NREHub

MCQ Production Plan — Fresh Sets to Oct 11 2026

The reference artifact is QBank/Mock Tests/ENT-Mock-Fresh-Set-25 MCQ 25.json

3,522 words ~16 min

MCQ Production Plan — Fresh Sets to Oct 11 2026

Purpose: produce MORE MCQs in the exact style of the existing custom sets (ENT-Mock-Fresh-Set-25 + MedCORE Production MCQ pipeline), aligned to the 60-Day Battle Plan, NRE50 rotation, confirmed recall intelligence, and the official NRE 2024 weightage. This is a production plan — not study content. Generated 2026-08-19 · Day 8 · Block 2 (GI + Hepatology).

1. What “similar” means — the format blueprint

The reference artifact is QBank/Mock Tests/ENT-Mock-Fresh-Set-25 MCQ 25.json (+ its Unsolved/Answers DOCX pair and Practice HTML). Every new set must match this blueprint exactly:

JSON schema (per question)

Field Requirement
id SUBJ-SET-NN (e.g. ENT-FS-01)
stem Clinical vignette: age + sex + presentation + key finding + single question line
options Exactly 4 (A–D), single best answer, no “all of the above”
correctAnswer Letter A–D
testedConcept One named concept, e.g. “BPPV — diagnosis and treatment”
cognitiveTask Recall / Interpretation / Analysis-synthesis (label precisely)
discriminator The single data point that separates right from wrong
whyCorrect 1–2 sentence clinical reasoning connecting discriminator → answer
trapType One named trap per question (see §1.3)
futureAlert One-line trigger phrase for next encounter
reviseTopic Topic to revise if missed
syllabusTopic Official syllabus tag (e.g. “ENT — Vertigo and balance disorders”)
correctRule One-line rule to remember
distractors Dict: per-option one-line why-wrong
distractorExplanations List mirroring options order
errorLogRecord Pre-filled record ready to paste into Error Log

Set-level rules (from the ENT set’s actual distribution)

  • 25 MCQs per set, 4 options each.
  • Cognitive mix per set: ~15% Recall, ~45% Interpretation, ~30% Analysis/synthesis, ~10% mixed Recall/interpretation. Pure-recall sets are forbidden — the exam targets “higher cognition.”
  • Topic spread: cover the subject’s FULL syllabus in one set (the ENT set hit 22 distinct syllabus topics across ear/nose/throat/larynx/neck/salivary).
  • One unique trap type per question — no repeated trap within a set.
  • Every question must have a named discriminator and future alert. No question without all three (discriminator + trap + alert) ships.
  • Vignettes must be realistic NRE register: age, sex, duration, one physical sign or one investigation result, then “most likely diagnosis / most appropriate next step / best first-line treatment / gold standard.”

Trap taxonomy (tag from this list — matches Error Log README)

Partial Truth · Opposite · Synonym Swap · Overthinking · Unit/Number · Guideline Version · Clinical Mimic · Treatment misdirection · Investigation hierarchy · Organism homogeneity · Severity under-recognition · Priority inversion · “What NOT to do” · Age-appropriate differential · Iatrogenic trap · Negative discrimination · Treatment ladder · Pattern recognition.

2. Current inventory vs NRE weightage — the gap

Source: MedCORE Production/07 MCQ Source JSON (2246 MCQs) + ENT fresh set.

Subject Exam MCQs (per 200) Current sets Current MCQs Proportional target* Gap
Medicine & Allied 52 28 sets 700 700 (benchmark) —
Surgery & Allied 48 21 sets 525 646 −121
Mock Exams (full mocks) — 2 396 — needs Mock C
Paediatrics 14 5 125 188 −63
Obstetrics & Gynaecology 14 5 125 188 −63
Applied Pathology 11 2 50 148 −98
Applied Anatomy 11 2 50 148 −98
Applied Physiology 10 2 50 135 −85
Clinical Pharmacology 8 2 50 108 −58
ENT 6 2 50 81 −31
Ophthalmology 6 1 25 81 −56
Applied Biochemistry 6 1 25 81 −56
Community Medicine 6 1 25 81 −56
Forensic Medicine 4 1 25 54 −29

*Proportional target = Medicine’s 700 scaled by exam weightage ratio.

