ECG ST elevation myocardial infarction STEMI

ECG atrial fibrillation electrocardiogram

ECG ventricular tachycardia electrocardiogram

ECG complete heart block atrioventricular block

ECG Wolff-Parkinson-White delta wave pre-excitation

ECG left ventricular hypertrophy hypertrophic cardiomyopathy

ECG pericarditis saddle-shaped ST elevation

ECG pulmonary embolism S1Q3T3 right heart strain

ECG Brugada syndrome coved ST elevation V1 V2

ECG long QT syndrome torsades de pointes

ls -la /home/daytona/workspace/ 2>/dev/null || echo "workspace empty or missing"

mkdir -p /home/daytona/workspace/heart-ecg-pdf

~/heart-ecg-pdf/generate_pdf.py

import json
import subprocess
import base64
import os
from reportlab.lib.pagesizes import A4
from reportlab.lib import colors
from reportlab.lib.styles import getSampleStyleSheet, ParagraphStyle
from reportlab.lib.units import cm, mm
from reportlab.platypus import (
    SimpleDocTemplate, Paragraph, Spacer, Table, TableStyle,
    PageBreak, Image as RLImage, HRFlowable, KeepTogether
)
from reportlab.lib.enums import TA_CENTER, TA_LEFT, TA_JUSTIFY
from reportlab.platypus import BalancedColumns
from io import BytesIO

OUTPUT = "/home/daytona/workspace/heart-ecg-pdf/Heart_Diseases_ECG.pdf"
IMG_DIR = "/home/daytona/workspace/heart-ecg-pdf/imgs"
os.makedirs(IMG_DIR, exist_ok=True)

# ── Image URLs ──────────────────────────────────────────────────────────────
IMAGE_URLS = {
    "stemi":     "https://cdn.orris.care/cdss_images/pmc_clinical_VQA_16738d4b125195130b5e53a811645d4064182fc2bbd0de16b0e00a57af7ff6ef.jpg",
    "af":        "https://cdn.orris.care/cdss_images/pmc_clinical_VQA_55b3585ba004b0ca69f159c22eb5fbb4d8995fc007e77627c0d37922e4d88973.jpg",
    "chb":       "https://cdn.orris.care/cdss_images/pmc_clinical_VQA_755c449b940e537dbbe6dd7bf9c73561fc7bed3c059859efcb9b52c8878d99b4.jpg",
    "vt":        "https://cdn.orris.care/cdss_images/pmc_clinical_VQA_d0c8263390a11db77a98f37c73e35131257aee921845e98272875e8c3fe8caec.jpg",
    "wpw":       "https://cdn.orris.care/cdss_images/pmc_clinical_VQA_53c53d9ed7100de15d6d512e52e2a269252a79c85c12a60e54a4ce2fc4d574ec.jpg",
    "hcm":       "https://cdn.orris.care/cdss_images/pmc_clinical_VQA_2910b31808487d7cd1b8678e2a6942b2ca7518030d03f48787ac4b092bd7e92e.jpg",
    "pericarditis": "https://cdn.orris.care/cdss_images/pmc_clinical_VQA_cd4008531c1b1a2a99ab6d9b4282dcdcedfcf77c302b6ecbb7f7e01c5c985cf9.jpg",
    "pe":        "https://cdn.orris.care/cdss_images/pmc_clinical_VQA_5f391caa5a24ec425e9f48d4d9f73620a3b91f98df1f92cea400318772fda12e.jpg",
    "brugada":   "https://cdn.orris.care/cdss_images/pmc_clinical_VQA_6a00d07e5f8f18709c0b945773550567b91d5cce1d6e873c08d66dcc8caa5600.jpg",
    "lqts":      "https://cdn.orris.care/cdss_images/HARRISON_1763034143291_d1a2f9c1-7f66-4dec-a3f8-19b81f2ba927.png",
}

# ── Download images ──────────────────────────────────────────────────────────
print("Downloading images...")
urls = list(IMAGE_URLS.values())
result = json.loads(subprocess.check_output(
    ["node", "/home/daytona/skills/shared/scripts/fetch_images.js"] + urls
))

img_paths = {}
keys = list(IMAGE_URLS.keys())
for i, item in enumerate(result):
    key = keys[i]
    if item.get("base64"):
        raw = base64.b64decode(item["base64"].split(",")[1])
        ext = "png" if "png" in item["base64"] else "jpg"
        path = f"{IMG_DIR}/{key}.{ext}"
        with open(path, "wb") as f:
            f.write(raw)
        img_paths[key] = path
        print(f"  ✓ {key}")
    else:
        print(f"  ✗ {key}: {item.get('error')}")

