Dr. Avinash kumar gupta

VibeRounds — Master Case Analysis Protocol

Complete Workflow: From Case Link to Full Clinical Intelligence Output

Framework Source: VibeRounds Combined Modules 1–20 Protocol Version: 1.0 — June 2026 Validated Against: 60F with Coma, E Coli Sepsis, Cervical Myelopathy, and Albumino-Cytological Dissociation in CSF

How to Use This Protocol

Upload this file to Claude (or any capable LLM) along with the VibeRounds Modules 1–20 file.
Paste a case link (public blog, open patient record, or de-identified case narrative URL) in your message.
Send the trigger prompt at the bottom of this document.
The AI will execute all six pipeline stages in sequence and deliver three output .md files:
- VibeRounds-TopInsights-[CaseName].md — Top 10 clinical insights (standalone, shareable summary)
- VibeRounds-CaseAnalysis-[CaseName].md — Prompt mapping, ranking, and full high-value prompt answers
- VibeRounds-CARE-AdvocateDebrief-[CaseName].md — CARE case report and advocate debrief

Data Safety Note: Only use de-identified or consent-obtained public case records. Do not paste identifying information (full name, date of birth, address, hospital number) into any LLM session. This protocol is an educational tool, not a clinical decision-making service. All outputs require independent clinical verification before any action is taken.

Pipeline Overview

INPUT
  ├── VibeRounds Modules 1–20 (uploaded file)
  └── Case link or narrative (provided by user)
         │
         ▼
  STAGE 1 — Case Ingestion
  Read and extract structured clinical details from the case source
         │
         ▼
  STAGE 2 — Prompt Mapping
  Match VibeRounds prompts from all 20 modules to the patient's clinical features
         │
         ▼
  STAGE 3 — Prompt Ranking
  Rank all matched prompts 1–10 on clinical importance for this specific case
         │
         ▼
  STAGE 4 — High-Value Prompt Execution
  Answer all prompts rated 8–10 in full clinical depth
         │
         ▼
  STAGE 5 — Insight Synthesis
  Extract the top 10 clinical insights from all analysis above
         │
         ▼
  STAGE 6 — CARE Report + Advocate Debrief
  Produce CARE-format case write-up and structured advocate journey analysis
         │
         ▼
OUTPUT
  ├── File 1: VibeRounds-TopInsights-[CaseName].md
  ├── File 2: VibeRounds-CaseAnalysis-[CaseName].md
  └── File 3: VibeRounds-CARE-AdvocateDebrief-[CaseName].md

Stage Specifications

STAGE 1 — Case Ingestion

Objective: Extract a complete, structured summary of the clinical case from the provided source.

Instructions for AI:

Fetch the case from the provided URL using web_fetch or read the pasted narrative
Extract and structure the following fields, leaving any field explicitly marked [NOT DOCUMENTED] if absent:

Patient demographics: age, sex, occupation, social/geographical background
Chief complaint and mode of presentation
Background history (duration, prior diagnoses, prior treatments)
Symptom timeline (chronological — when each symptom appeared or changed)
Medications (name, dose, frequency, route, duration, any self-modifications)
Examination findings (vitals, systemic, neurological, others)
Investigations (results, dates where available, units)
Procedures performed
Working diagnoses at presentation
Management given
Outcome
Investigations NOT performed but clinically indicated
Patient/advocate narrative (if present)

Quality gate before proceeding: Confirm the case has enough clinical content to support at least 8 usable VibeRounds prompts. If not, flag this and ask the user for additional case detail before continuing.

STAGE 2 — Prompt Mapping

Objective: Identify every VibeRounds prompt across Modules 1–20 that is directly applicable to this patient’s clinical features.