Verdict: Medicine and Surgery are healthy. Every other subject is under-covered. The small subjects (Ophtho, Biochem, CM, Forensic, ENT) and the applied basics (Path, Anat, Physio, Pharm) are the real gaps — and they are exactly the blocks where the May 2026 attempt leaked marks.

3. Grounding rules — what every set must be built on

Before writing a single question, the generator MUST consult, in order:

  1. Tier-1 recalls — Past Papers/ (NRE 2026 MAY, Dec 2025) and QBank/TXT/PMDC/NRE Recalled 189.txt. If the topic appeared, mirror its exact discriminator and disguise.
  2. Intelligence Report — Exam Intel/Intelligence Reports/ NRE_Intelligence_Report_May2026.docx (116 confirmed topics with discriminators + traps). New MCQs should re-test these discriminators.
  3. Prediction sheets — Exam Intel/Predictions/Prediction_Sheet_1_ Oct2026_High_Yield_Topic_List.docx (recognition trigger + discriminator + trap per topic).
  4. NRE50 rotation map — Exam Intel/NRE50 System/Rotation Master Map.pdf (which topics belong to which quiz slot).
  5. Existing QBank — check the subject’s separated bank + existing JSON sets to avoid duplicating stems. New sets must be fresh, not re-skins.
  6. Syllabus — Syllabus/NRE_Syllabus_2024.docx for official topic names (used in syllabusTopic tags).

Rule: never let generic medical completeness override recalled NRE pattern evidence. If a recall says the exam tests the discriminator a specific way, that is the way the new MCQ tests it.

4. Generation workflow (per set)

1. SELECT topic set   → from backlog (§5), battle-plan block, NRE50 rotation
2. GROUND in evidence → pull discriminators/traps from recalls + intel + predictions
3. CHECK for dupes    → scan existing JSON sets + separated banks for the topic
4. WRITE 25 MCQs      → JSON schema from §1, cognitive mix from §1.2
5. SELF-QA            → every Q has discriminator + trap + alert; answer key
                        internally consistent; no repeated trap; topic spread
6. RENDER DOCX pair   → Unsolved + Answers via existing pipeline
                        (MedCORE Production generate_mcq_docx.js /
                        build_mcq_qa_docx.py; Scripts/build_mcq_docx.py)
7. DELIVER            → QBank/Mock Tests/ (fresh sets) or MedCORE Production/
                        Source JSON/MCQ Sets/<Subject>/<Topic>/ (day sets)
8. FEED THE LOOP      → wrong answers from practice → Error Log/YYYY-MM-DD.md
                        (trap type + discriminator + future alert)

Delivery convention (matches memory + existing artifacts):

  • Unsolved and Answers in SEPARATE files, never merged.
  • Naming: <Subject>-Mock-Fresh-Set-N MCQ 25 Unsolved.docx / ... MCQ 25 Answers.docx + ... MCQ 25.json + optional Practice.html.
  • Place fresh sets in QBank/Mock Tests/.

5. Production backlog — block-aligned to the 60-Day Battle Plan

30 fresh 25-MCQ sets (750 MCQs) + NRE50 Set 2 (200) + Mock C (200) = ~1,150 new MCQs by Oct 10. Each set is a single agent task.

Block 2 · GI + Hepatology + Nutrition (Aug 19–23) — NOW

# Set Subject Source grounding
1 GI Emergencies (UGIB, varices, perforation) Medicine/Surgery Varices→banding, perforated PU→laparotomy (prediction #10, #13) · ✅ delivered 2026-08-19
2 Hepatology (cirrhosis complications, hepatitis, pancreatitis) Medicine NH₃→glutamine, falling albumin (battle HOT) · ✅ delivered 2026-08-19
3 Nutrition + Malabsorption Medicine anti-tTG, PEM (battle HOT) · ✅ delivered 2026-08-19

Block 3 · Renal + Urology + Acid-Base (Aug 24–29)

# Set Subject Source grounding
4 AKI / CKD / Glomerular diseases Medicine APSGN timing trap, IgA during URTI (prediction #5, #6) · ✅ delivered 2026-08-19
5 Electrolytes + Acid-Base Applied Physiology Metabolic alkalosis—vomiting (intel #8) · ✅ delivered 2026-08-27 (REBUILT — v1 failed QA: 8 topics, B:15 key)
6 Urology (stones, obstruction, prostate, testis) Surgery Testicular torsion (prediction #9) · ✅ delivered 2026-08-27 (Block 3 COMPLETE)