# ── Styles ───────────────────────────────────────────────────────────────────
styles = getSampleStyleSheet()

DARK_RED   = colors.HexColor("#8B0000")
MED_RED    = colors.HexColor("#C0392B")
LIGHT_RED  = colors.HexColor("#FADBD8")
DARK_GRAY  = colors.HexColor("#2C3E50")
MID_GRAY   = colors.HexColor("#566573")
LIGHT_GRAY = colors.HexColor("#F2F3F4")
WHITE      = colors.white
GOLD       = colors.HexColor("#D4AC0D")

title_style = ParagraphStyle(
    "TitleStyle", parent=styles["Title"],
    fontSize=26, textColor=WHITE, alignment=TA_CENTER,
    spaceAfter=6, fontName="Helvetica-Bold", leading=32,
)
subtitle_style = ParagraphStyle(
    "SubtitleStyle", parent=styles["Normal"],
    fontSize=12, textColor=colors.HexColor("#F0B27A"),
    alignment=TA_CENTER, spaceAfter=4, fontName="Helvetica",
)
source_style = ParagraphStyle(
    "SourceStyle", parent=styles["Normal"],
    fontSize=8, textColor=colors.HexColor("#A9CCE3"),
    alignment=TA_CENTER, fontName="Helvetica-Oblique",
)
disease_num_style = ParagraphStyle(
    "DiseaseNum", parent=styles["Normal"],
    fontSize=10, textColor=MED_RED,
    fontName="Helvetica-Bold", spaceAfter=2,
)
disease_title_style = ParagraphStyle(
    "DiseaseTitle", parent=styles["Heading1"],
    fontSize=18, textColor=DARK_RED,
    fontName="Helvetica-Bold", spaceAfter=6, spaceBefore=4,
    borderPad=4,
)
section_head_style = ParagraphStyle(
    "SectionHead", parent=styles["Normal"],
    fontSize=11, textColor=WHITE, fontName="Helvetica-Bold",
    backColor=MED_RED, leftIndent=6, rightIndent=6,
    spaceAfter=4, spaceBefore=8, leading=16,
    borderPad=4,
)
body_style = ParagraphStyle(
    "Body", parent=styles["Normal"],
    fontSize=9.5, textColor=DARK_GRAY, fontName="Helvetica",
    spaceAfter=3, leading=14, alignment=TA_JUSTIFY,
)
bullet_style = ParagraphStyle(
    "Bullet", parent=styles["Normal"],
    fontSize=9.5, textColor=DARK_GRAY, fontName="Helvetica",
    leftIndent=14, bulletIndent=4, spaceAfter=2, leading=13,
    bulletFontName="Helvetica", bulletFontSize=10,
)
caption_style = ParagraphStyle(
    "Caption", parent=styles["Normal"],
    fontSize=8, textColor=MID_GRAY, fontName="Helvetica-Oblique",
    alignment=TA_CENTER, spaceAfter=6, spaceBefore=2, leading=11,
)
table_header_style = ParagraphStyle(
    "TableHeader", parent=styles["Normal"],
    fontSize=8.5, textColor=WHITE, fontName="Helvetica-Bold",
    alignment=TA_CENTER,
)
table_cell_style = ParagraphStyle(
    "TableCell", parent=styles["Normal"],
    fontSize=8.5, textColor=DARK_GRAY, fontName="Helvetica",
    alignment=TA_LEFT, leading=11,
)

# ── Helper: add image ────────────────────────────────────────────────────────
def ecg_image(key, caption_text, width=15*cm):
    items = []
    if key in img_paths:
        try:
            img = RLImage(img_paths[key], width=width, height=width*0.45)
            img.hAlign = "CENTER"
            items.append(img)
        except Exception as e:
            print(f"  Image error {key}: {e}")
    items.append(Paragraph(caption_text, caption_style))
    return items

def section_header(text):
    return Paragraph(f"  {text}", section_head_style)

def bullet(text):
    return Paragraph(f"• {text}", bullet_style)

def hr():
    return HRFlowable(width="100%", thickness=0.5, color=colors.HexColor("#D5D8DC"), spaceAfter=4, spaceBefore=4)

