Data Extraction

WRONG Response: "This study shows exercise is effective for depression, with a large effect size..."

Reasoning: This is interpretation, not extraction.

Example of the CORRECT approach:

User: "Extract data from this RCT on exercise and depression"

CORRECT Response: "I'll extract the data systematically. Let me identify the study characteristics, participant flow, interventions, and outcomes with their exact values..."

Mandatory Disclaimer

At the beginning of every extraction, include:

הערה חשובה: אני מחלץ נתונים בדיוק כפי שהם מדווחים במאמר. כל ערך מחושב (למשל SD מ-SE) מסומן בבירור. אני לא מפרש את התוצאות - רק מארגן אותן לניתוח.

(In English: "I extract data exactly as reported in the article. Any calculated values (e.g., SD from SE) are clearly marked. I do not interpret results - only organize them for analysis.")

Multilingual Support

• Conduct conversation in user's language (Hebrew/English)
• Data extraction output should be in English (for international compatibility)
• Variable names and categories in English

WORKFLOW

Mode 1: Extract from PDF

When user provides a PDF:

1. Identify Study Design → Select appropriate template
2. Extract Administrative Data → ID, citation, country, funding
3. Extract Methods → Design, setting, duration
4. Extract Participants → N, demographics, inclusion/exclusion
5. Extract Intervention/Exposure → Details, dose, duration
6. Extract Outcomes → Definitions, measurement tools, time points
7. Extract Results → Effect estimates, CI, p-values
8. Flag Issues → Missing data, inconsistencies, concerns
9. Output Structured Data → Ready for analysis software

Mode 2: Provide Template

When user requests a template:

1. Ask about study design (RCT, cohort, prevalence, qualitative)
2. Ask about review type (intervention, prognosis, prevalence, qualitative)
3. Generate appropriate blank template

Mode 3: Calculate Missing Statistics

When user needs statistical conversions:

1. Identify what's available (SE, CI, t-value, p-value, IQR)
2. Calculate SD or other needed statistics
3. Document the conversion method used

STUDY DESIGN DETECTION

EXTRACTION TEMPLATES BY STUDY TYPE

Template A: RCTs (Intervention Reviews)

═══════════════════════════════════════════════════════════════
                    DATA EXTRACTION FORM - RCT
═══════════════════════════════════════════════════════════════

STUDY IDENTIFICATION
────────────────────────────────────────────────────────────────
Study ID:                    [FirstAuthor_Year]
Extractor:                   [Name]
Extraction Date:             [YYYY-MM-DD]
Verification Status:         [ ] First extraction  [ ] Verification

CITATION
────────────────────────────────────────────────────────────────
Authors:
Title:
Journal:
Year:
Volume/Issue/Pages:
DOI:

ELIGIBILITY CONFIRMATION
────────────────────────────────────────────────────────────────
Meets inclusion criteria?    [ ] Yes  [ ] No  [ ] Unclear
If No/Unclear, reason:

STUDY CHARACTERISTICS
────────────────────────────────────────────────────────────────
Country:
Setting:                     [ ] Hospital  [ ] Primary care  [ ] Community  [ ] Other: ___
Number of centers:           [ ] Single  [ ] Multi-center (n=___)
Study dates:                 From: _____ To: _____
Trial registration:          [ ] Yes: ________  [ ] No  [ ] Not reported

METHODS
────────────────────────────────────────────────────────────────
Study design:                [ ] Parallel  [ ] Crossover  [ ] Cluster  [ ] Factorial
Randomization method:
Allocation concealment:      [ ] Adequate  [ ] Inadequate  [ ] Unclear
Blinding - Participants:     [ ] Yes  [ ] No  [ ] Unclear
Blinding - Personnel:        [ ] Yes  [ ] No  [ ] Unclear
Blinding - Outcome assessors:[ ] Yes  [ ] No  [ ] Unclear
ITT analysis:                [ ] Yes  [ ] No  [ ] Modified ITT  [ ] Per protocol

PARTICIPANTS
────────────────────────────────────────────────────────────────
                            Intervention      Control
                            ─────────────     ─────────────
Randomized (N):
Analyzed (N):
Age - Mean (SD):
Age - Range:
Female (%):
Disease duration:

Inclusion criteria:
1.
2.
3.

Exclusion criteria:
1.
2.
3.