Block 4 · Endocrine + DM + Endocrine Surgery (Aug 30–Sep 4)

# Set Subject Source grounding
7 Endocrine (Thyroid / Adrenal / Pituitary) Medicine TRAb+exophthalmos, pheochromocytoma metanephrines · ✅ delivered 2026-08-27
8 DM + complications + emergencies Medicine DKA→insulin now, metformin first-line (battle HOT) · ✅ delivered 2026-08-27 (fresh-angle set; DKA/HHS/metformin-CKD/sulfonylurea/SGLT2/ACEi avoided — saturated)
9 Endocrine Surgery (thyroid, parathyroid, adrenal) Surgery RLN+Chvostek (battle HOT) · ✅ delivered 2026-08-27 (Block 4 COMPLETE)

Block 5 · Neuro + Psych + Eye + ENT (Sep 5–10) — small-subject surge [✅ COMPLETE]

# Set Subject Source grounding
10 Neurology (stroke, coma, epilepsy, headache) Medicine SAH normal CT→LP (battle HOT) · ✅ delivered 2026-08-27
11 Psychiatry Medicine Anxiety/depression/psychosis pattern · ✅ delivered 2026-08-27
12 Ophthalmology Fresh Set 1 Ophthalmology Pilocarpine/glaucoma (intel), ACG→mannitol, chemical eye→irrigate · ✅ delivered 2026-08-30 (Block 5)
13 ENT Fresh Set 2 ENT Epiglottitis airway-first, nasal FB (intel) · ✅ delivered 2026-08-31 (Block 5 COMPLETE)

Block 6 · Haematology + Oncology + Gen Path (Sep 11–16) [✅ COMPLETE]

# Set Subject Source grounding
14 Haematology (anaemias, leukaemias, lymphomas) Medicine Auer rods=AML, CML BCR-ABL, ferritin-first (intel #3, #4) · ✅ delivered 2026-08-31 (Block 6)
15 Oncology + General Pathology Applied Pathology Hypersensitivity types, Barrett’s metaplasia (intel #1, #2) · ✅ delivered 2026-08-31 (Block 6 COMPLETE)
16 Applied Pathology Fresh Set Applied Pathology Type III vs IV, AML/ALL cross-wire

Block 7 · Surgery I: Acute Abdomen + Trauma + Shock (Sep 17–22)

# Set Subject Source grounding
17 Trauma + Burns Surgery Parkland half-in-8h (battle HOT)
18 Acute Abdomen + Shock Surgery Littre/Amyand/Richter, Charcot triad (battle HOT)

Block 8 · Surgery II: Airway + Vascular + Colorectal (Sep 23–28)

# Set Subject Source grounding
19 Vascular + Airway Emergencies Surgery 6 Ps limb ischaemia, MH→dantrolene (battle HOT)
20 Colorectal + Hernias Surgery apple-core, femoral strangulation (battle HOT)

Block 9 · OBG I + Paeds I (Sep 29–Oct 4)

# Set Subject Source grounding
21 Obstetrics Fresh Set ObGyn Anti-D 72h, praevia vs abruption, pre-eclampsia labetalol (prediction #2)
22 Paediatrics Fresh Set 1 Paediatrics Neonatal 3:1, CPR ratio, congenital (intel)

Block 10 · OBG II + Paeds II + CommMed + Forensic (Oct 5–10)

# Set Subject Source grounding
23 Gynaecology Fresh Set ObGyn Fibroids, GTN, contraception (intel)
24 Paediatrics Fresh Set 2 Paediatrics Immunisation, vaccine-preventable, poisoning
25 Community Medicine + Biostatistics Fresh Set CM Vaccines, median, doubling time (intel + recalled Q1)
26 Forensic Medicine + Toxicology Fresh Set Forensic Kerosene ingestion (intel), antidotes
27 Applied Biochemistry Fresh Set Biochemistry G6PD, PKU, porphyrins, vitamins
28 Applied Anatomy Fresh Set Anatomy Nerve lesions: foot drop, wrist drop, axillary (intel #5–7)
29 Applied Physiology Fresh Set Physiology LH surge, acid-base, RAAS
30 Clinical Pharmacology Fresh Set Pharmacology Antidotes, drug of choice pattern (intel)