# ── Cover Page ───────────────────────────────────────────────────────────────
def cover_page(story):
    story.append(Spacer(1, 2.5*cm))

    # Title block with dark background table
    cover_data = [[
        Paragraph("10 Common Heart Diseases", title_style),
    ], [
        Paragraph("ECG Changes & Electrocardiographic Findings", subtitle_style),
    ], [
        Spacer(1, 0.3*cm),
    ], [
        Paragraph("A Comprehensive Clinical Reference", subtitle_style),
    ], [
        Spacer(1, 0.3*cm),
    ], [
        Paragraph("Sources: Harrison's Principles of Internal Medicine 22E · Braunwald's Heart Disease · Tintinalli's Emergency Medicine · Goldman-Cecil Medicine · Fuster and Hurst's The Heart 15E", source_style),
    ]]
    cover_table = Table(cover_data, colWidths=[17*cm])
    cover_table.setStyle(TableStyle([
        ("BACKGROUND",   (0,0), (-1,-1), DARK_GRAY),
        ("ROWPADDING",   (0,0), (-1,-1), 10),
        ("BOX",          (0,0), (-1,-1), 2, MED_RED),
        ("ROUNDEDCORNERS", [8]),
    ]))
    story.append(cover_table)
    story.append(Spacer(1, 1.2*cm))

    # Quick reference summary table on cover
    story.append(Paragraph("  Quick Reference Summary", ParagraphStyle(
        "QRTitle", parent=styles["Normal"], fontSize=13,
        textColor=DARK_RED, fontName="Helvetica-Bold", spaceAfter=6,
    )))

    qr_headers = ["#", "Disease", "Key ECG Finding", "Critical Clue"]
    qr_rows = [
        ["1", "STEMI", "Convex ST elevation + Q waves", "Territorial + reciprocal changes"],
        ["2", "Atrial Fibrillation", "No P waves, irregular RR", '"Irregularly irregular"'],
        ["3", "Complete Heart Block", "P/QRS dissociation, slow escape", "No fixed PR relationship"],
        ["4", "Ventricular Tachycardia", "Wide QRS tachycardia, AV dissociation", "Fusion/capture beats"],
        ["5", "WPW Syndrome", "Short PR + delta wave + wide QRS", "Slurred QRS upstroke"],
        ["6", "Hypertrophic Cardiomyopathy", "LVH voltage + giant T inversions", "Deep V2-V5 T inversions (apical)"],
        ["7", "Acute Pericarditis", "Diffuse saddle ST + PR depression", "All leads, not territorial"],
        ["8", "Pulmonary Embolism", "S1Q3T3 + sinus tachycardia + RBBB", "Right heart strain pattern"],
        ["9", "Brugada Syndrome", "Coved ST elevation V1-V2 + inv T", "Only Type 1 is diagnostic"],
        ["10", "Long QT Syndrome", "QTc >500 ms + T-wave changes", "Torsades de pointes risk"],
    ]

    table_data = [[Paragraph(h, table_header_style) for h in qr_headers]]
    for row in qr_rows:
        table_data.append([Paragraph(cell, table_cell_style) for cell in row])

    qr_table = Table(table_data, colWidths=[1*cm, 4.5*cm, 6.5*cm, 5*cm])
    qr_table.setStyle(TableStyle([
        ("BACKGROUND",   (0,0), (-1,0), DARK_RED),
        ("ROWBACKGROUNDS", (0,1), (-1,-1), [WHITE, LIGHT_GRAY]),
        ("TEXTCOLOR",    (0,0), (-1,0), WHITE),
        ("ALIGN",        (0,0), (0,-1), "CENTER"),
        ("FONTNAME",     (0,0), (-1,0), "Helvetica-Bold"),
        ("FONTSIZE",     (0,0), (-1,-1), 8.5),
        ("GRID",         (0,0), (-1,-1), 0.4, colors.HexColor("#D5D8DC")),
        ("ROWPADDING",   (0,0), (-1,-1), 5),
        ("VALIGN",       (0,0), (-1,-1), "MIDDLE"),
    ]))
    story.append(qr_table)
    story.append(PageBreak())

