Metadata-Version: 2.4
Name: sentinel-clinical-reg-tools
Version: 1.0.0
Summary: ALCOA+ Regulatory Medical Writing, EU CTR 536/2014 PLS Scorer, ClinicalTrials.gov v2 Parser & ICH E3/CONSORT Linter
Author-email: Kaled Bouchagraoui <kaledbouchagraoui7@gmail.com>
License: MIT
Keywords: clinical-trials,regulatory-affairs,medical-writing,alcoa-plus,prisma-2020,ich-e3
Requires-Python: >=3.10
Description-Content-Type: text/markdown
License-File: LICENSE
Dynamic: license-file

<div align="center">
  <img src="assets/sentinel_logo_editorial.png" alt="Sentinel — Clinical Data, PRISMA 2020 & Sequence QC Suite" width="640" />
</div>

# Clinical Literature, Biostatistics & Sequence QC Automation

**ALCOA+ Reproducible Pipelines for EU MDR Annex XIV (MEDDEV 2.7/1 Rev 4), ICH E6(R2) / E9 / M4, and EFSA / Cosmetovigilance Evidence Dossiers.**

[![Python 3.10+](https://img.shields.io/badge/python-3.10%2B-blue)](https://www.python.org/) [![License: MIT](https://img.shields.io/badge/license-MIT-green)](LICENSE) [![Open in Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/github/kaledbouchagraoui7-bot/computational-biology-portfolio/blob/main/computational-biology-portfolio/02_clinical_trial_eda.ipynb)

> **100% Free & Open-Source (MIT) · Zero-Install Demo:** Built by a 17-year-old high-school student in France for researchers, students, and medical writers needing reproducible PRISMA 2020 mechanical prep, ICH E9 baseline statistics, and version-pinned NCBI sequence checks. **[Try the Zero-Install Browser Web Demo](https://kaledbouchagraoui7-bot.github.io/computational-biology-portfolio/)** (runs directly in your browser against PubMed & Europe PMC with pre-loaded benchmark datasets).

---

## Executive Capabilities

This repository automates the three highest-friction mechanical workflows in **clinical literature synthesis, biostatistics QA, and sequence QC** under strict **ALCOA+ traceability** and **zero-fabrication guarantees**:

| Module | Regulatory Standard | Key Automated Deliverables |
|---|---|---|
| **01 — Reference Transcript & Variant QC**<br>[`01_ncbi_sequence_analyzer.py`](computational-biology-portfolio/01_ncbi_sequence_analyzer.py) | **ICH E6(R2) §4.9 · FDA/IVDR Genetic Biomarker Dossiers** | Versioned NCBI RefSeq/Entrez mRNA retrieval (`NM_000492.4`), ambiguity-aware GC screening, 3-frame ORF verification, and immutable accession-version logging. |
| **02 — Clinical Trial Data Cleaning & Baseline Table 1**<br>[`02_clinical_trial_eda.py`](computational-biology-portfolio/02_clinical_trial_eda.py) | **ICH E9 §5.2 · CDISC Harmonisation · CONSORT** | Auditable de-duplication and outlier trail (`cleaned_cohort.csv`), baseline **Table 1** (Welch t-test & Pearson χ² validated to `<1e-9` vs. SciPy), and transparent distribution figure ([`table1_demographics.png`](computational-biology-portfolio/table1_demographics.png)). |
| **03 — Systematic Literature & PRISMA 2020 Packager**<br>[`03_pubmed_literature_synthesizer.py`](computational-biology-portfolio/03_pubmed_literature_synthesizer.py) + [`pilot_package_generator.py`](computational-biology-portfolio/pilot_package_generator.py) | **EU MDR 2017/745 Annex XIV · MEDDEV 2.7/1 Rev 4 · ICH M4** | Dual-source NCBI PubMed + Europe PMC REST queries, exact PRISMA 2020 conservation accounting, Vancouver-cited **Landscape Word Evidence Table (`.docx`)**, **4-sheet Excel Screening Matrix (`.xlsx`)** with soft amber QA highlights on missing fields, and **300-DPI PRISMA 2020 Flow Diagram (`.png` / `.pdf`)**. |

## Automation Boundaries: What Code Does vs. Human Judgment

| What These Scripts Automate (Deterministic) | What Requires Qualified Human Judgment (Clinical / Biostats) |
|---|---|
| Cross-database PubMed + Europe PMC querying, PMID/DOI deduplication, and exact PRISMA flow arithmetic | PICO clinical inclusion/exclusion decisions, risk-of-bias (RoB-2 / ROBINS-I) appraisal, and CER equivalence justification |
| Exact Welch t-test / χ² / Shapiro-Wilk baseline Table 1 math and outlier flagging | Clinical significance interpretation, handling of missing-not-at-random bias, and protocol deviation adjudication |
| Version-pinned RefSeq transcript QC and 3-frame ORF verification | Variant pathogenicity classification (ACMG/AMP) and clinical biomarker validation |

