File size: 62,391 Bytes
84602b6
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
#!/usr/bin/env python3
"""
Paper-Native Clinical PICO & Entity Parser for UnsolvedMedical v3.0.0.

Transforms raw peer-reviewed medical paper titles and metadata into:
  1. Complete, non-truncated, grammatically clean clinical topics (`clean_topic`)
  2. Paper-verified `condition_or_health_state` and `population` (eliminating cross-condition chimeras
     when a paper addresses a condition outside the 20 sub-domain defaults)
  3. Paper-specific `intervention_exposure_test_or_predictor` (`name`, `kind`, `details`)
  4. Paper-specific `comparator_or_reference`
  5. 100 natural clinical research question formulations (5 distinct structural variants x 20 question types)
     with distinct `statement`, concise `clean_statement`, and patient-accessible `plain_language_statement`.
"""

import html
import re

SUBTITLE_BOILERPLATE_RE = re.compile(
    r"\s*[:–—-]\s*(a |an |the )?(systematic review|meta-analysis|narrative review|scoping review|"
    r"literature review|umbrella review|comprehensive review|integrative review|minireview|mini-review|"
    r"review of|updated review|critical review|overview of|retrospective|prospective|multicenter|"
    r"multi-center|single-center|single center|randomized|randomised|phase [i1234v/]+|pilot study|"
    r"cohort study|cross-sectional|observational study|clinical trial|real-world|secondary analysis|"
    r"post-hoc|post hoc|sub-study|substudy|preliminary|comparative study|mendelian randomization|"
    r"network meta-analysis|pooled analysis|propensity score|outcomes from|results from|insights from|"
    r"findings from|lessons from|experience from|report from|evidence-based|consensus statement|"
    r"position paper|clinical practice guideline|expert consensus|multidisciplinary|"
    r"international delphi|delphi consensus|national survey|cross-sectional survey).*$",
    re.I,
)

PREFIX_BOILERPLATE_RE = re.compile(
    r"^(a |an |the )?(systematic review and meta-analysis of|systematic review of|meta-analysis of|"
    r"narrative review of|scoping review of|comprehensive review of|clinical overview of|overview of|"
    r"update on|updates in|current perspectives on|emerging role of|advances in|new insights into|"
    r"recent advances in|state of the art in|clinical management of|management of|current state of|"
    r"exploring the|understanding the|investigating the|evaluating the|assessing the|"
    r"reassessing the role of|reassessing|role of|the role of|development of a|development of|"
    r"predictive value of|prognostic value of|diagnostic value of|clinical utility of|"
    r"clinical significance of|significance of|expert consensus on the|expert consensus on|"
    r"consensus statement on|international consensus on|diagnosis and monitoring of|diagnosis and management of|"
    r"diagnosis and treatment of|prevention and treatment of|efficacy and safety of|effectiveness and safety of|"
    r"efficacy of|effectiveness of|impact of the|impact of|effect of|effects of|"
    r"influence of|the influence of|association between|association of|relationship between)\s+",
    re.I,
)

DANGLING_TAIL_RE = re.compile(
    r"\s+\b(a|an|the|in|of|for|with|without|and|or|by|to|from|on|at|among|between|after|before|"
    r"during|under|into|via|versus|vs|through|across|within|towards|toward|against|upon|using|"
    r"including|following|regarding|concerning|based|associated|related|compared|comparing|"
    r"as|such|whether|while|when|where|who|which|that|whose|undergoing|receiving|treated|"
    r"diagnosed|presenting|living|incorporating|combining|targeting|evaluating|assessing|"
    r"predicting|preventing|managing|early|late|acute|chronic|severe|moderate|mild|advanced|"
    r"metastatic|primary|secondary|novel|new|current|emerging|clinical|human|pediatric|adult|"
    r"older|elderly|high-risk|low-risk|refractory|relapsed|recurrent)\b[:;,\-–—.]*$",
    re.I,
)

# Non-distinctive modifier words that must NOT trigger a specific sub-domain disease match on their own
GENERIC_CONDITION_MODIFIERS = {
    "with", "without", "from", "after", "before", "during", "early", "stage", "late", "acute",
    "chronic", "severe", "moderate", "mild", "advanced", "metastatic", "unresectable", "resectable",
    "borderline", "locally", "localized", "high", "risk", "low", "relapsed", "relapsing", "refractory",
    "recurrent", "first", "second", "line", "complete", "remission", "positive", "negative", "wildtype",
    "proficient", "deficient", "dependent", "resistant", "progressing", "associated", "related",
    "induced", "mediated", "primary", "secondary", "adult", "adults", "children", "pediatric",
    "neonatal", "adolescent", "older", "elderly", "maternal", "pregnant", "postpartum", "patients",
    "patient", "care", "health", "management", "treatment", "therapy", "disease", "diseases",
    "syndrome", "syndromes", "disorder", "disorders", "condition", "conditions", "infection",
    "infections", "cancer", "carcinoma", "adenocarcinoma", "malignancy", "tumor", "tumors",
    "cell", "small", "large", "diffuse", "factor", "inhibitors", "inhibitor", "mutations", "driver",
    "pathway", "inhibition", "targetable", "neutralizing", "viii", "active", "grade", "serous",
    "chemoimmunotherapy", "blockade", "allogeneic", "hsct", "transplant", "transfusion",
    "long", "term", "short", "type", "types", "form", "forms", "rate", "rates", "systemic",
    "idiopathic", "hereditary", "familial", "congenital", "acquired", "uncomplicated", "complicated",
    "stanford", "persistent", "paroxysmal", "sustained", "protracted", "maintenance", "discontinuation",
    "withdrawal", "postoperative", "perioperative", "preoperative", "intensive", "critical",
    "critically", "emergency", "outpatient", "inpatient", "community", "hospital", "surgical",
    "medical", "clinical", "human", "cohort", "onset", "status", "loss", "pain", "failure", "injury",
}

# Broad umbrella disease nouns that require at least one accompanying qualifier token when present in base_cond_name
UMBRELLA_DISEASE_NOUNS = {
    "amyloidosis", "dissection", "aneurysm", "stenosis", "regurgitation", "insufficiency",
    "cardiomyopathy", "neuropathy", "nephropathy", "retinopathy", "hepatitis", "cirrhosis",
    "fibrosis", "arthritis", "vasculitis", "colitis", "epilepsy", "dementia", "depression",
    "anemia", "thrombocytopenia", "hypertension", "diabetes", "asthma", "stroke", "hemorrhage",
    "infarction", "thrombosis", "embolism", "leukemia", "lymphoma", "myeloma", "sarcoma",
    "melanoma", "glioma", "pneumonia", "meningitis", "encephalitis", "endocarditis", "myocarditis",
    "pancreatitis", "sepsis", "glaucoma", "uveitis", "dermatitis", "psoriasis", "alopecia",
    "lupus", "myositis", "osteoporosis", "osteoarthritis", "fracture", "fractures", "infertility",
    "dystrophy", "hemophilia", "thalassemia",
}