Assembly tasks (from the fresh pool)

Task MCQs When
NRE50 Set 2 — Q1 (53) / Q2 (47) / Q3 (53) / Q4 (47) 200 Build after Block 5; Q2/Q4 = harder vignettes (atypical presentations, contraindications, mechanism, negative stems)
Mock C — Full 200, exam weightage 200 Assemble by Day 60 (Oct 10) from all fresh sets

6. Quality bar — success criteria per set

A set SHIPS only when all pass:

  • 25 MCQs, 4 options each, single best answer, no “all of the above”
  • Cognitive mix: ≤15% pure recall; ≥75% interpretation/analysis
  • Every question has discriminator + trapType + futureAlert (non-empty)
  • No trap type repeats within the set
  • Topic spread covers the subject’s full syllabus (≥15 distinct syllabusTopic tags)
  • No stem duplicates or near-duplicates vs existing QBank/JSON sets
  • Answer key verified: correctAnswer matches distractorExplanations
  • errorLogRecord pre-filled for every question
  • Unsolved + Answers delivered as SEPARATE DOCX files (+ JSON)
  • Source grounding documented: which recall/intel/prediction each high-yield Q mirrors

7. Generation Prompt Template — copy-paste per set

Hand this to any agent (or use it yourself) for each backlog item. Fill the four <PLACEHOLDER> fields. The exemplar is QBank/Mock Tests/ ENT-Mock-Fresh-Set-25 MCQ 25.json — match its structure exactly.

ROLE: NRE-I MCQ writer. You produce exam-style MCQs in the exact format of
the reference file `QBank/Mock Tests/ENT-Mock-Fresh-Set-25 MCQ 25.json`.

TASK: Create a fresh set of 25 MCQs for:
  SUBJECT: <SUBJECT, e.g. Medicine & Allied>
  TOPIC RANGE: <TOPIC RANGE, e.g. GI Emergencies: UGIB, varices, perforation>
  SYLLABUS TAG PREFIX: <e.g. "Medicine — ">

GROUNDING (do this BEFORE writing, in order):
1. Read `Past Papers/` recalls and `QBank/TXT/PMDC/NRE Recalled 189.txt`
   for this topic. If the topic appeared, mirror its discriminator and disguise.
2. Read `Exam Intel/Intelligence Reports/NRE_Intelligence_Report_May2026.docx`
   confirmed-topic list. Re-test those exact discriminators.
3. Read `Exam Intel/Predictions/Prediction_Sheet_1_Oct2026_High_Yield_Topic_List.docx`
   for recognition triggers + traps.
4. Check `MedCORE Production/07 MCQ Source JSON/` and the subject's separated
   bank in `QBank/Mock Tests/Separated/subjects/` — DO NOT duplicate stems.
5. Tag every question's `syllabusTopic` with the official name from
   `Syllabus/NRE_Syllabus_2024.docx`.

FORMAT (per question, exact JSON keys as in the ENT exemplar):
  id, stem, options (exactly 4, A-D, single best answer, no "all of the
  above"), correctAnswer, testedConcept, cognitiveTask (Recall /
  Interpretation / Analysis-synthesis), discriminator (the single data point
  separating right from wrong), whyCorrect (1-2 sentences), trapType (one
  named trap), futureAlert (one-line trigger phrase), reviseTopic,
  syllabusTopic, correctRule (one line), distractors (dict, per-option
  why-wrong), distractorExplanations (list in option order), errorLogRecord
  (pre-filled: date, source, questionId, subjectSystem, syllabusTopic,
  correctAnswer, missedDiscriminator, trapType, correctRule, futureAlert).

STEM STYLE: clinical vignette in NRE register — age + sex + presentation +
duration + one physical sign or investigation result + single question line
("most likely diagnosis" / "most appropriate next step" / "best first-line
treatment" / "gold standard investigation"). No trick stems, no negatives
unless the topic demands it (e.g. "which is NOT").

