates, section revisions, and exports.
What AccuScripter does
AccuScripter is a report-writing workspace for systematic and targeted evidence reviews. It brings protocol information, selected study documents, AI-prepared analysis, study linkage, report structure, evidence tables, and writing into one guided sequence.
The output is intended to be reviewed. AccuScripter prepares structured drafts and supporting tables, while researchers remain responsible for checking included evidence, citations, interpretation, and final wording before using the report downstream.
| What you provide | What AccuScripter prepares | What you review or export |
|---|---|---|
| Research question, objectives, protocol, and writing context | Writing hints and scope boundaries | Whether the report objective matches the review |
| Selected main PDFs and optional supplementary PDFs | Text extraction, mini extraction, and study summaries | Extraction status, skipped files, failed files, and eligible evidence |
| Confirmed analysis and structure choices | Study linkage, report hierarchy, evidence tables, and draft narrative | Editable evidence tables, report versions, revision outputs, and downloadable files |
The complete AccuScripter workflow
AccuScripter has strict review gates. Later stages depend on earlier data, so changes to PDFs, analysis, structure, or table columns can make downstream outputs stale.
- 1
Set project context
Create the project, confirm report type, and enter or review the protocol, research question, synopsis, and objectives.
- 2
Upload or sync PDFs
Add standalone PDFs or sync workflow-selected PDFs, then wait for extracted text before analysis.
- 3
Analyze PDFs
Run mini extraction so AccuScripter has study summaries and key evidence for structure and table generation.
- 4
Review study linkage
For clinical reports, review grouped publications, primary records, secondary records, and unlinked studies.
- 5
Confirm report structure
Generate and review the proposed report hierarchy, references, planned tables, and any unplaced evidence.
- 6
Prepare evidence tables
Generate tables from the confirmed structure, inspect values, and add or extract custom columns when required.
- 7
Write and version the report
Start writing, review the generated report, create Version 1, and use revisions or report updates when the evidence changes.
Choose standalone or workflow-linked writing
AccuScripter can run as a standalone writing workspace or as the writing stage of a Master Project. Workflow-linked projects inherit context and source files from earlier stages, while standalone projects start from the PDFs uploaded directly into AccuScripter.
| Mode | Best used when | Important behavior |
|---|---|---|
| Standalone AccuScripter | The team already has the full-text PDFs and wants to build a report directly. | Users upload main and supplementary PDFs in AccuScripter and control the writing evidence set there. |
| Workflow-linked AccuScripter | The writing stage belongs to a Master Project with search, screening, and PDF Manager context. | Eligible files and some protocol fields are inherited from upstream workflow stages and managed through the connected workflow. |
Set protocol, goals, and report type before analysis
The protocol and goals shape the writing task. Use them to describe the review question, intended scope, objectives, and any clinical or consensus context that should affect structure and narrative.
| Input | Purpose | Review guidance |
|---|---|---|
| Research question | Defines the central question the report should answer. | Keep it focused enough to guide Results and Discussion sections. |
| Protocol input / PICOS | Provides population, intervention, comparator, outcomes, study design, and other boundaries. | In workflow mode, review inherited PICOS and update the source workflow stage if it is wrong. |
| Objectives | Shapes the report goals and section emphasis. | Enter one objective per line so the writing task stays explicit. |
| Project synopsis | Gives short context for background and executive summary writing. | Describe the review context rather than repeating every table field. |
| Consensus themes | Defines themes used in consensus-program analysis. | Load and check theme questions before running PDF analysis. |
Prepare PDFs and wait for extracted text
AccuScripter writes from selected main PDFs and optional supplementary PDFs. Standalone users upload PDFs directly. Workflow-linked users select PDFs in Workflow PDF Manager and sync them into the writing stage.