# Explicit comparison pattern: "<Arm A> versus/vs./compared with <Arm B>"
EXPLICIT_COMP_RE = re.compile(
    r"(?P<arm_a>[A-Za-z0-9][A-Za-z0-9\s,\-/+()]{3,75}?)\s+"
    r"(?:versus|vs\.?|compared with|compared to)\s+"
    r"(?P<arm_b>[A-Za-z0-9][A-Za-z0-9\s,\-/+()]{4,75}?)"
    r"(?:\s+(?:in|for|among|after|during|with|on|to)\b|$)",
    re.I,
)

# "Role/Efficacy/Safety/Value/Impact of <Intervention/Factor> in/for/on <Condition/Outcome>"
OF_PATTERN_RE = re.compile(
    r"^(?:the\s+)?(?:role|efficacy|effectiveness|safety|impact|effects?|influence|value|utility|feasibility|"
    r"performance|accuracy|use|application|addition|combination|integration|implementation|"
    r"optimization|evaluation|assessment|association|relationship|correlation|"
    r"predictive value|prognostic value|diagnostic value|diagnostic accuracy|clinical utility|"
    r"clinical significance|significance|therapeutic potential|mechanisms?|pathogenesis|determinants|predictors|biomarkers|"
    r"reassessing the role|development|genetics)\s+of\s+(?:a\s+|an\s+|the\s+)?"
    r"(?P<focus>[A-Za-z0-9][A-Za-z0-9\s,\-/+()]{5,90}?)"
    r"(?:\s+(?:in|for|among|on|to guide|after|during|with)\s+(?P<context>[A-Za-z0-9].+))?$",
    re.I,
)

# "<Intervention/Biomarker> for/in the treatment/management/diagnosis/prediction/prevention of <Condition>"
FOR_ACTION_RE = re.compile(
    r"^(?P<focus>[A-Za-z0-9][A-Za-z0-9\s,\-/+()]{5,85}?)\s+"
    r"(?:for|in|to guide)\s+(?:the\s+)?(?:treatment|management|prevention|diagnosis|early detection|"
    r"prediction|prognosis|risk stratification|monitoring|rehabilitation|mitigation|control|adjuvant|neoadjuvant)\s*(?:of\s+)?"
    r"(?P<context>[A-Za-z0-9].+)$",
    re.I,
)

# "<Intervention/Exposure> and <Outcome/Condition> in <Population>"
AND_IN_RE = re.compile(
    r"^(?P<focus>[A-Za-z0-9][A-Za-z0-9\s,\-/+()]{5,75}?)\s+and\s+"
    r"(?P<outcome>[A-Za-z0-9][A-Za-z0-9\s,\-/+()]{4,70}?)\s+"
    r"(?:in|among|after|during)\s+(?P<context>[A-Za-z0-9].+)$",
    re.I,
)

# Participial/Gerund lead: "Targeting/Predicting/De-escalating <X> in <Y>"
GERUND_RE = re.compile(
    r"^(?:targeting|inhibiting|modulating|predicting|stratifying|monitoring|optimizing|"
    r"de-escalating|reassessing|mapping|decoding|unraveling|evaluating|comparing|enhancing|"
    r"preventing|overcoming|identifying|quantifying|characterizing)\s+(?:the\s+|a\s+|an\s+)?"
    r"(?P<focus>[A-Za-z0-9][A-Za-z0-9\s,\-/+()]{5,85}?)"
    r"(?:\s+(?:in|for|among|after|during|with|to)\s+(?P<context>[A-Za-z0-9].+))?$",
    re.I,
)

# Extract condition phrase after "in patients with <X>", "in <X>", "of <X>", "for <X>"
COND_PREP_RE = re.compile(
    r"\b(?:in patients with|among patients with|in adults with|in children with|in women with|in men with|in individuals with|"
    r"in the management of|in the treatment of|for the treatment of|in|among)\s+"
    r"(?P<cond>[A-Za-z0-9][A-Za-z0-9\s,\-/+()]{6,95})$",
    re.I,
)


def _balance_parens(text: str) -> str:
    t = text.strip()
    while t.startswith("(") and t.endswith(")") and t[1:-1].count("(") == t[1:-1].count(")"):
        t = t[1:-1].strip()
    if t.count("(") > t.count(")"):
        last_open = t.rfind("(")
        if len(t) - last_open <= 15:
            t = t + ")"
        else:
            t = t[:last_open].strip()
    elif t.count(")") > t.count("("):
        t = t.replace(")", "").strip()
    return t


def clean_complete_topic(raw_title: str) -> str:
    """
    Extract a complete, self-contained clinical topic phrase from a peer-reviewed paper title
    without mid-phrase truncation or dangling prepositions/adjectives.
    """
    t = html.unescape(raw_title or "")
    t = re.sub(r"<[^>]+>", "", t).strip()
    t = re.sub(r"\s+", " ", t).rstrip(".:;?-–— ")
    t = SUBTITLE_BOILERPLATE_RE.sub("", t).strip()

    if ":" in t:
        p0, p1 = [p.strip() for p in t.split(":", 1)]
        if p0.lower() in {"research review", "clinical review", "special report", "consensus statement", "clinical practice guideline", "original article", "review article"} and len(p1) >= 24:
            t = p1
        elif len(p0) >= 28:
            t = p0
        elif len(p1) >= 28 and not SUBTITLE_BOILERPLATE_RE.search(": " + p1):
            t = f"{p0} ({p1})" if len(p0) + len(p1) < 120 else p1

    t = re.sub(
        r"^(a |an |the )?(systematic review and meta-analysis of|systematic review of|meta-analysis of|"
        r"narrative review of|scoping review of|comprehensive review of|clinical overview of|overview of|"
        r"update on|updates in|current perspectives on)\s+",
        "",
        t,
        flags=re.I,
    ).strip()

    if len(t) > 215:
        clause_break = re.search(r"[,;–—]\s+", t[110:])
        if clause_break:
            t = t[: 110 + clause_break.start()].strip()
        else:
            t = t[:210].rsplit(" ", 1)[0].strip()

    t = _balance_parens(t)
    t = re.sub(r"[:;,\-–—/]+\s*$", "", t).strip()
    while True:
        new_t = DANGLING_TAIL_RE.sub("", t).strip()
        new_t = re.sub(r"[:;,\-–—/]+\s*$", "", new_t).strip()
        if new_t == t or len(new_t) < 22:
            break
        t = new_t

    t = _balance_parens(t)
    if t and t[0].islower():
        t = t[0].upper() + t[1:]
    return t