SET-LEVEL RULES:
- 25 MCQs, 4 options each.
- Cognitive mix: ~15% Recall, ~45% Interpretation, ~30% Analysis/synthesis,
  ~10% mixed. Pure-recall sets are forbidden.
- Topic spread: cover the FULL subject syllabus or the full topic range —
  aim for >=15 distinct syllabusTopic tags per set.
- One unique trapType per question; no trap repeats within the set.
- Every question ships with discriminator + trapType + futureAlert.
- Use the trap taxonomy from the Error Log README: Partial Truth, Opposite,
  Synonym Swap, Overthinking, Unit/Number, Guideline Version, Clinical Mimic,
  Treatment misdirection, Investigation hierarchy, Organism homogeneity,
  Severity under-recognition, Priority inversion, "What NOT to do",
  Age-appropriate differential, Iatrogenic trap, Negative discrimination,
  Treatment ladder, Pattern recognition.

DELIVERABLES (separate files, never merged):
1. `<Subject>-Mock-Fresh-Set-N MCQ 25.json` — full data
2. `<Subject>-Mock-Fresh-Set-N MCQ 25 Unsolved.docx` — questions only
3. `<Subject>-Mock-Fresh-Set-N MCQ 25 Answers.docx` — questions + correct
   answer + whyCorrect + per-option distractor explanations
4. Optional: `<Subject>-Mock-Fresh-Set-N Practice.html`

SELF-QA BEFORE DELIVERING (all must pass):
- correctAnswer matches distractorExplanations for every question
- No repeated trapType within the set
- Cognitive mix within the band above
- No stem duplicates vs existing banks
- errorLogRecord complete for every question
- Source grounding documented: list which recall/intel/prediction each
  high-yield question mirrors

Worked example — Set #1 (GI Emergencies), 2 of 25 questions

Produced by the template above, grounded in the Oct 2026 Prediction Sheet (prediction #13 varices→banding, #10 perforated PU→laparotomy). This is the exact format each backlog item must ship in.