# ── Disease Entries ──────────────────────────────────────────────────────────
diseases = [
    {
        "num": "1",
        "title": "ST-Elevation Myocardial Infarction (STEMI)",
        "img_key": "stemi",
        "caption": "Anterior STEMI: Convex 'tombstone' ST-segment elevation across V1-V6, most pronounced in V2-V4, with reciprocal ST depression in inferior leads (II, III, aVF). Sinus rhythm with hyperacute T waves.",
        "overview": "STEMI results from complete occlusion of a coronary artery causing transmural myocardial injury. Immediate reperfusion (primary PCI within 90 min) is life-saving.",
        "ecg_points": [
            "Hyperacute T waves — earliest sign; tall, broad, symmetric peaked T waves",
            "ST-segment elevation ≥1 mm in ≥2 contiguous limb leads or ≥2 mm in ≥2 precordial leads (convex/tombstone morphology)",
            "Reciprocal ST depression in mirror-image leads (e.g., inferior STEMI → depression in aVL/I)",
            "Pathological Q waves (>40 ms wide, >1/4 R-wave height) — indicate transmural necrosis",
            "T-wave inversion follows as infarct evolves (hours to days)",
            "Localisation: Anterior (V1-V4 = LAD), Inferior (II,III,aVF = RCA), Lateral (I,aVL,V5-V6 = LCx)",
        ],
        "extra": None,
    },
    {
        "num": "2",
        "title": "Atrial Fibrillation (AF)",
        "img_key": "af",
        "caption": "Atrial fibrillation with rapid ventricular response: absent P waves replaced by fine fibrillatory baseline (best seen V1, lead II), and an irregularly irregular R-R interval pattern throughout all leads.",
        "overview": "AF is the most common sustained cardiac arrhythmia (~1-2% prevalence), caused by disorganised atrial electrical activity. It carries a 5x stroke risk, managed with rate/rhythm control and anticoagulation.",
        "ecg_points": [
            "Absent P waves — replaced by chaotic fibrillatory (f) waves at 350-600 impulses/min; best seen in V1 and II",
            "Irregularly irregular R-R intervals — the hallmark finding",
            "Narrow QRS complexes (unless aberrant conduction/bundle branch block co-exists)",
            "Ventricular rate: 100-180/min if uncontrolled; <100/min if rate-controlled",
            "Low-voltage QRS in limb leads may suggest infiltrative disease (e.g., amyloidosis)",
        ],
        "extra": None,
    },
    {
        "num": "3",
        "title": "Complete (Third-Degree) AV Block",
        "img_key": "chb",
        "caption": "Complete heart block: slow ventricular escape rhythm (~44 bpm) with wide QRS complexes, complete AV dissociation — P waves march at a faster independent rate through QRS complexes and T waves. Medical emergency requiring pacing.",
        "overview": "Complete AV block is a life-threatening arrhythmia where no atrial impulses conduct to the ventricles. Causes include inferior MI (RCA), Lyme disease, drugs (beta-blockers, digoxin), and idiopathic fibrosis (Lenegre disease).",
        "ecg_points": [
            "Complete AV dissociation — P waves and QRS complexes fire independently with no fixed PR interval",
            "Regular P-P intervals (atrial rate 60-100/min, faster than ventricles)",
            "Regular R-R intervals at a slower escape rate (junctional: 40-60/min, narrow QRS; ventricular: 20-40/min, wide QRS)",
            "Wide QRS if escape rhythm is infra-Hisian/ventricular (>120 ms)",
            "Secondary ST-T changes from abnormal ventricular activation sequence",
            "P waves may appear before, within, or after QRS complexes — none are conducted",
        ],
        "extra": None,
    },
    {
        "num": "4",
        "title": "Ventricular Tachycardia (VT)",
        "img_key": "vt",
        "caption": "Monomorphic ventricular tachycardia at ~150 bpm: broad regular QRS complexes (>120 ms), AV dissociation. Red circle highlights a capture beat — a narrow QRS indicating transient sinus capture of the ventricles, pathognomonic of VT.",
        "overview": "VT is a life-threatening arrhythmia originating below the bundle of His. Sustained VT (>30 sec) requires urgent cardioversion. Most common cause is structural heart disease (post-MI scar).",