def _clean_extracted_entity(phrase: str) -> str:
    p = phrase.strip(" ,:;.-–—")
    p = PREFIX_BOILERPLATE_RE.sub("", p).strip(" ,:;.-–—")
    p = re.sub(r"\s+\b(patients|individuals|adults|children|subjects|cases)\s*$", "", p, flags=re.I).strip()
    while True:
        np = DANGLING_TAIL_RE.sub("", p).strip(" ,:;.-–—")
        if np == p or len(np) < 8:
            break
        p = np
    p = _balance_parens(p)
    if p and p[0].islower():
        p = p[0].upper() + p[1:]
    return p


def _distinctive_condition_words(cond_name: str) -> list[str]:
    return [
        w.lower()
        for w in re.findall(r"[A-Za-z0-9]{3,}", cond_name)
        if len(w) >= 4 and w.lower() not in GENERIC_CONDITION_MODIFIERS
    ]


def _extract_condition_from_paper(clean_topic: str, cat_display_name: str) -> str | None:
    """Extract a paper-native disease/condition phrase from clean_topic when it differs from the sub-domain default."""
    trimmed_topic = re.sub(r"\s+in the (?:era|context|setting|wake|presence) of\s+[A-Za-z0-9\s,\-/+()]+$", "", clean_topic, flags=re.I).strip()
    m_with = re.search(
        r"\b(?:children and young people|children and adolescents|adolescents and adults|older adults|pregnant women|postpartum women|children|adolescents|adults|women|men|patients|individuals|infants|neonates)\s+with\s+"
        r"(?P<cond>[A-Za-z0-9][A-Za-z0-9\s,\-/+()]{6,90})$",
        trimmed_topic,
        flags=re.I,
    )
    if m_with:
        c = _clean_extracted_entity(m_with.group("cond"))
        if 8 <= len(c) <= 90:
            return c
    m = COND_PREP_RE.search(trimmed_topic)
    if m:
        c = _clean_extracted_entity(m.group("cond"))
        if 8 <= len(c) <= 90 and c.lower() not in {"clinical practice", "routine care", "primary care", "the united states", "europe", "china", "japan", "turkey"}:
            return c
    m_of = re.search(
        r"\b(?:genetics|pathogenesis|diagnosis|management|treatment|epidemiology|prognosis|classification|biomarkers|monitoring)\s+of\s+"
        r"([A-Za-z0-9][A-Za-z0-9\s,\-/+()]{6,85})$",
        trimmed_topic,
        flags=re.I,
    )
    if m_of:
        c = _clean_extracted_entity(m_of.group(1))
        if 8 <= len(c) <= 90:
            return c
    return None


def _strip_redundant_population_tail(topic_text: str, cond_name: str) -> str:
    """
    Strip trailing ' in <disease/patients>' from a paper topic when it is embedded inside a
    sentence that already explicitly states 'In <population> with <condition>'.
    """
    trimmed_topic = re.sub(r"\s+in the (?:era|context|setting) of\s+[A-Za-z0-9\s,\-/+()]+$", "", topic_text, flags=re.I).strip()
    m_on_of = re.search(
        r"\s+(?:on|in|among|for)\s+(?:the\s+[a-z\s]{3,25}\s+of\s+)?(?:children|adolescents|adults|women|men|patients|individuals|infants|neonates)[A-Za-z0-9,\-\s/+()]{4,95}$",
        trimmed_topic,
        flags=re.I,
    )
    if m_on_of:
        trimmed = trimmed_topic[: m_on_of.start()].strip()
        if len(trimmed) >= 14:
            return _clean_extracted_entity(trimmed)
    cond_words = set(_distinctive_condition_words(cond_name))
    m = re.search(r"\s+(?:in|among)\s+([A-Za-z0-9,\-\s/+()]{8,85})$", trimmed_topic, flags=re.I)
    if m:
        tail = m.group(1).lower()
        if any(cw in tail for cw in cond_words) or any(k in tail for k in ("patients", "cancer", "disease", "carcinoma", "lymphoma", "leukemia", "amyloidosis", "syndrome", "dissection", "children", "women", "adults")):
            trimmed = trimmed_topic[: m.start()].strip()
            if len(trimmed) >= 14:
                return _clean_extracted_entity(trimmed)
    return _clean_extracted_entity(trimmed_topic)


