Analyze the stability testing data in Columns A through H:
- Column A: Batch Number
- Column B: Storage Condition (e.g., 25°C/60% RH, 40°C/75% RH)
- Column C: Time Point (Months: 0, 3, 6, 9, 12)
- Column D: Active Pharmaceutical Ingredient (API) Assay % (Specification: 95.0% - 105.0%)
- Column E: Total Degradation Impurities % (Specification: NMT 1.5%)

Tasks:
1. Provide an Excel formula for Column F that flags "OUT OF SPEC (OOS)" if Assay < 95.0% or > 105.0%, or if Impurities > 1.5%. Otherwise output "PASS".
2. Write a formula to calculate the degradation rate (slope of impurities vs time) for each batch.
3. Suggest conditional formatting rules to highlight accelerating degradation before it breaches OOS limits.

In Column C, I have messy clinical site demographic entries that mix age and gender (e.g., "54 yr M", "Female, 62", "M - 41 years old", "Age: 29 / F").

Provide formulas or Power Query steps to:
1. Extract numerical AGE into Column D as an integer.
2. Standardize GENDER into Column E strictly as either "M", "F", or "UNKNOWN".
3. Provide a data validation formula for Column F that flags rows where Age is outside the protocol inclusion criteria (18 to 75 years inclusive).

Using the attached Phase IIb Efficacy Summary for Compound [XYZ-101], generate an outline for a high-impact 6-slide executive presentation for the Senior Scientific Advisory Board.

Structure each slide with:
- Slide Title (action-oriented, e.g. "Primary Endpoint Met with Statistically Significant Efficacy")
- Visual Concept (describe recommended chart: Kaplan-Meier curve, waterfall plot, or 3-column metric cards)
- 3 Bullet Points (succinct, bold key metric first, max 15 words per bullet)
- Speaker Talking Points (2-3 sentences providing regulatory and clinical context)

Slides required:
Slide 1: Study Design & Patient Demographics (n=450)
Slide 2: Primary Efficacy Endpoint (p-value, hazard ratio, 95% CI)
Slide 3: Secondary Biomarker Outcomes
Slide 4: Safety & Tolerability Profile (treatment-emergent adverse events)
Slide 5: Competitive Landscape & Differentiation
Slide 6: Phase III Recommendation & Regulatory Timeline

Create a 4-slide competitive intelligence briefing based on the newly published FDA approval package for [Competitor Drug Name].

Focus on:
1. Mechanism of Action (MoA) Comparison vs our in-house molecule.
2. Dosing Convenience & Administration Routes (Oral QD vs SC Bi-weekly).
3. Boxed Warnings or Special Monitoring Requirements.
4. Opportunities for our medical affairs positioning.

Make each slide clean, professional, and suitable for commercial sales and medical science liaisons (MSLs).

You are an automated Pharmacovigilance (PV) data extraction node in a high-throughput GxP ETL pipeline.

Your task is to ingest unstructured patient incident reports or physician notes and output strictly validated, parseable JSON conforming to ICH E2A safety guidelines.

INPUT TEXT:
"""
[Insert raw clinician note / EHR excerpt]
"""

EXTRACTION RULES:
1. Extract every reported adverse event into an array under the key "adverse_events".
2. For each event, include:
   - "verbatim_term": Exact symptom phrase quoted directly from the input text.
   - "suspect_drug": Trade name or active ingredient mentioned, or "NOT_SPECIFIED".
   - "severity": Must be strictly one of ["MILD", "MODERATE", "SEVERE", "LIFE_THREATENING"].
   - "causality_assessment": Strictly one of ["UNRELATED", "POSSIBLE", "PROBABLE", "DEFINITE"].
   - "action_taken_with_drug": Strictly one of ["DOSE_NOT_CHANGED", "DOSE_REDUCED", "DRUG_INTERRUPTED", "DRUG_WITHDRAWN", "UNKNOWN"].
   - "regulatory_seriousness": Boolean (true if hospitalization, disability, life-threatening, or death occurred; false otherwise).
   - "confidence_score": Decimal between 0.0 and 1.0 indicating extraction certainty.
3. If no adverse events are detected, return an empty array for "adverse_events".
4. Set "requires_expedited_15_day_review": true if any event meets regulatory expedited criteria.

OUTPUT CONSTRAINT:
Output ONLY raw, valid JSON with zero conversational filler, zero markdown backticks, and zero preamble.

You are an audio intelligence and clinical documentation node in an automated medical speech-to-text pipeline.

You will ingest raw, noisy speech-to-text transcripts generated from investigator meetings, advisory boards, or patient monitoring calls.

RAW TRANSCRIPT:
"""
[Insert raw Whisper audio transcript here]
"""

TRANSFORMATION DIRECTIVES:
1. Medical Vocabulary Normalization:
   - Detect phonetically misheard pharmaceutical terms and correct them to standard INN (International Nonproprietary Names) or trade names (e.g., correct "pembro" or "keytruder" to "Keytruda / pembrolizumab", "humira" to "adalimumab", "ozempick" to "Ozempic / semaglutide").
   - Standardize clinical trial abbreviations (AUC, Cmax, OOS, SAE, MedDRA, eCRF).

2. Automated PHI / PII Redaction (HIPAA Safe Harbor):
   - Replace any identifiable patient names, hospital room numbers, physician phone numbers, dates of birth, or Medical Record Numbers (MRNs) with standard redaction tokens like [PATIENT_NAME_REDACTED], [MRN_REDACTED], [PHONE_REDACTED].

3. Structured Clinical Synthesis:
   - Provide an Executive Scientific Summary (bulleted trial efficacy, pharmacokinetics, and safety observations).
   - Document any Unresolved Protocol Questions or Investigator Concerns.
   - Output an Action Item Table with columns: [Action Item, Responsible Function (QA / Clinical Ops / Data Mgmt), Target Milestone].

OUTPUT FORMAT:
Return clean, audit-ready Markdown with strict separation between clinical facts and operational deliverables.