        "ecg_points": [
            "Wide QRS complex tachycardia (QRS >120 ms) at rate 100-250/min",
            "AV dissociation — independent P waves seen in ~50% of cases",
            "Fusion beats — partial ventricular capture producing intermediate-morphology QRS",
            "Capture beats — brief normal narrow QRS (pathognomonic of VT, rare but diagnostic)",
            "Brugada criteria: concordance in precordials, RS interval >100 ms, no RS in any precordial lead → VT",
            "Ventricular fibrillation: chaotic disorganised baseline, no QRS — cardiac arrest",
        ],
        "extra": None,
    },
    {
        "num": "5",
        "title": "Wolff-Parkinson-White (WPW) Syndrome",
        "img_key": "wpw",
        "caption": "WPW syndrome: Shortened PR interval (<120 ms), delta waves (slurred initial QRS upstroke, red arrows in leads II-III), and widened QRS complexes throughout all leads. Negative delta waves in inferior leads suggest posteroseptal pathway.",
        "overview": "WPW is caused by a congenital accessory pathway (Bundle of Kent) bypassing the AV node. Risk of sudden death via antidromic AF with rapid ventricular rates. Curative catheter ablation is first-line for symptomatic patients.",
        "ecg_points": [
            "Short PR interval (<120 ms) — early ventricular activation via accessory pathway",
            "Delta wave — slurred upstroke at the start of QRS (ventricular pre-excitation)",
            "Wide QRS complex (>120 ms) — fusion of pre-excited and normal ventricular activation",
            "Secondary ST-T changes discordant to QRS vector",
            "Pathway localisation: negative deltas in II,III,aVF → posteroseptal; positive V1 → left free wall",
            "Risk: AF with rapid conduction → ventricular fibrillation (avoid AV nodal blockers in AF+WPW)",
        ],
        "extra": None,
    },
    {
        "num": "6",
        "title": "Hypertrophic Cardiomyopathy (HCM)",
        "img_key": "hcm",
        "caption": "HCM (septal type): High-amplitude R waves in V2-V4 (LVH voltage criteria), ST-segment depression, and deep symmetric T-wave inversions V2-V6 with strain pattern. Absent septal Q waves in I, aVL, V5-V6. Positive T wave in aVR (arrhythmic risk marker).",
        "overview": "HCM is the most common inherited cardiac condition (1:500), caused by sarcomere protein mutations. The leading cause of sudden cardiac death in young athletes. ECG is abnormal in >90% of patients.",
        "ecg_points": [
            "LVH voltage: Sokolow-Lyon (SV1 + RV5/V6 >35 mm); Cornell (RaVL + SV3 >28 mm males)",
            "ST-segment depression and T-wave inversion (strain pattern) in I, aVL, V4-V6",
            "Giant deep symmetric T-wave inversions in V2-V5 (apical HCM/Yamaguchi syndrome, >10 mm)",
            "Absence of septal Q waves in lateral leads (V5, V6, I, aVL)",
            "Left axis deviation; left atrial enlargement (P mitrale in lead II)",
            "Positive T wave in aVR — associated with increased arrhythmic risk",
        ],
        "extra": None,
    },
    {
        "num": "7",
        "title": "Acute Pericarditis",
        "img_key": "pericarditis",
        "caption": "Acute pericarditis: Diffuse concave ('saddle-shaped') ST elevation across I, II, III, aVF, V2-V6 with PR depression in lead II (most sensitive sign); reciprocal ST depression and PR elevation only in aVR. Sinus tachycardia reflects systemic inflammation.",
        "overview": "Pericarditis is inflammation of the pericardium, most commonly viral. ECG evolves through 4 stages over days to weeks. Key differentiation from STEMI: diffuse non-territorial changes, concave morphology, PR depression.",
        "ecg_points": [
            "Stage I (Days 1-2): Diffuse concave 'saddle-shaped' ST elevation in all leads except aVR and V1; PR depression (most sensitive)",
            "Stage II (Days 3-7): ST returns to baseline; T waves flatten",
            "Stage III (Weeks 1-3): T-wave inversion in previously elevated leads",
            "Stage IV (Weeks-months): ECG normalises",