Instructions for AI:

Review all 20 modules systematically
For each module, identify which specific steps/prompts are triggered by features present in this case
A prompt is “usable” if it would generate clinically meaningful output specific to this patient — not generic educational content
Present as a table with columns: # | Module | Step | Prompt Purpose | Patient-Context Trigger
Minimum 15 usable prompts expected for a complex multi-system case
For single-system or simpler cases, minimum 8 prompts

Mapping triggers by module (reference guide):

Module	Triggered when case has…
M1 — Socratic Clinical Reasoning	Any diagnostic uncertainty; reasoning gaps; learning context
M2 — Patient-Advocate Documentation	Family/caregiver narrative present; non-medical documentation of course
M3 — Extended Monitoring	Longitudinal deterioration over weeks/months documented
M4 — Ward Round Preparation	Acute admission; multi-system complexity; handover moments
M5 — Real-Time Case Review	Longitudinal data (glucose logs, medication logs, serial bloods)
M6 — Registry Analytics	Population-level patterns implied; case belongs to a known registry
M7 — Cross-Case Learning	Similar cases exist or are mentioned; registry context
M8 — Socratic Design QA	Session being designed for teaching; prompt quality review needed
M9 — N-of-1 Research Protocol	Complex, multi-system, unusual case suitable for case report
M10 — Article Reading	Specific paper directly relevant to case findings
M11 — Patient Education	Patient/family information needs visible in narrative
M12 — Differential Diagnosis Deepdive	Working diagnosis formed but contested; anchoring risk present
M13 — Polypharmacy Audit	5+ medications; drug-disease conflicts; prescribing cascade risk
M14 — Resource-Constrained Reasoning	Investigations unavailable; transport limitations; low-resource setting
M15 — Illness Script Acquisition	Typical/atypical presentation; script mismatch visible
M16 — Basic Science Integration	Mechanism-to-diagnosis link important for this case
M17 — Semantic Qualifiers	Problem representation needs to be corrected or sharpened
M18 — Causal Network Reasoning	Multiple findings interact conditionally; one finding changes weight of another
M19 — Community & Social Medicine	Social determinants of health visible; occupation/environment relevant
M20 — Recognition-Primed Decision	Time-critical moment present; acute deterioration; stat call scenario

STAGE 3 — Prompt Ranking

Objective: Rank all mapped prompts by clinical importance for this specific case.

Instructions for AI:

Score each prompt 1–10 using the following criteria:
- 10 — Directly addresses the central diagnostic or management failure in the case; answering this prompt would change what a clinician does
- 8–9 — Addresses a high-stakes clinical question, safety risk, or major cognitive error present in this case
- 6–7 — Adds meaningful clinical context; addresses a secondary diagnostic or management question
- 4–5 — Useful for teaching or learning; does not directly alter management
- 1–3 — Applicable in principle but low additional yield for this specific patient
Present as a table with columns: Rank | Score | Prompt | Justification
Sort descending by score
All prompts scored 8–10 are automatically promoted to Stage 4 execution

STAGE 4 — High-Value Prompt Execution

Objective: Answer every prompt scored 8–10 in full clinical depth, applied specifically to this patient.

Instructions for AI:

Address each high-value prompt as a named section with the prompt title, score, and module reference
Each answer must be patient-specific — no generic textbook answers
Use tables, structured differentials, clinical reasoning chains, and mechanism explanations where they add clarity
Where the prompt requires adversarial reasoning (Module 12), apply the full devil’s advocate frame
Where the prompt requires causal network reasoning (Module 18), explicitly map how each finding conditionally changes the weight of others
Where the prompt involves pharmacology (Module 13), produce a structured drug-disease conflict table
Where the prompt involves time-critical decision-making (Module 20), produce the recognition → plan → forward simulation → escalation threshold structure

Minimum answer length per high-value prompt: Sufficient to be clinically actionable. A bulleted list of five words is not an answer. A paragraph that names the mechanism, the clinical consequence, and the action is the minimum standard.

STAGE 5 — Insight Synthesis

Objective: Extract the top 10 clinical insights from all analysis above.