def select_coherent_outcomes(cat_id: int, raw_title: str, cond_name: str) -> tuple[str, str, str]:
    """
    Return domain-coherent (prim_out, safe_out, pro_out) phrases matched to the category and specific
    disease/sub-specialty so no unrelated sub-domain scales (e.g. endometriosis scales on PCOS, or dental
    scales on stroke rehabilitation) ever appear in a question statement.
    """
    tl = f"{raw_title} {cond_name}".lower()
    if cat_id == 1:
        if any(k in tl for k in ("thromboembol", "vte", "anticoagul", "thrombosis")):
            return (
                "recurrent venous thromboembolism and clinically relevant bleeding",
                "ISTH major bleeding and clinically relevant non-major bleeding",
                "anticoagulation treatment burden and health-related quality of life",
            )
        if any(k in tl for k in ("hemophilia", "haemophilia", "sickle cell", "thalassemia", "aplastic anemia", "hemoglobinuria", "pnh", "thrombocytopenic", "ttp", "itp", "von willebrand")):
            return (
                "annualized bleeding or crisis rate, hemoglobin stabilization, and event-free survival",
                "thromboembolic events, hemolysis or bleeding breakthroughs, and immunogenicity",
                "FACIT-Fatigue score, physical function, and transfusion or infusion burden",
            )
        return (
            "overall survival and progression-free or disease-free survival",
            "Grade >=3 treatment-related and immune-mediated adverse events",
            "EORTC QLQ-C30 health-related quality of life and symptomatic toxicity burden",
        )
    if cat_id == 2:
        if "aortic" in tl or "dissection" in tl or "aneurysm" in tl:
            return (
                "aortic-related mortality, stroke, and reintervention-free survival",
                "spinal cord ischemia, stroke, and major perioperative bleeding",
                "functional independence and health-related quality of life",
            )
        return (
            "cardiovascular death, myocardial infarction, and ischemic stroke",
            "ISTH major bleeding, intracranial hemorrhage, and arrhythmia or conduction disturbances",
            "Kansas City Cardiomyopathy Questionnaire (KCCQ) score and functional independence",
        )
    if cat_id == 3:
        if "epilep" in tl or "seizure" in tl:
            return (
                "sustained seizure freedom and >=50% seizure frequency reduction",
                "neuropsychiatric and cognitive adverse events and treatment discontinuation",
                "Quality of Life in Epilepsy (QOLIE-31) score and cognitive function",
            )
        return (
            "confirmed disability progression and longitudinal neurological function",
            "treatment-related neurotoxicity, serious infection, and infusion or procedural complications",
            "cognitive function, fatigue, and patient-reported neurological quality of life (Neuro-QoL)",
        )
    if cat_id == 4:
        return (
            "sustained symptomatic remission and relapse prevention",
            "suicidal ideation or behavior, treatment-emergent adverse events, and withdrawal symptoms",
            "psychosocial functioning, social participation, and patient-reported mental health quality of life",
        )
    if cat_id == 5:
        return (
            "28-day and 90-day all-cause mortality and microbiological or clinical cure",
            "organ toxicity and treatment-emergent antimicrobial resistance selection",
            "time to symptom resolution, hospital-free days, and treatment completion",
        )
    if cat_id == 6:
        if "sleep" in tl or "apnea" in tl:
            return (
                "apnea-hypopnea index reduction, nocturnal oxygenation, and cardiovascular event rate",
                "treatment intolerance, mask or device complications, and daytime residual sedation",
                "Epworth Sleepiness Scale score, sleep quality, and daytime functional alertness",
            )
        return (
            "annualized severe exacerbation rate and lung function trajectory (FEV1 or FVC)",
            "pneumonia, airway infection, and treatment-related systemic adverse events",
            "St George's Respiratory Questionnaire (SGRQ) score and dyspnea symptom burden",
        )
    if cat_id == 7:
        return (
            "glycemic and cardiometabolic control, major adverse cardiovascular events, and organ preservation",
            "severe hypoglycemia, lean body mass loss, and gastrointestinal or hepatobiliary adverse events",
            "metabolic health-related quality of life, physical function, and treatment satisfaction",
        )
    if cat_id == 8:
        if any(k in tl for k in ("liver", "hepat", "cirrhosis", "masld", "nafld", "nash", "biliary", "cholang")):
            return (
                "liver-related event-free survival, histologic fibrosis regression, and decompensation rate",
                "hepatic decompensation, variceal bleeding, and serious infection",
                "Chronic Liver Disease Questionnaire (CLDQ) score and hospital-free days",
            )
        return (
            "steroid-free clinical and endoscopic or transmural remission",
            "serious opportunistic infection, gastrointestinal bleeding, and bowel perforation",
            "digestive symptom burden, IBD-Disk disability score, and quality of life",
        )
    if cat_id == 9:
        return (
            "sustained eGFR preservation, progression to kidney failure, and renal or patient survival",
            "hyperkalemia, acute kidney injury, volume overload, and serious infection",
            "Kidney Disease Quality of Life (KDQOL-36) score and symptom burden",
        )
    if cat_id == 10:
        return (
            "sustained steroid-free clinical remission and prevention of organ damage accrual",
            "serious opportunistic infection, thromboembolic events, and treatment-related toxicity",
            "HAQ-DI physical function, FACIT-Fatigue score, and cumulative glucocorticoid toxicity",
        )
    if cat_id == 11:
        if any(k in tl for k in ("psoriasis", "dermatitis", "eczema", "hidradenitis", "alopecia", "vitiligo", "skin", "cutaneous", "pemphig", "urticaria", "acne", "melasma", "wound", "ulcer")):
            return (
                "validated skin clearance or lesion resolution and durable disease control",
                "mucocutaneous or systemic infection and treatment-related adverse events",
                "Dermatology Life Quality Index (DLQI), pruritus or pain intensity, and patient satisfaction",
            )
        return (
            "pain intensity reduction, functional joint survival, and revision-free durability",
            "periprosthetic infection or fracture, thromboembolism, and procedural complications",
            "PROMIS pain interference, physical function, and return to independent mobility",
        )
    if cat_id == 12:
        if any(k in tl for k in ("polycystic", "pcos", "ovary", "ovarian", "infertility", "fertility", "ivf", "menopaus", "endometriosis", "fibroid", "adenomyosis", "contracept", "menstrual", "gynecol", "pelvic")):
            return (
                "reproductive, endocrine, and cardiometabolic health outcomes",
                "treatment-related thromboembolic, metabolic, and gynecologic adverse events",
                "validated condition-specific quality of life, symptom severity, and reproductive well-being",
            )
        return (
            "severe maternal morbidity and perinatal survival without severe morbidity",
            "preeclampsia, fetal growth restriction, neonatal complications, and maternal hemorrhage",
            "maternal functional recovery, postpartum depression (EPDS), and patient-reported birth experience",
        )
    if cat_id == 13:
        if any(k in tl for k in ("neonat", "preterm", "infant", "necrotizing", "bronchopulmonary", "hypoxic-ischemic")):
            return (
                "survival without neurodevelopmental impairment and major neonatal morbidity",
                "sepsis, necrotizing enterocolitis, and growth suppression",
                "neurodevelopmental trajectory and family caregiver burden",
            )
        return (
            "pediatric disease control, age-appropriate growth, and long-term organ function",
            "treatment-related toxicity, infection, and developmental or endocrine adverse effects",
            "Pediatric Quality of Life Inventory (PedsQL) score and family caregiver burden",
        )
    if cat_id == 14:
        if any(k in tl for k in ("alzheimer", "dementia", "amyloid", "cognitive", "lewy", "frontotemporal")):
            return (
                "CDR-SB cognitive and functional trajectory and days alive and independent at home",
                "amyloid-related imaging abnormalities (ARIA), intracerebral hemorrhage, and falls",
                "neuropsychiatric symptom burden, Zarit caregiver burden score, and quality of life",
            )
        return (
            "days alive and independent at home, functional preservation, and survival",
            "falls, delirium, polypharmacy-related adverse drug events, and emergency hospitalization",
            "symptom burden, Zarit caregiver burden score, and goal-concordant care",
        )
    if cat_id == 15:
        return (
            "30-day and 90-day all-cause mortality and organ support-free days",
            "major perioperative or critical care complications, organ ischemia, and reintervention",
            "EQ-5D-5L functional recovery, 6-minute walk distance, and return to independent living",
        )
    if cat_id == 16:
        return (
            "diagnostic accuracy, time to definitive diagnosis, and downstream clinical outcome impact",
            "false-positive invasive biopsy cascades, overdiagnosis, and diagnostic delay",
            "diagnostic yield, clinician decision confidence, and patient decisional regret",
        )
    if cat_id == 17:
        return (
            "all-cause and cause-specific mortality and incident disease rate",
            "overdiagnosis, false-positive interventions, and preventive regimen adverse events",
            "health equity gap reduction across socioeconomic strata and quality-adjusted life years",
        )
    if cat_id == 18:
        return (
            "30-day mortality, preventable adverse events, and guideline-concordant care delivery",
            "unintended care delays, missed diagnoses, and clinician workload burden",
            "patient-reported care coordination, equity disparity reduction, and implementation fidelity",
        )
    if cat_id == 19:
        if any(k in tl for k in ("retinal", "macular", "glaucoma", "myopia", "visual", "ophthalm", "ocular", "cornea", "uveitis", "cataract", "optic")):
            return (
                "best-corrected visual acuity (BCVA) and structural retinal or optic nerve preservation",
                "intraocular inflammation, endophthalmitis, and ocular hypertension",
                "National Eye Institute Visual Function Questionnaire (NEI VFQ-25) score and visual independence",
            )
        if any(k in tl for k in ("dental", "periodont", "caries", "implant", "oral", "orthodont", "endodont", "maxillofacial", "tooth", "teeth")):
            return (
                "tooth or implant survival, clinical attachment gain, and oral disease resolution",
                "peri-implantitis, alveolar bone loss, and postoperative infection",
                "Oral Health Impact Profile (OHIP-14) score, masticatory function, and patient comfort",
            )
        if any(k in tl for k in ("hearing", "cochlear", "tinnitus", "vestibular", "speech", "aphasia", "dysphagia", "auditory", "deaf")):
            return (
                "speech-in-noise recognition, communication recovery, and swallowing safety",
                "device failure, aspiration pneumonia, and vestibular dysfunction",
                "Tinnitus Functional Index, communication participation, and WHODAS 2.0 score",
            )
        return (
            "validated motor and functional recovery (Fugl-Meyer, ARAT, or FIM)",
            "musculoskeletal injury, falls, and autonomic or cardiovascular instability",
            "WHODAS 2.0 community participation, independent mobility, and quality of life",
        )
    return (
        "event-free survival, organ-specific disease stabilization, and validated functional response",
        "treatment-related immunogenicity, hepatotoxicity, and severe multisystem adverse events",
        "time to definitive diagnosis, functional independence, and patient or caregiver quality of life",
    )