Supplementary-file naming
To attach supplementary PDFs to a main file, keep the same base file name and add a supported suffix such as _SUPP1, _SUPPL, _APPENDIX1, or _DATASET1. AccuScripter supports up to three supplementary PDFs per main file.
| Status | Meaning | User action |
|---|---|---|
| Pending or queued | The file is waiting for text extraction. | Wait or refresh before continuing to analysis. |
| Processing or retrying | Text extraction is active or being attempted again. | Do not start downstream analysis until required files are ready. |
| Completed | Text is available for analysis and writing. | Include the file if it belongs in the writing evidence set. |
| Failed | The system could not extract usable text. | Retry extraction or replace the file when possible. |
| Skipped | The file cannot be used for writing, such as when it exceeds the page limit. | Exclude it or provide a suitable replacement. |
Use mini extraction to prepare evidence for structure and tables
Analyze PDFs prepares the compact study information AccuScripter needs before linkage, structure, evidence tables, and writing. Clinical projects use the standard writing-analysis fields. Consensus-program projects use the consensus themes and loaded theme questions to categorize responses.
| Analysis state | What it means | What to check |
|---|---|---|
| Ready to analyze | Required PDFs have extracted text and no analysis is currently running. | Confirm file inclusion before spending credits on analysis. |
| In progress | Mini extraction or consensus analysis is running. | Wait for completion and monitor progress messages. |
| Partial | Some PDFs were analyzed while others remain pending, skipped, or failed. | Review the affected files and decide whether to retry or continue with the available evidence. |
| Completed | Analysis results are available for the selected files. | Export or inspect analysis rows before linkage, structure, and writing. |
| Failed | Analysis did not complete successfully. | Review the error, retry failed summaries or extraction, then rerun analysis. |
Group related publications before building the report
For clinical projects, linkage analysis groups records that appear to describe the same study, such as publications sharing an acronym or NCT identifier. This helps AccuScripter avoid treating linked papers as unrelated evidence.
| Linkage output | Meaning | How to use it |
|---|---|---|
| Linked groups | Records with shared trial identifiers, acronyms, or defensible relationship signals. | Review primary and secondary publications before relying on grouped evidence. |
| Primary record | The publication AccuScripter treats as the lead record for the group. | Check that the lead record is appropriate for reporting. |
| Unlinked records | Records without a confident linkage group. | Review manually when a known relationship may be missing from the PDFs. |
Review the proposed report hierarchy before writing
Report structure organizes sections, references, and planned evidence tables. This is the main checkpoint between evidence analysis and narrative generation.
| Structure item | Why it matters | Review guidance |
|---|---|---|
| Sections | They define the report hierarchy and writing targets. | Check titles, order, and whether every planned discussion area is represented. |
| References | They connect sections to the evidence used for narrative generation. | Confirm that eligible and relevant studies are allocated to the right sections. |
| Evidence tables | They define the structured table outputs used in the report. | Review table purpose, assigned studies, and whether extra columns are needed. |
| Unplaced evidence | It signals that some studies had no defensible Results placement. | Review before writing, especially during report updates. |
| Suggested additions | Update workflows can propose new sections or tables. | Approve only additions that match the review scope. |
Generate and manage evidence tables from the confirmed structure
Evidence tables are generated from the saved structure and analyzed study evidence. They can be exported to Excel, and their columns can be managed when the report needs additional extracted fields.
| Table control | Use it when | Downstream effect |
|---|---|---|
| Generate evidence tables | The report structure is ready and no table output exists yet. | Creates the baseline table rows for review and export. |
| Update evidence tables | A Version 2 update batch has finalized placement into the report structure. | Adds new evidence rows and records table-level changes. |
| Manage columns | A table needs additional extracted fields beyond its current schema. | New or edited columns become pending extraction work. |
| Extract pending columns | Custom column changes have been saved. | Runs column-study extraction in batches and updates the table output. |
| Export to Excel | The team needs reviewable table output outside the app. | Downloads the current evidence-table view. |
Generate the draft and preserve report versions
Writing sends the confirmed structure, analysis, linkage results, evidence tables, and included studies to the writing service. The job runs asynchronously and returns a report output that can be reviewed and exported.
Create a Version 2 update from new evidence
Report updates start after Version 1 exists. New PDFs are processed as an update batch, analyzed, linked, placed into the existing report structure, finalized, and then used to update evidence tables and writing.
| Update stage | Purpose | Gate before continuing |
|---|---|---|
| Create update batch | Collect new PDFs for the next report version. | Version 1 must exist first. |
| Sync or upload update PDFs | Prepare new files without changing the frozen baseline evidence. | Text extraction must complete or failed files must be excluded. |
| Analyze update PDFs | Append analyzed references for the new version. | Included update files must have usable analysis. |
| Place new evidence | Map new studies to existing or suggested report sections and tables. | Placements must be reviewed and finalized. |
| Update evidence tables | Add the new evidence rows and table change details. | Evidence-table update must complete before report update writing. |
| Generate Version 2 | Create the updated report and comparison output. | Review touched sections, new references, table changes, warnings, and unplaced evidence. |
Revise selected narrative sections with impact review
Section revision starts from a completed report version. The user selects a section, writes revision instructions, runs impact analysis, reviews required and recommended affected sections, generates the revised narrative, assembles the result, and approves it.