Instructions for AI:

Insights must be case-specific — not general medical knowledge
Each insight should be one of: a diagnostic finding that was missed or underweighted, a cognitive bias that operated, a pharmacological risk that was unrecognised, a clinical signal that was misinterpreted, a system-level failure, or a decision point where investigation or management should have pivoted
Number 1–10; order from most clinically important to least
Each insight: 3–5 sentences — name the finding, explain why it was significant, state what the correct response would have been

STAGE 6 — CARE Report + Advocate Debrief

Objective: Produce two structured documents — a CARE-format case report and a structured advocate journey debrief.

6A — CARE-Format Case Report

Follow all 12 CARE guideline fields in sequence:

1. Title Format: [Age][Sex] with [chief complaint(s)] and [key investigation finding]: [educational contribution of case]

2. Abstract Four paragraphs: Background (why this case matters clinically) | Case Summary (≤150 words) | Key Learning Points (numbered, minimum 3) | Conclusion (one sentence on the educational contribution)

3. Introduction Why this case is clinically and educationally important. What gap in practice it exposes. What it contributes that existing literature does not.

4. Patient Information Structured table: demographics, occupation, social/geographical background, comorbidities, prior diagnoses.

5. Clinical Findings Two sub-sections:

5a: Presenting symptoms (bulleted, with duration)
5b: Examination findings (table: parameter finding clinical significance)
5c: Investigations (table: investigation result interpretation; explicitly flag missing investigations)

6. Timeline Chronological flow diagram or structured timeline from first symptoms to outcome, with each major clinical event, decision point, and missed pivot labelled.

7. Diagnostic Assessment

Working diagnoses at time of death/discharge
Unestablished diagnoses — differential table (diagnosis supporting evidence against investigation required)
The central diagnostic question: what remained unanswered, and why

8. Therapeutic Interventions Table: intervention | timing | rationale | outcome. Include a separate row for each intervention that was NOT initiated but was indicated, with justification.

9. Follow-Up and Outcomes What happened. If fatal: what post-mortem data was or was not obtained, and what that means for learning.

10. Discussion Three to four named teaching points, each structured as: what happened → what should have happened → why the gap occurred → what would prevent it.

11. Patient Perspective Reconstructed from any patient or advocate narrative in the case source. If none exists, mark as [Not documented — patient perspective unavailable].

12. Informed Consent Statement Source and consent status of the case material.

6B — Advocate Debrief

Structure:

Opening: What is an advocate debrief? One paragraph defining its purpose — not criticism of the family, but clinical systems analysis of what healthcare encounters failed to provide.

The Advocate’s Role: What specific roles did the family caregiver play in this case? (Historian, medication administrator, wound care coordinator, decision-maker, etc.)

Inflection Point Analysis: For each major clinical turning point in the case (minimum 5 inflection points), produce:

### Inflection Point [N] — [Name of moment]
**What happened:** [clinical event]
**What the advocate understood:** [their likely interpretation]
**What the advocate needed to know:** [specific information, in plain language]
**The question the advocate needed to ask:** [exact wording they could have used]
**VibeRounds module applied:** [Module X, Step Y]
[Module 11-style red flag table or Module 2-style information brief where applicable]

What the Advocate Did Well: Explicit acknowledgement of what the family managed correctly — this section is mandatory and must not be omitted.

Advocate Learning Summary: Apply Module 2 Step 2.8 (Bloom’s Remember → Understand → Apply) as three questions and model answers the advocate should have been able to answer at discharge.

Recommendations for Future Similar Cases: Four to six specific, actionable recommendations — for the clinical team, for the prescribing clinician, for the discharge process, and for the healthcare system.

Output File Specifications

File 1: `VibeRounds-TopInsights-[CaseName].md`

Purpose: A standalone, shareable summary of the most important clinical learning from this case. Designed to be read independently — without needing to open the full analysis file.