---

## Sample Turnkey Output (`dossier_glp1_safety/`)

- **Live Literature Matrix:** [`computational-biology-portfolio/dossier_glp1_safety/literature_matrix.md`](computational-biology-portfolio/dossier_glp1_safety/literature_matrix.md)
- **Full Documentation & Quickstart:** [`computational-biology-portfolio/README.md`](computational-biology-portfolio/README.md)
- **Regulatory & Privacy Compliance Framework:** [`LEGAL_COMPLIANCE_AND_ETHICS.md`](LEGAL_COMPLIANCE_AND_ETHICS.md)

## Public Dataset Case Studies & Empirical Ground-Truth Benchmark

- **[Case Study 02 — Ground-Truth Benchmark vs. Published Review (`PMC13739473`)](computational-biology-portfolio/case_studies/02_published_prisma_ground_truth_benchmark.md)**: Re-runs the exact published PubMed query from *Sports genomics research in Turkish athlete populations (2019–2026)* (`226` vs. `225` reported MEDLINE records, `99.56%` match), deduplicates `422` combined PubMed + Europe PMC records down to `251` unique records, and documents why human full-text eligibility screening remains mandatory.
- **[Case Study 01 — GLP-1 Cardiorenal PRISMA 2020 Screening Audit (`135 -> 25` studies)](computational-biology-portfolio/case_studies/01_glp1_prisma2020_reproducibility_study.md)** · ([Reproducible Notebook](computational-biology-portfolio/case_studies/01_glp1_prisma2020_reproducibility_study.ipynb)): Dual-database PubMed + Europe PMC retrieval (`11` duplicates stripped; automates mechanical prep & count accounting only — not a standalone systematic review) paired with an explicit clinical false-positive precision audit.

## Regulatory Medical Writing Portfolio (12 ALCOA+ Specimens)

Prepared from public ClinicalTrials.gov (`NCT03719066`), openFDA, and PubMed/Europe PMC records with cross-document numeric consistency and explicit non-sponsor portfolio disclaimers in [`regulatory_writing_specimens/`](regulatory_writing_specimens/):

1. [Specimen #1 — ICH E3 CSR Synopsis (`NCT03719066`)](regulatory_writing_specimens/Specimen_01_ICH_E3_CSR_Synopsis_NCT03719066.md)
2. [Specimen #2 — EU CTR 536/2014 Plain Language Summary](regulatory_writing_specimens/Specimen_02_EU_CTR_Plain_Language_Summary_NCT03719066.md)
3. [Specimen #3 — ICH E6(R3) §7 Investigator's Brochure Update](regulatory_writing_specimens/Specimen_03_ICH_E6_Investigators_Brochure_Update.md)
4. [Specimen #4 — EU MDR Class IIa CER Skeleton (`CardioTrack-IIa`)](regulatory_writing_specimens/Specimen_04_EU_MDR_CER_Skeleton_CardioTrack-IIa.md)
5. [Specimen #5 — Medical Information Standard Response Document (SRD)](regulatory_writing_specimens/Specimen_05_Standard_Response_Document_SRD.md)
6. [Specimen #6 — PRISMA 2020 Systematic Review (`99.56%` `PMC13739473` Benchmark)](regulatory_writing_specimens/Specimen_06_PRISMA_2020_Systematic_Review.md)
7. [Specimen #7 — ICH E2F DSUR Executive Summary](regulatory_writing_specimens/Specimen_07_ICH_E2F_DSUR_Executive_Summary.md)
8. [Specimen #8 — ICH M11 Clinical Trial Protocol Synopsis](regulatory_writing_specimens/Specimen_08_ICH_M11_Protocol_Synopsis.md)
9. [Specimen #9 — ICH M4E(R2) CTD Module 2.5 Clinical Overview](regulatory_writing_specimens/Specimen_09_CTD_Module_2_5_Clinical_Overview.md)
10. [Specimen #10 — ICH M4E(R2) CTD Module 2.7.3 Clinical Efficacy](regulatory_writing_specimens/Specimen_10_CTD_Module_2_7_3_Clinical_Efficacy.md)
11. [Specimen #11 — ICH M4E(R2) CTD Module 2.7.4 Clinical Safety](regulatory_writing_specimens/Specimen_11_CTD_Module_2_7_4_Clinical_Safety.md)
12. [Specimen #12 — EU MDR Annex XIV Part B PMCF Evaluation Report](regulatory_writing_specimens/Specimen_12_EU_MDR_PMCF_Evaluation_Report.md)

## Quickstart

```bash
cd computational-biology-portfolio
pip install -r requirements.txt
python 01_ncbi_sequence_analyzer.py
python 02_clinical_trial_eda.py
python 03_pubmed_literature_synthesizer.py --query "cardiovascular implantable medical device safety" --max 30 --years 5 --output-dir dossier_sample
python -m unittest test_pilot_package.py -v
```

---
*MIT © 2026 Sentinel Regulatory & Technical Operations — Strict public-registry and synthetic-cohort traceability only. Zero patient PHI ingested.*