def _check_subdomain_disease_match(raw_title: str, base_cond_name: str) -> bool:
    """
    Verify whether `raw_title` genuinely addresses `base_cond_name` using whole-word token matching,
    subtype conflict checks (`type a` vs `type b`, `type 1` vs `type 2`), and qualifier checks for umbrella nouns.
    """
    tl = raw_title.lower()
    bcl = base_cond_name.lower()
    if ("type a" in bcl and "type b" in tl) or ("type b" in bcl and "type a" in tl):
        return False
    if ("type 1" in bcl and "type 2" in tl) or ("type 2" in bcl and "type 1" in tl):
        return False

    title_tokens = set(re.findall(r"[a-z0-9]+", tl))
    dist_words = _distinctive_condition_words(base_cond_name)
    if not dist_words:
        return True

    qualifiers = [w for w in dist_words if w not in UMBRELLA_DISEASE_NOUNS]
    umbrellas = [w for w in dist_words if w in UMBRELLA_DISEASE_NOUNS]

    if qualifiers and umbrellas:
        return any(q in title_tokens for q in qualifiers) and any(u in title_tokens for u in umbrellas)
    return any(dw in title_tokens for dw in dist_words)


def extract_paper_pico(
    raw_title: str,
    clean_topic: str,
    qtype: str,
    cond_tuple: tuple,
    int_tuple: tuple,
    outcomes_list: list[dict],
    lower_fn,
    cat_display_name: str = "Clinical medicine",
    cat_id: int = 1,
) -> dict:
    """
    Extract paper-specific PICO elements (`cond_name`, `int_name`, `int_details`, `comp_name`, `pop_desc`,
    `stage_sev`, `prio_sub`, `focus_clean`, `prim_out`, `safe_out`, `pro_out`) from the paper title and
    harmonize with the sub-domain clinical baseline.
    Guarantees zero cross-disease chimeras and zero mismatched outcome scales.
    """
    base_cond_name, onto_ids, base_stage, phenotype, base_pop, age_stages, inc_feat, base_prio_sub = cond_tuple
    _, base_int_name, base_int_details, base_comp_name, horizon, settings, _ = int_tuple

    tl = raw_title.lower()
    matches_subdomain_disease = _check_subdomain_disease_match(raw_title, base_cond_name)

    if matches_subdomain_disease:
        cond_name = base_cond_name
        pop_desc = base_pop
        stage_sev = base_stage
        prio_sub = base_prio_sub
    else:
        # Paper addresses a specific condition in this medical category that differs from `base_cond_name`
        paper_cond = _extract_condition_from_paper(clean_topic, cat_display_name)
        if paper_cond:
            cond_name = paper_cond
            if any(p in paper_cond.lower() for p in ("children", "adolescents", "young people", "women", "adults", "patients", "infants", "neonates", "older adults")):
                pop_desc = paper_cond[0].upper() + paper_cond[1:]
            else:
                pop_desc = f"Patients with {lower_fn(paper_cond)}"
        else:
            cond_name = f"{clean_topic[:80].rsplit(' ', 1)[0]}" if len(clean_topic) > 80 else clean_topic
            pop_desc = f"Patients evaluated for {lower_fn(cond_name)} in {cat_display_name.lower()}"
        stage_sev = "Clinically characterized human cohort"
        prio_sub = ["high-risk clinical subgroups", "underrepresented age and comorbidity strata"]

    extracted_int = None
    extracted_comp = None

    # 1. Check explicit comparison in title ("X versus Y", "X compared with Y")
    m_comp = EXPLICIT_COMP_RE.search(clean_topic)
    if m_comp:
        a = _clean_extracted_entity(m_comp.group("arm_a"))
        b = _clean_extracted_entity(m_comp.group("arm_b"))
        if " of " in a.lower():
            a = _clean_extracted_entity(a.split(" of ")[-1])
        # Reconstruct shared head noun when `a` is a single modifier word (e.g., "hybrid versus total arch replacement")
        a_words = a.split()
        b_words = b.split()
        if len(a_words) == 1 and len(b_words) >= 2:
            a = f"{a} {' '.join(b_words[1:])}"
        if len(a) >= 6 and len(b) >= 6:
            extracted_int = a
            extracted_comp = b

    # 2. Check "Role/Value/Efficacy/Safety/Development of <Focus> in/for <Context>"
    if not extracted_int:
        m_of = OF_PATTERN_RE.match(clean_topic)
        if m_of:
            focus = _clean_extracted_entity(m_of.group("focus"))
            if 8 <= len(focus) <= 95:
                extracted_int = focus

    # 3. Check "<Focus> for/in the treatment/diagnosis/prediction of <Context>"
    if not extracted_int:
        m_for = FOR_ACTION_RE.match(clean_topic)
        if m_for:
            focus = _clean_extracted_entity(m_for.group("focus"))
            if 8 <= len(focus) <= 95:
                extracted_int = focus

    # 4. Check Gerund lead ("Targeting/Predicting/Stratifying <Focus> in <Context>")
    if not extracted_int:
        m_ger = GERUND_RE.match(clean_topic)
        if m_ger:
            focus = _clean_extracted_entity(m_ger.group("focus"))
            if 8 <= len(focus) <= 95:
                extracted_int = focus

    # 5. Check "<Focus> and <Outcome> in <Context>"
    if not extracted_int:
        m_and = AND_IN_RE.match(clean_topic)
        if m_and:
            focus = _clean_extracted_entity(m_and.group("focus"))
            if 8 <= len(focus) <= 85:
                extracted_int = focus