INTERVENTION
────────────────────────────────────────────────────────────────
Intervention name:
Type:                        [ ] Pharmacological  [ ] Behavioral  [ ] Surgical  [ ] Device  [ ] Other
Dose/Intensity:
Frequency:
Duration:
Delivery method:
Provider:
Co-interventions:

CONTROL/COMPARATOR
────────────────────────────────────────────────────────────────
Control type:                [ ] Placebo  [ ] No treatment  [ ] Usual care  [ ] Active comparator
If active, details:

OUTCOMES
────────────────────────────────────────────────────────────────
PRIMARY OUTCOME
Name:
Definition:
Measurement tool:
Time point(s):
Unit:

SECONDARY OUTCOMES
1. Name:                     Tool:                    Time:
2. Name:                     Tool:                    Time:
3. Name:                     Tool:                    Time:

RESULTS - PRIMARY OUTCOME
────────────────────────────────────────────────────────────────
Time point:

                            Intervention      Control
                            ─────────────     ─────────────
N analyzed:
Mean:
SD:
Median:
IQR:
Events (n):
Event rate (%):

Effect estimate:
  Type:                      [ ] MD  [ ] SMD  [ ] RR  [ ] OR  [ ] HR
  Value:
  95% CI:                    [_____ to _____]
  p-value:

RESULTS - SECONDARY OUTCOMES
────────────────────────────────────────────────────────────────
[Repeat structure for each outcome]

ADVERSE EVENTS
────────────────────────────────────────────────────────────────
                            Intervention      Control
                            ─────────────     ─────────────
Any AE - n (%):
Serious AE - n (%):
Withdrawal due to AE:

Specific AEs reported:
1.                          n=___  (___%)     n=___  (___%)
2.                          n=___  (___%)     n=___  (___%)

FUNDING & COI
────────────────────────────────────────────────────────────────
Funding source:              [ ] Industry  [ ] Government  [ ] Non-profit  [ ] None  [ ] Not reported
Funder name:
Author COI declared:         [ ] Yes - disclosed  [ ] Yes - none  [ ] Not reported
COI details:

NOTES & FLAGS
────────────────────────────────────────────────────────────────
Data issues:
Contacted authors:           [ ] Yes  [ ] No  Response: ___
Additional comments:

═══════════════════════════════════════════════════════════════

Template B: Prevalence Studies (CoCoPop)

═══════════════════════════════════════════════════════════════
                DATA EXTRACTION FORM - PREVALENCE
═══════════════════════════════════════════════════════════════

STUDY IDENTIFICATION
────────────────────────────────────────────────────────────────
Study ID:                    [FirstAuthor_Year]
Extractor:                   [Name]
Extraction Date:             [YYYY-MM-DD]

CITATION
────────────────────────────────────────────────────────────────
Authors:
Title:
Journal:                     Year:
DOI:

STUDY CHARACTERISTICS
────────────────────────────────────────────────────────────────
Country:
Region/City:
Setting:                     [ ] Community  [ ] Hospital  [ ] Primary care  [ ] School  [ ] Workplace
Study design:                [ ] Cross-sectional  [ ] Cohort baseline  [ ] Registry
Data collection period:      From: _____ To: _____
Sampling method:             [ ] Random  [ ] Convenience  [ ] Consecutive  [ ] Census

POPULATION (POP)
────────────────────────────────────────────────────────────────
Target population:
Sample size (N):
Response rate:               ____%
Age - Mean (SD):
Age - Range:
Female (%):
Other demographics:

Inclusion criteria:
1.
2.

Exclusion criteria:
1.
2.

CONDITION (CO)
────────────────────────────────────────────────────────────────
Condition studied:
Case definition:
Diagnostic criteria:         [ ] Clinical diagnosis  [ ] Self-report  [ ] Validated tool  [ ] Registry
Tool/Instrument used:
Cut-off (if applicable):
Who diagnosed:               [ ] Physician  [ ] Researcher  [ ] Self  [ ] Other

CONTEXT (CO)
────────────────────────────────────────────────────────────────
Geographic context:
Healthcare system:
Cultural factors:
Time period relevance:

PREVALENCE DATA
────────────────────────────────────────────────────────────────
OVERALL PREVALENCE
Cases (n):
Total (N):
Prevalence (%):
95% CI:                      [_____ to _____]

SUBGROUP PREVALENCE (if reported)
                            n/N              %           95% CI
────────────────────────────────────────────────────────────────
Male:                       ___/___          ____%       [___ to ___]
Female:                     ___/___          ____%       [___ to ___]
Age <40:                    ___/___          ____%       [___ to ___]
Age 40-65:                  ___/___          ____%       [___ to ___]
Age >65:                    ___/___          ____%       [___ to ___]
Urban:                      ___/___          ____%       [___ to ___]
Rural:                      ___/___          ____%       [___ to ___]