| Revision concept | Meaning | User responsibility |
|---|---|---|
| Revision instruction | The user-written direction for how a section should change. | Be specific about what to revise and what should remain unchanged. |
| Impact analysis | A dependency review that identifies required and recommended affected sections. | Run it again if the instruction or selected section changes. |
| Required sections | Sections that must be regenerated for a consistent revision. | Review them as part of the revision scope. |
| Recommended sections | Additional sections that may need updating. | Choose whether to include them based on scientific and editorial judgment. |
| Approval | The step that stores the assembled revision as a completed subversion. | Approve only after checking the revised narrative and affected outputs. |
Know when downstream outputs must be rerun
AccuScripter keeps a staged chain of evidence and writing outputs. If an upstream input changes, later outputs may no longer represent the current project state.
| Change made | Affected outputs | Recommended action |
|---|---|---|
| Protocol, question, synopsis, or objectives change | Analysis interpretation, structure, evidence tables, and writing may no longer match the intended report. | Review the stage chain and rerun affected analysis, structure, tables, or writing. |
| PDF evidence set changes | Analysis, linkage, structure, tables, and writing may be stale. | Run extraction and analysis for the current files, then regenerate downstream outputs. |
| Analysis rows are edited or replaced | Linkage, report structure, evidence tables, and writing can reflect older analysis. | Rerun linkage or regenerate the structure and downstream outputs as needed. |
| Linkage results are rerun | Grouped-study assumptions used by structure and writing may change. | Review structure allocation and regenerate affected tables or writing. |
| Report structure is edited or finalized again | Evidence tables and writing should follow the revised hierarchy. | Regenerate evidence tables when table definitions changed, then regenerate writing. |
| Evidence-table columns are added, removed, or edited | Table values and report text that references them may be stale. | Extract pending columns and review whether writing needs regeneration. |
| Revision instruction changes after impact analysis | The affected-section scope may be stale. | Analyze impact again before generating or approving the revision. |
Export the outputs that support review and handoff
AccuScripter exports are checkpoints for review, documentation, and external editing. Use them to verify the evidence trail before relying on a report version.
| Export | What it contains |
|---|---|
| Analysis Excel | Study summaries, extracted analysis fields, and consensus responses where applicable. |
| Linkage Excel | Linked study groups, primary or secondary relationships, unlinked records, and update linkage changes. |
| Evidence-table Excel | Generated evidence tables, rows, columns, and values for the current selected view. |
| Report DOCX | The generated report draft for review and editing outside the application. |
| Version comparison | Touched sections, new references, evidence-table changes, warnings, and review items for update or revision outputs. |
| Change log DOCX | A document-oriented summary of recorded changes between compared report outputs. |
Working reliably in AccuScripter
| Do | Avoid |
|---|---|
| check_circle Confirm the protocol and objectives before analyzing PDFs. | cancel Run analysis from vague or outdated writing context. |
| check_circle Use consistent main and supplementary PDF naming. | cancel Upload supplements with names that cannot be matched to a main PDF. |
| check_circle Wait for extracted text before running Analyze PDFs. | cancel Continue while required files are still pending or failed. |
| check_circle Review linkage groups and unlinked records before structure generation. | cancel Assume every related publication was grouped correctly. |
| check_circle Finalize the report structure before generating tables. | cancel Write from an unchecked or unstable outline. |
| check_circle Check evidence tables and custom columns before writing. | cancel Treat generated values as publication-ready without review. |
| check_circle Use versions, updates, and revisions deliberately. | cancel Mix baseline files, update files, and section revisions without checking their scope. |
Before using a report downstream
- check The report type and protocol context match the intended review.
- check The selected main PDFs and supplementary PDFs are correct.
- check Required files have completed text extraction.
- check PDF analysis results have been reviewed or exported where needed.
- check Clinical study linkage groups and unlinked records have been checked.
- check The report structure has been reviewed and finalized.
- check Evidence tables, columns, and extracted values have been verified.
- check The generated narrative, references, and conclusions have been reviewed by the research team.
- check Version, update, revision, and comparison outputs reflect the intended report state.