    # 6. Split before " in ", " among ", " for ", " after ", " during "
    if not extracted_int:
        m_prep = re.split(r"\s+(?:in|among|for|after|during|following|across)\s+", clean_topic, maxsplit=1, flags=re.I)
        if len(m_prep) == 2 and 10 <= len(m_prep[0].strip()) <= 85:
            cand_focus = _clean_extracted_entity(m_prep[0])
            if cand_focus.lower() != cond_name.lower() and len(cand_focus) >= 10:
                extracted_int = cand_focus

    # Guard against extracting the disease/condition name itself as `int_name` (e.g. from "Clinical Management of Acute GVHD")
    cond_tokens = set(_distinctive_condition_words(cond_name))
    if extracted_int:
        int_no_parens = re.sub(r"\([^)]*\)", "", extracted_int).strip()
        int_tokens = set(_distinctive_condition_words(int_no_parens))
        if not (int_tokens - cond_tokens) or int_no_parens.lower() in cond_name.lower():
            extracted_int = None

    # Construct final paper-native `int_name` and `int_details`
    if extracted_int:
        int_name = extracted_int
        int_details = f"Clinical and translational evaluation of {lower_fn(extracted_int)} in {lower_fn(cond_name)}"
    elif matches_subdomain_disease:
        int_name = base_int_name
        int_details = base_int_details
        if not extracted_comp:
            extracted_comp = base_comp_name
    else:
        stripped_topic = _strip_redundant_population_tail(clean_topic, cond_name)
        st_no_parens = re.sub(r"\([^)]*\)", "", stripped_topic).strip()
        int_tokens = set(_distinctive_condition_words(st_no_parens))
        if not (int_tokens - cond_tokens) or st_no_parens.lower() in cond_name.lower():
            int_name = f"Guideline-directed clinical management and biomarker stratification in {lower_fn(cond_name)}"
        else:
            int_name = _balance_parens(stripped_topic if len(stripped_topic) <= 90 else stripped_topic[:88].rsplit(" ", 1)[0])
        int_details = f"Evaluation of {lower_fn(int_name)} in {lower_fn(cond_name)}"

    # Concise, non-redundant focus phrase for embedding in sentences that already specify `pop_mid`
    focus_clean = _strip_redundant_population_tail(clean_topic, cond_name)
    fc_no_parens = re.sub(r"\([^)]*\)", "", focus_clean).strip()
    focus_tokens = set(_distinctive_condition_words(fc_no_parens))
    if len(focus_clean) < 14 or not (focus_tokens - cond_tokens) or fc_no_parens.lower() in cond_name.lower():
        focus_clean = int_name

    # Construct paper-native `comp_name` (avoiding the word "protocol" at the end so "protocol for" never occurs)
    if extracted_comp:
        comp_name = extracted_comp
    else:
        cond_short = lower_fn(cond_name)
        qtype_comparators = {
            "etiology": f"Unexposed or baseline reference risk cohort in {cond_short}",
            "mechanism": f"Canonical baseline pathway activity or treatment-sensitive phenotype in {cond_short}",
            "natural_history": f"Standard longitudinal clinical surveillance cohort in {cond_short}",
            "prevention": "Routine risk-factor modification or standard surveillance without targeted prophylaxis",
            "screening": "Symptom-prompted clinical evaluation or current guideline screening schedule",
            "diagnosis": "Conventional clinicopathologic, imaging, or laboratory reference standard",
            "classification": "Conventional clinicopathologic staging and consensus taxonomy",
            "prognosis": f"Standard clinical risk score without {lower_fn(int_name)} stratification",
            "treatment_effectiveness": f"Standard-of-care therapy or active control regimen in {cond_short}",
            "comparative_effectiveness": f"Current guideline-recommended standard regimen in {cond_short}",
            "treatment_strategy": f"Standard fixed-duration or unselected dosing regimen in {cond_short}",
            "safety": "Alternative active comparator or standard dose-monitoring regimen",
            "rehabilitation": "Usual physical and functional care without structured rehabilitation pathway",
            "supportive_palliative": "Usual symptom-directed care without integrated supportive pathway",
            "patient_experience": "Routine clinician-directed care without structured patient-reported decision tool",
            "implementation": "Usual guideline dissemination without multifaceted implementation strategy",
            "health_services": "Conventional specialty or hospital-centered care delivery model",
            "public_health": "Status-quo population health policy without targeted equity intervention",
            "measurement": "Legacy clinician-rated or unharmonized outcome assessment instrument",
            "methodology": "Conventional unadjusted analysis or traditional fixed-sample trial design",
        }
        comp_name = qtype_comparators.get(qtype, f"Standard-of-care management in {cond_short}")

    # Refine population and stage if paper title specifies a distinct clinical cohort modifier
    if "metastatic" in tl and "metastatic" not in stage_sev.lower():
        stage_sev = f"Metastatic or advanced stage ({stage_sev})"
    elif ("relapsed" in tl or "refractory" in tl) and "relapsed" not in stage_sev.lower():
        stage_sev = f"Relapsed or refractory stage ({stage_sev})"
    elif ("early-stage" in tl or "early stage" in tl) and "early" not in stage_sev.lower():
        stage_sev = f"Early-stage disease ({stage_sev})"
    elif ("pediatric" in tl or "children" in tl or "neonat" in tl) and "pediatric" not in pop_desc.lower() and "children" not in pop_desc.lower():
        if pop_desc.lower().startswith("adults "):
            pop_desc = "Children and adolescents " + pop_desc[7:]
        elif pop_desc.lower().startswith("patients "):
            pop_desc = "Pediatric and adolescent patients " + pop_desc[9:]
        else:
            pop_desc = f"Children and adolescents with {lower_fn(cond_name)}"
    elif ("elderly" in tl or "older adult" in tl or "geriatric" in tl) and "older" not in pop_desc.lower():
        if pop_desc.lower().startswith("adults "):
            pop_desc = "Older adults " + pop_desc[7:]
        elif pop_desc.lower().startswith("patients "):
            pop_desc = "Older adults and geriatric patients " + pop_desc[9:]
        else:
            pop_desc = f"Older adults with {lower_fn(cond_name)}"

    prim_out, safe_out, pro_out = select_coherent_outcomes(cat_id, raw_title, cond_name)

    return {
        "cond_name": cond_name,
        "int_name": int_name,
        "int_details": int_details,
        "comp_name": comp_name,
        "pop_desc": pop_desc,
        "stage_sev": stage_sev,
        "prio_sub": prio_sub,
        "focus_clean": focus_clean,
        "prim_out": prim_out,
        "safe_out": safe_out,
        "pro_out": pro_out,
    }