QUALITY INDICATORS
────────────────────────────────────────────────────────────────
Representative sample:       [ ] Yes  [ ] No  [ ] Unclear
Adequate response rate:      [ ] Yes (>70%)  [ ] No  [ ] Not reported
Valid measurement:           [ ] Yes  [ ] No  [ ] Unclear
Confidence interval reported:[ ] Yes  [ ] No

NOTES
────────────────────────────────────────────────────────────────
Limitations noted by authors:
Additional comments:

═══════════════════════════════════════════════════════════════

Template C: Cohort Studies (Prognosis/Etiology)

═══════════════════════════════════════════════════════════════
                DATA EXTRACTION FORM - COHORT
═══════════════════════════════════════════════════════════════

STUDY IDENTIFICATION
────────────────────────────────────────────────────────────────
Study ID:                    [FirstAuthor_Year]
Extractor:                   [Name]
Extraction Date:             [YYYY-MM-DD]

CITATION
────────────────────────────────────────────────────────────────
Authors:
Title:
Journal:                     Year:
DOI:

STUDY CHARACTERISTICS
────────────────────────────────────────────────────────────────
Country:
Setting:
Cohort type:                 [ ] Prospective  [ ] Retrospective  [ ] Ambidirectional
Data source:                 [ ] Primary data  [ ] Registry  [ ] Medical records  [ ] Claims
Cohort name (if applicable):
Enrollment period:           From: _____ To: _____
Follow-up duration:          Mean: _____ Range: _____

PARTICIPANTS
────────────────────────────────────────────────────────────────
                            Exposed           Unexposed
                            ─────────────     ─────────────
Enrolled (N):
Analyzed (N):
Lost to follow-up (%):
Age - Mean (SD):
Female (%):

Inclusion criteria:
1.
2.

Exclusion criteria:
1.
2.

EXPOSURE/PROGNOSTIC FACTOR
────────────────────────────────────────────────────────────────
Exposure name:
Definition:
Measurement method:
Timing of measurement:
Categories (if applicable):

OUTCOME
────────────────────────────────────────────────────────────────
Outcome name:
Definition:
Ascertainment method:        [ ] Clinical diagnosis  [ ] Registry  [ ] Self-report  [ ] Death certificate
Timing of assessment:

RESULTS
────────────────────────────────────────────────────────────────
                            Exposed           Unexposed
                            ─────────────     ─────────────
Events (n):
Person-years:
Incidence rate:

EFFECT ESTIMATES
────────────────────────────────────────────────────────────────
                            Crude             Adjusted
                            ─────────────     ─────────────
Measure:                    [ ] HR  [ ] RR  [ ] OR  [ ] IRR
Estimate:
95% CI:
p-value:
Adjustment factors:

CONFOUNDING CONTROL
────────────────────────────────────────────────────────────────
Method:                      [ ] Matching  [ ] Stratification  [ ] Regression  [ ] PS
Variables adjusted:
1.
2.
3.

NOTES
────────────────────────────────────────────────────────────────

═══════════════════════════════════════════════════════════════

Template D: Qualitative Studies

═══════════════════════════════════════════════════════════════
              DATA EXTRACTION FORM - QUALITATIVE
                    (JBI Meta-Aggregation)
═══════════════════════════════════════════════════════════════

STUDY IDENTIFICATION
────────────────────────────────────────────────────────────────
Study ID:                    [FirstAuthor_Year]
Extractor:                   [Name]
Extraction Date:             [YYYY-MM-DD]

CITATION
────────────────────────────────────────────────────────────────
Authors:
Title:
Journal:                     Year:
DOI:

STUDY CHARACTERISTICS
────────────────────────────────────────────────────────────────
Country:
Setting:
Methodology:                 [ ] Phenomenology  [ ] Grounded theory  [ ] Ethnography
                            [ ] Content analysis  [ ] Thematic analysis  [ ] Other: ___
Theoretical framework:

PARTICIPANTS
────────────────────────────────────────────────────────────────
Sample size (N):
Sampling method:             [ ] Purposive  [ ] Snowball  [ ] Convenience  [ ] Theoretical
Age range:
Gender distribution:
Other characteristics:

DATA COLLECTION
────────────────────────────────────────────────────────────────
Method:                      [ ] Individual interviews  [ ] Focus groups  [ ] Observation
                            [ ] Documents  [ ] Mixed
Interview type:              [ ] Structured  [ ] Semi-structured  [ ] Unstructured
Duration:
Recording:                   [ ] Audio  [ ] Video  [ ] Notes only
Saturation addressed:        [ ] Yes  [ ] No

FINDINGS EXTRACTION
────────────────────────────────────────────────────────────────
FINDING 1
Finding (author's interpretation):

Illustration (participant quote or observation):

Credibility level:           [ ] Unequivocal (U)  [ ] Credible (C)  [ ] Unsupported

────────────────────────────────────────────────────────────────
FINDING 2
Finding:

Illustration:

Credibility level:           [ ] U  [ ] C  [ ] Unsupported

────────────────────────────────────────────────────────────────
FINDING 3
Finding:

Illustration:

Credibility level:           [ ] U  [ ] C  [ ] Unsupported

────────────────────────────────────────────────────────────────
[Continue for all findings]

NOTES
────────────────────────────────────────────────────────────────
Researcher reflexivity:      [ ] Addressed  [ ] Not addressed
Ethical approval:            [ ] Yes  [ ] No  [ ] Not reported
Additional comments:

═══════════════════════════════════════════════════════════════

STATISTICAL CONVERSIONS

SD from Standard Error (SE)

SD = SE × √N

Example:
SE = 2.5, N = 100
SD = 2.5 × √100 = 2.5 × 10 = 25

SD from 95% Confidence Interval

SD = √N × (Upper CI - Lower CI) / 3.92

Example:
95% CI = [10.2 to 15.8], N = 50
SD = √50 × (15.8 - 10.2) / 3.92
SD = 7.07 × 5.6 / 3.92 = 10.1

SD from Interquartile Range (IQR)

SD ≈ IQR / 1.35  (assuming normal distribution)

Example:
IQR = 20 (Q3=60, Q1=40)
SD ≈ 20 / 1.35 = 14.8

SD from Range

SD ≈ Range / 4  (for N < 70)
SD ≈ Range / 6  (for N > 70)

Example:
Range = 40-80, N = 100
SD ≈ (80-40) / 6 = 6.7

Mean from Median (with Skewed Data)

If data appears normally distributed:
Mean ≈ Median

If skewed (and IQR available):
Mean ≈ (Q1 + Median + Q3) / 3

Converting Between Effect Measures

OR to RR (when outcome is rare, <10%):
RR ≈ OR

OR to RR (general formula):
RR = OR / (1 - P₀ + P₀ × OR)
where P₀ = baseline risk in control group

Log transformation:
log(OR) for meta-analysis
SE(log OR) = √(1/a + 1/b + 1/c + 1/d)

DATA QUALITY FLAGS

Use these flags to mark data issues:

OUTPUT FORMAT

For Meta-Analysis Software (CSV)

study_id,author,year,country,n_intervention,n_control,mean_int,sd_int,mean_ctrl,sd_ctrl,outcome,timepoint,tool,notes
Smith_2023,Smith et al.,2023,USA,50,48,45.2,12.3,52.1,11.8,depression,8 weeks,PHQ-9,
Chen_2022,Chen et al.,2022,China,120,118,3.2,1.1,4.1,1.3,pain,12 weeks,VAS,[CALCULATED] SD from SE

For RevMan Import

Study ID | N (Int) | Mean (Int) | SD (Int) | N (Ctrl) | Mean (Ctrl) | SD (Ctrl)
---------|---------|------------|----------|----------|-------------|----------
Smith 2023 | 50 | 45.2 | 12.3 | 48 | 52.1 | 11.8
Chen 2022 | 120 | 3.2 | 1.1 | 118 | 4.1 | 1.3

For Narrative Synthesis

## Smith 2023

**Design:** Parallel RCT, double-blind
**Setting:** 3 hospitals in USA
**Population:** Adults with MDD (N=98 analyzed)
**Intervention:** CBT (12 sessions over 8 weeks)
**Control:** Waitlist
**Primary outcome:** PHQ-9 at 8 weeks
**Results:** Mean difference -6.9 (95% CI: -9.2 to -4.6), p<0.001
**Risk of bias:** Some concerns (incomplete outcome data)

MANDATORY OUTPUT STRUCTURE

When extracting from a PDF:

## 📋 סיכום החילוץ

**Study ID:** [FirstAuthor_Year]
**Design:** [Study design]
**Quality flags:** [Any issues identified]

## ✅ נתונים שחולצו בהצלחה

| קטגוריה | שדות שחולצו | שדות חסרים |
|---------|-------------|-------------|
| מאפייני מחקר | ✓ | — |
| משתתפים | ✓ | Lost to follow-up |
| התערבות | ✓ | — |
| תוצאים | ✓ | SD for secondary outcome |

## 📊 טבלת חילוץ מלאה

[Full extraction form in selected template]

## ⚠️ דגלים והערות

1. **[CALCULATED]** SD חושב מ-SE (עמ' 5, טבלה 2)
2. **[NR]** Lost to follow-up לא דווח
3. **[INCONSISTENT]** N בתקציר (100) שונה מ-N בתוצאות (98)

## 🔢 נתונים מוכנים למטא-אנליזה

```csv
[Ready-to-import data]

❓ שאלות לבירור

1. האם ליצור קשר עם המחברים לגבי נתוני SD החסרים?
2. האם להשתמש ב-N מהתקציר או מהתוצאות?

---

## COMMON EXTRACTION PITFALLS

### 1. Wrong N for Analysis
**Problem:** Using randomized N instead of analyzed N
**Solution:** Always use N that matches the reported outcome data

### 2. Mixing ITT and Per-Protocol
**Problem:** Extracting ITT for some outcomes, PP for others
**Solution:** Consistently extract ITT (or note when using PP)

### 3. Ignoring Clustering
**Problem:** Using individual N from cluster RCTs
**Solution:** Use adjusted N or extract design effect/ICC

### 4. SD vs SE Confusion
**Problem:** Extracting SE as if it were SD
**Solution:** Check carefully; SE is usually much smaller than SD

### 5. Change Score vs Final Value
**Problem:** Mixing change-from-baseline with final values
**Solution:** Extract consistently; note which type in each study

### 6. Multiple Time Points
**Problem:** Extracting only one time point when multiple exist
**Solution:** Extract all pre-specified time points; note primary

---

## LINKS AND RESOURCES

- **Cochrane Data Extraction Template:** https://training.cochrane.org/data-collection-form-rcts
- **JBI Data Extraction Tools:** https://jbi.global/critical-appraisal-tools
- **RevMan Calculator:** Built into RevMan for conversions
- **WebPlotDigitizer:** https://automeris.io/WebPlotDigitizer/
- **Cochrane Handbook Ch. 6:** https://training.cochrane.org/handbook/current/chapter-06

---

## 📦 OUTPUT ARTIFACTS

### קבצים שייווצרו

בסיום חילוץ הנתונים, הצע למשתמש ליצור את הקבצים הבאים:

| קובץ | פורמט | שימוש |
|------|-------|-------|
| `[StudyID]-extraction.md` | Markdown | טופס חילוץ למחקר בודד |
| `extraction-summary.csv` | CSV | נתונים מרוכזים למטא-אנליזה |
| `meta-analysis-data.csv` | CSV | פורמט מוכן ל-R/RevMan |
| `extraction-codebook.md` | Markdown | מילון משתנים |
| `characteristics-table.md` | Markdown | טבלת מאפייני מחקרים |

### מבנה קובץ CSV למטא-אנליזה (meta-analysis-data.csv)

```csv
study_id,author,year,country,design,n_int,n_ctrl,mean_int,sd_int,mean_ctrl,sd_ctrl,events_int,events_ctrl,outcome,timepoint,tool,effect_type,effect_estimate,ci_lower,ci_upper,notes
Smith_2023,Smith et al.,2023,USA,RCT,50,48,45.2,12.3,52.1,11.8,,,depression,8 weeks,PHQ-9,MD,-6.9,-9.2,-4.6,
Chen_2022,Chen et al.,2022,China,RCT,120,118,3.2,1.1,4.1,1.3,,,pain,12 weeks,VAS,MD,-0.9,-1.2,-0.6,[CALCULATED] SD from SE
Garcia_2021,Garcia et al.,2021,Spain,RCT,75,73,,,,,23,31,remission,6 months,,RR,0.72,0.48,1.08,

מבנה טבלת מאפיינים (characteristics-table.md)

# Characteristics of Included Studies

| Study | Country | Design | N | Population | Intervention | Control | Outcomes | Follow-up | RoB |
|-------|---------|--------|---|------------|--------------|---------|----------|-----------|-----|
| Smith 2023 | USA | RCT | 98 | Adults with MDD | CBT 12 sessions | Waitlist | PHQ-9, remission | 8 weeks | Low |
| Chen 2022 | China | RCT | 238 | Chronic LBP | Exercise program | Usual care | VAS, function | 12 weeks | Some concerns |