Contains:

Case link on line 3 (immediately after title and subtitle)
One-sentence case identifier (age, sex, key diagnoses)
Top 10 clinical insights from Stage 5, numbered 1–10 in order of clinical importance
Each insight formatted as:
- Insight [N]: [Short title]
- Finding: what was observed or missed
- Significance: why it mattered clinically
- Correct response: what should have happened
Footer: VibeRounds disclaimer and reference to companion Files 2 and 3

File 2: `VibeRounds-CaseAnalysis-[CaseName].md`

Contains:

Case link on line 3 (immediately after title and subtitle)
Case summary (structured, from Stage 1)
Workflow description (one paragraph per stage)
Section 1: Full case summary table
Section 2: Prompt mapping table (Stage 2)
Section 3: Ranked prompt table (Stage 3)
Section 4: High-value prompt answers (Stage 4) — one named subsection per prompt
Footer: VibeRounds disclaimer and reference to companion Files 1 and 3

File 3: `VibeRounds-CARE-AdvocateDebrief-[CaseName].md`

Contains:

Case link on line 3
Reference to Files 1 and 2 as companion documents
Part A: Full CARE case report (all 12 fields)
Part B: Full advocate debrief (all sections)
Footer: VibeRounds disclaimer

Naming convention: [CaseName] = abbreviated identifier derived from the case (e.g., 60F-EColi-Sepsis, 45M-DKA-Tropical, 72F-Dementia-Falls). Use age + sex + two to three key clinical features, hyphen-separated, no spaces.

Quality Checks Before Delivering Output

Before writing any output file, the AI must confirm:

Trigger Prompt

Copy and send this prompt to Claude (with this protocol file and the VibeRounds Modules 1–20 file uploaded):

#VibeRounds Run the full Master Case Analysis Protocol on the following case:

[PASTE CASE LINK OR CASE NARRATIVE HERE]

Execute all six pipeline stages in sequence:
Stage 1 — Case ingestion and structured extraction
Stage 2 — Prompt mapping across all 20 modules
Stage 3 — Prompt ranking by clinical importance (1–10)
Stage 4 — Full answers to all prompts scored 8–10
Stage 5 — Top 10 clinical insights
Stage 6 — CARE-format case report (all 12 fields) + Advocate debrief (all sections)

Deliver three output .md files:
File 1: VibeRounds-TopInsights-[CaseName].md
File 2: VibeRounds-CaseAnalysis-[CaseName].md
File 3: VibeRounds-CARE-AdvocateDebrief-[CaseName].md

Follow all quality checks in the protocol before writing the output files.
State which stage you are in before beginning each one.
Do not collapse stages or skip ahead.

What This Protocol Is and Is Not

This protocol IS:

A structured educational framework for deep clinical case analysis
A tool for applying the full VibeRounds cognitive library to individual patient cases
A method for producing publishable-quality case reports and caregiver education briefs
A system for surfacing cognitive biases, diagnostic gaps, and missed clinical pivots in retrospective case review

This protocol IS NOT:

A clinical decision support system
A substitute for licensed clinical judgment
A diagnostic oracle — the AI generates hypotheses and frameworks, not diagnoses
A tool to be used with identifiable patient data without appropriate consent and de-identification

Protocol Maintenance Notes

This protocol was developed and validated in June 2026 against one complex multi-system case. As additional cases are run through the pipeline, the following should be updated:

Mapping triggers table (Stage 2): Add new module-to-case-feature mappings as they emerge from novel case types
Ranking criteria (Stage 3): Calibrate score thresholds against clinical outcomes if follow-up data becomes available
Quality checklist: Add new checklist items as output gaps are identified
Naming convention: Extend if case types expand beyond single-admission presentations (e.g., longitudinal monitoring cases from Module 3)

VibeRounds Master Case Analysis Protocol v1.0 — June 2026 All outputs generated using this protocol are educational. Independent clinical verification is required before acting on any content. This protocol does not constitute clinical advice, diagnostic guidance, or a substitute for professional medical judgment.