def build_natural_statements(
    prob_num: int,
    qtype: str,
    clean_topic: str,
    focus_clean: str,
    cond_name: str,
    pop_desc: str,
    stage_sev: str,
    int_name: str,
    comp_name: str,
    prim_out: str,
    safe_out: str,
    pro_out: str,
    horizon: str,
    settings: list[str],
    prio_sub: list[str],
    lower_fn,
) -> tuple[str, str, str, str]:
    """
    Return `(title, statement, clean_statement, plain_language_statement)` using 100 natural
    clinical question templates (5 variants per question_type x 20 question_types) with zero
    awkward double-splicing.
    """
    variant = prob_num % 5
    focus_mid = lower_fn(focus_clean)
    cond_mid = lower_fn(cond_name)
    pop_mid = lower_fn(re.sub(r"\s*\([^)]{10,}\)", "", pop_desc).strip())
    int_mid = lower_fn(int_name)
    comp_mid = lower_fn(re.sub(r"\s*\([^)]{10,}\)", "", comp_name).strip())
    prio_mid = ", ".join(lower_fn(p) for p in prio_sub[:2])
    setting_mid = ", ".join(s.replace("_", " ") for s in settings[:2])

    cond_sig_words = _distinctive_condition_words(cond_name)
    q_label = qtype.replace("_", " ").title()
    if any(w in clean_topic.lower() for w in cond_sig_words) or cond_name.lower() in clean_topic.lower():
        title = f"{clean_topic} — Open {q_label} Question"
    else:
        title = f"{clean_topic} in {cond_name} ({q_label})"

    clean_statement = (
        f"In {pop_mid}, how does {int_mid} compare with {comp_mid} regarding {prim_out} over {horizon}?"
    )