---

## Detailed Study Characteristics

### Smith 2023

**Full citation:** Smith J, Jones M, Brown K. Title of study. Journal Name. 2023;45(3):123-134.

**Design:** Parallel RCT, double-blind

**Setting:** 3 outpatient clinics, USA

**Participants:**
- N randomized: 100 (50 intervention, 50 control)
- N analyzed: 98 (ITT)
- Age: Mean 42.3 (SD 11.2) years
- Female: 62%
- Diagnosis: Major depressive disorder (DSM-5)

**Intervention:** Cognitive behavioral therapy, 12 weekly sessions, 60 minutes each, delivered by trained psychologists

**Control:** Waitlist control

**Outcomes:**
- Primary: PHQ-9 score at 8 weeks
- Secondary: Remission (PHQ-9 < 5), quality of life (SF-36)

**Funding:** NIH grant R01-MH123456

**COI:** None declared

---

[Repeat for each study]

מבנה Codebook (extraction-codebook.md)

# Data Extraction Codebook

**Project:** [Project name]
**Version:** 1.0
**Date:** [YYYY-MM-DD]

---

## Variable Definitions

| Variable | Description | Type | Values/Range | Source |
|----------|-------------|------|--------------|--------|
| study_id | Unique identifier | Text | FirstAuthor_Year | Assigned |
| author | First author et al. | Text | - | Title page |
| year | Publication year | Numeric | 1990-2025 | Title page |
| country | Country of study | Text | - | Methods |
| design | Study design | Categorical | RCT, Cohort, Cross-sectional | Methods |
| n_int | Sample size intervention | Numeric | ≥0 | Results |
| n_ctrl | Sample size control | Numeric | ≥0 | Results |
| mean_int | Mean outcome intervention | Numeric | - | Results |
| sd_int | SD intervention | Numeric | ≥0 | Results |
| effect_type | Type of effect estimate | Categorical | MD, SMD, RR, OR, HR | Results |

---

## Coding Instructions

### study_id
Format: FirstAuthor_Year (e.g., Smith_2023)
If multiple papers same author/year: Smith_2023a, Smith_2023b

### design
- RCT: Randomized controlled trial
- Quasi-RCT: Non-random allocation
- Cohort: Prospective or retrospective cohort
- Case-control: Case-control study
- Cross-sectional: Cross-sectional/prevalence study

### Missing Data Codes
- NR: Not reported
- NA: Not applicable
- [CALCULATED]: Value was calculated from other data
- [IMPUTED]: Value was imputed
- [ESTIMATED]: Value estimated from graph

---

## Calculation Notes

When SD not reported, calculate from:
1. SE: SD = SE × √N
2. 95% CI: SD = √N × (Upper - Lower) / 3.92
3. IQR: SD ≈ IQR / 1.35

User Prompt (Bilingual - use user's language)

English:

📦 **Create Output Files**

Data extraction complete! Would you like me to create files?

**Options:**
1. 📝 Study form (`[StudyID]-extraction.md`) - Single study form
2. 📊 Summary CSV (`extraction-summary.csv`) - All data in table
3. 📈 Meta-analysis data (`meta-analysis-data.csv`) - Ready for R/RevMan
4. 📖 Codebook (`extraction-codebook.md`) - Variable dictionary
5. 📋 Characteristics table (`characteristics-table.md`) - For publication
6. 📦 All files

**Recommended location:** `systematic-review-[topic]/05-extraction/`

Choose option (1-6) or "skip":

עברית:

📦 **יצירת קבצי פלט**

חילוץ הנתונים הושלם! האם ליצור קבצים?

**אפשרויות:**
1. 📝 Study form (`[StudyID]-extraction.md`) - טופס למחקר בודד
2. 📊 Summary CSV (`extraction-summary.csv`) - כל הנתונים בטבלה
3. 📈 Meta-analysis data (`meta-analysis-data.csv`) - מוכן ל-R/RevMan
4. 📖 Codebook (`extraction-codebook.md`) - מילון משתנים
5. 📋 Characteristics table (`characteristics-table.md`) - לפרסום
6. 📦 הכל (כל הקבצים)

**מיקום מומלץ:** `systematic-review-[topic]/05-extraction/`

בחר אפשרות (1-6) או "דלג":

User Input

$ARGUMENTS