            "Key differentiators from STEMI: diffuse (not territorial), concave not convex, PR depression present, no reciprocal depression except aVR, no Q waves",
            "Spodick's sign: downsloping TP segment (subtle but specific)",
        ],
        "extra": {
            "type": "stage_table",
        },
    },
    {
        "num": "8",
        "title": "Pulmonary Embolism (PE)",
        "img_key": "pe",
        "caption": "Acute PE (S1Q3T3 pattern): Prominent deep S wave in lead I (S1), pathological Q wave in lead III (Q3), inverted T wave in lead III (T3), all annotated with circles/arrows. Sinus tachycardia >100 bpm reflecting acute right heart strain.",
        "overview": "PE causes acute right ventricular pressure overload. The ECG has low sensitivity (~20-30%) but high specificity for severe PE when S1Q3T3 + RBBB are present. Sinus tachycardia is the most common finding.",
        "ecg_points": [
            "Sinus tachycardia — most common finding (>40% of cases)",
            "S1Q3T3 pattern (McGinn-White sign): prominent S in lead I, Q wave in III, T-wave inversion in III",
            "Right axis deviation and right bundle branch block (complete or incomplete, rSR' in V1)",
            "T-wave inversions V1-V4 (right ventricular strain)",
            "P pulmonale — peaked P waves >2.5 mm in lead II (right atrial enlargement)",
            "New atrial fibrillation or flutter may appear in massive PE",
        ],
        "extra": None,
    },
    {
        "num": "9",
        "title": "Brugada Syndrome",
        "img_key": "brugada",
        "caption": "Brugada syndrome Type 1 (only diagnostic pattern): Coved ST-segment elevation ≥2 mm in V1-V2 starting from an elevated J-point, descending with convex morphology into a symmetrically inverted T wave. RBBB-like rSR' morphology.",
        "overview": "Brugada syndrome is an inherited SCN5A sodium channel channelopathy causing polymorphic VT and sudden cardiac death, predominantly in young men. Type 1 pattern on ECG (spontaneous or drug-induced) is required for diagnosis.",
        "ecg_points": [
            "Type 1 (Coved — DIAGNOSTIC): J-point ≥2 mm, convex downsloping ST segment, negative (inverted) T wave in V1-V2",
            "Type 2 (Saddle-back): J-point ≥2 mm, saddle-back ST, positive T wave — requires pharmacological challenge",
            "Changes confined to V1-V2; sensitivity improves with electrodes in 2nd-3rd ICS",
            "RBBB-like rSR' pattern in V1; first-degree AV block; prolonged PR interval",
            "Arrhythmia: polymorphic VT, often nocturnal, triggered by fever or sodium channel blockers",
            "Fragmented QRS (multiple notches in QRS) indicates delayed heterogeneous depolarisation — high-risk marker",
        ],
        "extra": {
            "type": "brugada_table",
        },
    },
    {
        "num": "10",
        "title": "Long QT Syndrome (LQTS)",
        "img_key": "lqts",
        "caption": "Long QT syndrome: Panel A — baseline ECG showing markedly prolonged QTc (680 ms) with prominent U waves (arrows in V1). Panel B — short-long-short sequence triggering polymorphic VT (torsades de pointes): rapid QRS complexes twisting around the isoelectric line.",
        "overview": "LQTS is an inherited repolarisation disorder (KCNQ1, KCNH2, SCN5A mutations). Prolonged QTc delays repolarisation, creating risk of early afterdepolarisations and torsades de pointes. QTc >500 ms is high risk.",
        "ecg_points": [
            "Prolonged QTc: ≥450 ms (males) or ≥460 ms (females) borderline; ≥500 ms high-risk",
            "Bazett formula: QTc = QT / sqrt(RR) — correct for heart rate",
            "LQT1 (KCNQ1): broad-based smooth T wave",
            "LQT2 (KCNH2): bifid/notched T wave (most common subtype)",
            "LQT3 (SCN5A): late-onset peaked T wave with long isoelectric ST segment",
            "Prominent U waves in V2-V3 (may fuse with T wave, masking true QT prolongation)",
            "T-wave alternans: beat-to-beat variation in T-wave amplitude/morphology — electrical instability sign",
            "Torsades de Pointes: polymorphic VT, QRS complexes twist around isoelectric baseline",
        ],
        "extra": None,
    },
]