{
  "id": "MED-GI-01",
  "stem": "A 55-year-old man with cirrhosis presents with massive hematemesis. Upper GI endoscopy shows tortuous dilated veins in the distal esophagus. What is the first-line therapeutic intervention?",
  "options": [
    "A) Endoscopic band ligation",
    "B) Sclerotherapy",
    "C) Octreotide infusion",
    "D) Balloon tamponade"
  ],
  "correctAnswer": "A",
  "testedConcept": "Esophageal varices — first-line treatment of acute hemorrhage",
  "cognitiveTask": "Interpretation (management hierarchy in variceal bleed)",
  "discriminator": "Cirrhosis + hematemesis + distal esophageal veins → variceal bleed → band ligation is first-line",
  "whyCorrect": "Endoscopic band ligation is the first-line intervention for acute esophageal variceal hemorrhage; it is superior to sclerotherapy (fewer perforation/stricture complications). Octreotide is an adjunct, not primary.",
  "trapType": "Treatment ladder — adjunct (octreotide) chosen over definitive (banding)",
  "futureAlert": "Cirrhosis + hematemesis = band first; octreotide/terlipressin is the bridge, not the answer",
  "reviseTopic": "Upper GI bleed: varices vs peptic ulcer management",
  "syllabusTopic": "Medicine — Cirrhosis and complications (portal hypertension)",
  "correctRule": "Acute variceal bleed → endoscopic band ligation first-line; TIPS is rescue",
  "distractors": {
    "B": "Sclerotherapy is second-line — banding has better outcomes and fewer complications.",
    "C": "Octreotide reduces portal pressure as a pharmacologic adjunct, not the primary intervention.",
    "D": "Balloon tamponade is a temporary bridge only, with high complication rate."
  },
  "distractorExplanations": [
    "A: Correct — band ligation is first-line.",
    "B: Wrong — sclerotherapy is second-line after banding.",
    "C: Wrong — octreotide is adjunctive pharmacotherapy.",
    "D: Wrong — tamponade is temporary rescue only."
  ],
  "errorLogRecord": {
    "date": "2026-08-19", "source": "MedCORE MCQ - Medicine-GI Emergencies",
    "questionId": "MED-GI-01", "subjectSystem": "Medicine - GI Emergencies",
    "syllabusTopic": "Medicine — Cirrhosis and complications",
    "myAnswer": "", "correctAnswer": "A",
    "missedDiscriminator": "Variceal bleed → band ligation first-line",
    "trapType": "Treatment ladder — adjunct over definitive",
    "correctRule": "Acute variceal bleed → banding first; octreotide is bridge",
    "futureAlert": "Cirrhosis + hematemesis = band first"
  }
}
{
  "id": "MED-GI-02",
  "stem": "A 48-year-old man presents with sudden severe epigastric pain of 6 hours duration. On examination, the abdomen is rigid and board-like. Erect chest X-ray shows free air under the diaphragm. What is the most appropriate next step?",
  "options": [
    "A) High-dose IV PPI and observation",
    "B) Urgent upper GI endoscopy",
    "C) Exploratory laparotomy with closure",
    "D) CT abdomen with contrast and outpatient follow-up"
  ],
  "correctAnswer": "C",
  "testedConcept": "Perforated peptic ulcer — definitive management",
  "cognitiveTask": "Analysis/synthesis (rigidity + free air → surgical emergency)",
  "discriminator": "Board-like rigidity + free air under diaphragm → perforation → surgery, not endoscopy or PPI",
  "whyCorrect": "Sudden severe pain with board-like rigidity and pneumoperitoneum is a perforated viscus; the definitive step is exploratory laparotomy with closure (or laparoscopy where available). PPI is adjunctive, endoscopy is contraindicated with perforation.",
  "trapType": "Partial Truth — PPI is correct for PUD but not definitive for perforation",
  "futureAlert": "Board-like rigidity + free air = laparotomy, never endoscopy",
  "reviseTopic": "Acute abdomen: perforated viscus vs other surgical emergencies",
  "syllabusTopic": "Surgery — Perforated peptic ulcer",
  "correctRule": "Pneumoperitoneum + rigidity → immediate surgical exploration",
  "distractors": {
    "A": "PPI treats the ulcer but not the perforation — definitive management is surgical.",
    "B": "Endoscopy is contraindicated when perforation is suspected (air insufflation worsens leak).",
    "D": "Imaging is already diagnostic; delaying surgery worsens peritonitis."
  },
  "distractorExplanations": [
    "A: Wrong — PPI is adjunctive, not definitive for perforation.",
    "B: Wrong — endoscopy contraindicated with suspected perforation.",
    "C: Correct — exploratory laparotomy is the definitive step.",
    "D: Wrong — delay to surgery worsens outcome."
  ],
  "errorLogRecord": {
    "date": "2026-08-19", "source": "MedCORE MCQ - Medicine-GI Emergencies",
    "questionId": "MED-GI-02", "subjectSystem": "Medicine - GI Emergencies",
    "syllabusTopic": "Surgery — Perforated peptic ulcer",
    "myAnswer": "", "correctAnswer": "C",
    "missedDiscriminator": "Free air + rigidity → surgery, not PPI/endoscopy",
    "trapType": "Partial Truth — PPI over surgical management",
    "correctRule": "Pneumoperitoneum + rigidity → immediate laparotomy",
    "futureAlert": "Board-like rigidity + free air = laparotomy"
  }
}

Note: both examples ship with discriminator + trap + alert, unique trap types, 4 options, single best answer, and pre-filled errorLogRecord — the full blueprint. The remaining 23 questions of Set #1 follow the same recipe across the GI-emergency topic range (UGIB, varices, perforation, obstruction, ischemic bowel, gallstone disease, pancreatitis).

8. Open questions (assumed, correct me if wrong)

  1. Fresh-set destination — assumed QBank/Mock Tests/ (like the ENT set) rather than MedCORE Production/07 MCQ Source JSON. Confirm if day-sets should also flow into MedCORE Production.
  2. NRE50 Set 2 — assumed built from the fresh pool per the Rotation Master Map. If you want it independent of the fresh sets, say so.
  3. Batch size — assumed 25 MCQs per set (matches every existing set). If you want 50-MCQ sets for the big subjects, adjust the backlog.
  4. Priority override — if any block is weaker than assumed, the backlog order can be re-sequenced; the small-subject surge (Block 5) is the highest-value gap to close.