# ── Build PDF ─────────────────────────────────────────────────────────────────
doc = SimpleDocTemplate(
    OUTPUT,
    pagesize=A4,
    leftMargin=2*cm, rightMargin=2*cm,
    topMargin=2*cm, bottomMargin=2*cm,
    title="10 Common Heart Diseases — ECG Changes",
    author="Orris Medical AI",
    subject="Cardiology ECG Reference",
)

story = []
cover_page(story)

for disease in diseases:
    # Disease number + title
    story.append(Paragraph(f"DISEASE {disease['num']} OF 10", disease_num_style))
    story.append(Paragraph(disease["title"], disease_title_style))
    story.append(hr())

    # Overview
    story.append(section_header("Overview"))
    story.append(Paragraph(disease["overview"], body_style))
    story.append(Spacer(1, 6))

    # ECG Changes
    story.append(section_header("ECG Changes"))
    for point in disease["ecg_points"]:
        story.append(bullet(point))
    story.append(Spacer(1, 6))

    # Special tables
    if disease.get("extra"):
        if disease["extra"]["type"] == "stage_table":
            story.append(section_header("ECG Stage Evolution"))
            stage_data = [
                [Paragraph(h, table_header_style) for h in ["Stage", "Timing", "ECG Finding"]],
                [Paragraph("I", table_cell_style), Paragraph("Days 1-2", table_cell_style), Paragraph("Diffuse concave ST elevation; PR depression", table_cell_style)],
                [Paragraph("II", table_cell_style), Paragraph("Days 3-7", table_cell_style), Paragraph("ST returns to baseline; T waves flatten", table_cell_style)],
                [Paragraph("III", table_cell_style), Paragraph("Weeks 1-3", table_cell_style), Paragraph("T-wave inversion in previously elevated leads", table_cell_style)],
                [Paragraph("IV", table_cell_style), Paragraph("Weeks-months", table_cell_style), Paragraph("ECG normalises", table_cell_style)],
            ]
            stage_table = Table(stage_data, colWidths=[2*cm, 3.5*cm, 9.5*cm])
            stage_table.setStyle(TableStyle([
                ("BACKGROUND",   (0,0), (-1,0), DARK_RED),
                ("ROWBACKGROUNDS", (0,1), (-1,-1), [WHITE, LIGHT_GRAY]),
                ("GRID",         (0,0), (-1,-1), 0.4, colors.HexColor("#D5D8DC")),
                ("ROWPADDING",   (0,0), (-1,-1), 5),
                ("VALIGN",       (0,0), (-1,-1), "MIDDLE"),
            ]))
            story.append(stage_table)
            story.append(Spacer(1, 6))

        elif disease["extra"]["type"] == "brugada_table":
            story.append(section_header("Brugada ECG Pattern Classification"))
            brug_data = [
                [Paragraph(h, table_header_style) for h in ["Type", "ST Morphology", "T Wave", "Diagnostic?"]],
                [Paragraph("Type 1 (Coved)", table_cell_style), Paragraph("J-point ≥2 mm, convex downsloping", table_cell_style), Paragraph("Negative (inverted)", table_cell_style), Paragraph("YES", ParagraphStyle("Yes", parent=table_cell_style, textColor=colors.green, fontName="Helvetica-Bold"))],
                [Paragraph("Type 2 (Saddle-back)", table_cell_style), Paragraph("J-point ≥2 mm, saddle-back", table_cell_style), Paragraph("Positive", table_cell_style), Paragraph("NO (needs challenge)", ParagraphStyle("No", parent=table_cell_style, textColor=MED_RED))],
                [Paragraph("Type 3", table_cell_style), Paragraph("<2 mm ST elevation", table_cell_style), Paragraph("Positive", table_cell_style), Paragraph("NO", ParagraphStyle("No2", parent=table_cell_style, textColor=MED_RED))],
            ]
            brug_table = Table(brug_data, colWidths=[3.5*cm, 4.5*cm, 4*cm, 3*cm])
            brug_table.setStyle(TableStyle([
                ("BACKGROUND",   (0,0), (-1,0), DARK_RED),
                ("ROWBACKGROUNDS", (0,1), (-1,-1), [WHITE, LIGHT_GRAY]),
                ("GRID",         (0,0), (-1,-1), 0.4, colors.HexColor("#D5D8DC")),
                ("ROWPADDING",   (0,0), (-1,-1), 5),
                ("VALIGN",       (0,0), (-1,-1), "MIDDLE"),
            ]))
            story.append(brug_table)
            story.append(Spacer(1, 6))

    # ECG Image
    story.append(section_header("ECG Image"))
    story.append(Spacer(1, 4))
    for item in ecg_image(disease["img_key"], disease["caption"]):
        story.append(item)

    story.append(PageBreak())

# ── Build ─────────────────────────────────────────────────────────────────────
doc.build(story)
print(f"\nPDF saved to: {OUTPUT}")