    templates = {
        "comparative_effectiveness": [
            f"In {pop_mid}, does {focus_mid} improve {prim_out} and reduce {safe_out} over {horizon} compared with {comp_mid}?",
            f"Among {pop_mid}, what is the head-to-head comparative effectiveness of {focus_mid} versus {comp_mid} for achieving superior {prim_out} over {horizon}?",
            f"For {pop_mid} managed in {setting_mid} settings, does {focus_mid} provide a net clinical advantage over {comp_mid} on {prim_out} and {safe_out} through {horizon}?",
            f"When treating {pop_mid} (including {prio_mid}), is {focus_mid} superior or non-inferior to {comp_mid} with respect to {prim_out} at {horizon}?",
            f"Across diverse clinical cohorts of {pop_mid}, how does {focus_mid} compare against {comp_mid} in balancing {prim_out} against {safe_out} over {horizon}?",
        ],
        "treatment_effectiveness": [
            f"Among {pop_mid} ({lower_fn(stage_sev)}), what is the confirmatory clinical efficacy of {focus_mid} compared with {comp_mid} on {prim_out} over {horizon}?",
            f"In {pop_mid}, does therapeutic intervention with {focus_mid} produce durable improvements in {prim_out} over {horizon} relative to {comp_mid}?",
            f"For {pop_mid}, can {focus_mid} significantly improve {prim_out} without unacceptable {safe_out} over {horizon} compared with {comp_mid}?",
            f"What is the magnitude and durability of treatment benefit from {focus_mid} versus {comp_mid} on {prim_out} over {horizon} in {pop_mid}?",
            f"In randomized evaluation of {pop_mid}, does {focus_mid} achieve clinically meaningful gains in {prim_out} and {pro_out} over {horizon} compared with {comp_mid}?",
        ],
        "treatment_strategy": [
            f"For {pop_mid} in {setting_mid} settings, what is the optimal sequencing, dosing threshold, or treatment duration of {focus_mid} versus {comp_mid} over {horizon}?",
            f"In {pop_mid}, does a biomarker- or response-adapted strategy for {focus_mid} outperform {comp_mid} on {prim_out} and {safe_out} over {horizon}?",
            f"Among {pop_mid}, when can therapy be safely de-escalated, switched, or intensified using {focus_mid} compared with {comp_mid} over a {horizon} horizon?",
            f"What is the optimal clinical algorithm for initiating and sequencing {focus_mid} relative to {comp_mid} to maximize {prim_out} over {horizon} in {pop_mid}?",
            f"In {pop_mid}, does dynamic treatment adaptation guided by {focus_mid} improve {prim_out} over {horizon} compared with {comp_mid}?",
        ],
        "diagnosis": [
            f"In {pop_mid} presenting for clinical evaluation, what is the diagnostic accuracy and clinical utility of {focus_mid} compared with {comp_mid} for improving {prim_out} over {horizon}?",
            f"Among {pop_mid}, does integrating {focus_mid} into the diagnostic pathway outperform {comp_mid} in sensitivity, specificity, and downstream {prim_out} over {horizon}?",
            f"For {pop_mid} in {setting_mid} care, can {focus_mid} reduce diagnostic delay and misclassification compared with {comp_mid} while improving {prim_out} over {horizon}?",
            f"What are the externally validated diagnostic performance characteristics of {focus_mid} versus {comp_mid} in {pop_mid}, and does testing improve {prim_out} over {horizon}?",
            f"In {pop_mid}, does diagnostic triage with {focus_mid} provide superior clinical net benefit over {comp_mid} across {horizon}?",
        ],
        "screening": [
            f"In asymptomatic or at-risk {pop_mid}, does structured screening via {focus_mid} compared with {comp_mid} improve early detection and {prim_out} without excess {safe_out} over {horizon}?",
            f"Among {pop_mid} (including {prio_mid}), what is the net balance of benefits, false-positive burden, and overdiagnosis when screening with {focus_mid} versus {comp_mid} over {horizon}?",
            f"For risk-stratified {pop_mid}, does longitudinal surveillance using {focus_mid} reduce late-stage presentation and improve {prim_out} over {horizon} compared with {comp_mid}?",
            f"What screening interval and positivity threshold for {focus_mid} optimize {prim_out} and minimize {safe_out} compared with {comp_mid} in {pop_mid} over {horizon}?",
            f"In population and clinic-based cohorts of {pop_mid}, does early detection through {focus_mid} translate into improved {prim_out} over {horizon} relative to {comp_mid}?",
        ],
        "classification": [
            f"In {pop_mid}, does biological endotyping or phenotypic classification via {focus_mid} outperform {comp_mid} for predicting {prim_out} and guiding therapy over {horizon}?",
            f"Among {pop_mid}, can a reproducible subphenotype taxonomy based on {focus_mid} resolve clinical heterogeneity and improve {prim_out} stratification over {horizon} versus {comp_mid}?",
            f"For {pop_mid}, do disease subtypes defined by {focus_mid} exhibit distinct treatment responses and {prim_out} trajectories over {horizon} compared with {comp_mid}?",
            f"Does integrating multi-modal classification from {focus_mid} refine risk categories beyond {comp_mid} for {prim_out} over {horizon} in {pop_mid}?",
            f"In {pop_mid}, how accurately do diagnostic and molecular criteria derived from {focus_mid} partition patients into actionable strata for {prim_out} over {horizon} versus {comp_mid}?",
        ],
        "prognosis": [
            f"Among {pop_mid}, what is the externally validated prognostic discrimination, calibration, and decision utility of {focus_mid} versus {comp_mid} for stratifying {prim_out} over {horizon}?",
            f"In {pop_mid}, does prognostic modeling incorporating {focus_mid} provide incremental predictive value over {comp_mid} for {prim_out} through {horizon}?",
            f"Across independent cohorts of {pop_mid} (including {prio_mid}), how reliably does {focus_mid} predict long-term {prim_out} and {safe_out} over {horizon} compared with {comp_mid}?",
            f"For {pop_mid}, can baseline and serial assessment of {focus_mid} accurately stratify individual risk of {prim_out} over {horizon} relative to {comp_mid}?",
            f"What is the transportability and clinical net benefit of risk prediction using {focus_mid} compared with {comp_mid} for {prim_out} over {horizon} in {pop_mid}?",
        ],
        "natural_history": [
            f"Across longitudinal cohorts of {pop_mid}, what is the natural trajectory, progression rate, and determinant profile of {prim_out} over {horizon} in relation to {focus_mid} versus {comp_mid}?",
            f"In {pop_mid}, how does the long-term clinical course of {prim_out} evolve over {horizon} when stratified by {focus_mid} compared with {comp_mid}?",
            f"Among {pop_mid}, what are the transition probabilities between clinical states and cumulative incidence of {prim_out} over {horizon} associated with {focus_mid}?",
            f"In unselected prospective follow-up of {pop_mid}, what factors modify longitudinal progression to {prim_out} over {horizon} under {focus_mid} versus {comp_mid}?",
            f"How do baseline phenotypic features and {focus_mid} shape the multi-year natural history of {prim_out} over {horizon} in {pop_mid}?",
        ],
        "etiology": [
            f"In {pop_mid}, what is the causal contribution of {focus_mid}—and its interaction with host susceptibility—to disease development and {prim_out} over {horizon} compared with {comp_mid}?",
            f"Among {pop_mid}, does exposure or predisposition characterized by {focus_mid} independently increase disease incidence and {prim_out} over {horizon} after accounting for confounding?",
            f"Through what causal pathways and dose-response relationships does {focus_mid} drive disease onset and {prim_out} over {horizon} in {pop_mid} relative to {comp_mid}?",
            f"In longitudinal etiologic studies of {pop_mid}, what fraction of disease risk and {prim_out} over {horizon} is attributable to {focus_mid} versus {comp_mid}?",
            f"Using triangulated human causal inference in {pop_mid}, is {focus_mid} a modifiable causal driver of {prim_out} over {horizon}?",
        ],
        "mechanism": [
            f"In human tissues and clinical cohorts of {pop_mid}, what cellular and molecular pathways link {focus_mid} to {prim_out} over {horizon}, and is this axis causally rate-limiting?",
            f"Among {pop_mid}, how does mechanistic perturbation of {focus_mid} alter downstream pathobiology and clinical {prim_out} over {horizon} compared with {comp_mid}?",
            f"In {pop_mid}, what are the primary immunologic, metabolic, or signaling drivers of treatment resistance and disease progression mediated by {focus_mid} over {horizon}?",
            f"Does target engagement of the biological pathway underlying {focus_mid} reverse pathological remodeling and improve {prim_out} over {horizon} in {pop_mid}?",
            f"In translational human studies of {pop_mid}, how do host-microenvironment or genetic interactions involving {focus_mid} govern {prim_out} over {horizon}?",
        ],
        "prevention": [
            f"In at-risk {pop_mid}, does primary or secondary prevention with {focus_mid} compared with {comp_mid} reduce disease incidence and improve {prim_out} over {horizon}?",
            f"Among {pop_mid} (including {prio_mid}), what is the preventive efficacy and number needed to treat of {focus_mid} versus {comp_mid} for averting {prim_out} over {horizon}?",
            f"For high-risk {pop_mid}, does targeted prophylaxis or risk reduction via {focus_mid} durably lower disease incidence and {prim_out} over {horizon} without excess {safe_out}?",
            f"In {pop_mid}, what is the optimal timing and risk threshold for initiating preventive {focus_mid} compared with {comp_mid} to prevent {prim_out} over {horizon}?",
            f"Across community and clinical cohorts of {pop_mid}, does sustained preventive intervention with {focus_mid} reduce long-term {prim_out} over {horizon} relative to {comp_mid}?",
        ],
        "safety": [
            f"Among {pop_mid} exposed to {focus_mid} versus {comp_mid}, what is the comparative incidence, severity, and reversibility of {safe_out} relative to {prim_out} benefit over {horizon}?",
            f"In {pop_mid} (especially {prio_mid}), what patient-level risk factors and monitoring thresholds predict severe {safe_out} during {focus_mid} over {horizon}?",
            f"For {pop_mid} receiving treatment in {setting_mid} settings, how does the long-term safety and pharmacovigilance profile of {focus_mid} compare with {comp_mid} over {horizon}?",
            f"What is the causal risk-benefit trade-off between gains in {prim_out} and cumulative {safe_out} over {horizon} when using {focus_mid} versus {comp_mid} in {pop_mid}?",
            f"In {pop_mid}, what mitigation or dose-modification strategy best prevents treatment-limiting {safe_out} from {focus_mid} while preserving {prim_out} over {horizon}?",
        ],
        "rehabilitation": [
            f"In {pop_mid}, does structured rehabilitation incorporating {focus_mid} improve functional recovery and {prim_out} over {horizon} compared with {comp_mid}?",
            f"Among {pop_mid}, what intensity, timing, and modality of {focus_mid} maximize long-term functional independence and {pro_out} over {horizon} relative to {comp_mid}?",
            f"For {pop_mid} in {setting_mid} rehabilitation pathways, does individualized therapy with {focus_mid} produce durable gains in {prim_out} over {horizon} versus {comp_mid}?",
            f"In {pop_mid}, how does combining neuroplastic or physical conditioning via {focus_mid} compare against {comp_mid} in restoring mobility, function, and {prim_out} over {horizon}?",
            f"Across heterogeneous recovery profiles in {pop_mid}, which subgroups derive clinically meaningful functional benefit from {focus_mid} over {horizon} compared with {comp_mid}?",
        ],
        "supportive_palliative": [
            f"For {pop_mid} experiencing symptom burden, does integrated supportive or palliative care via {focus_mid} improve {pro_out} and {prim_out} over {horizon} compared with {comp_mid}?",
            f"In {pop_mid}, what is the comparative effectiveness of early, structured {focus_mid} versus {comp_mid} for relieving refractory symptoms and improving {pro_out} over {horizon}?",
            f"Among {pop_mid} and their caregivers, does a multidisciplinary supportive pathway centered on {focus_mid} reduce acute care utilization and improve {pro_out} over {horizon}?",
            f"In managing complex symptom clusters in {pop_mid}, how does {focus_mid} compare against {comp_mid} with respect to symptom control, {safe_out}, and {pro_out} over {horizon}?",
            f"For {pop_mid} across {setting_mid} settings, does goal-concordant supportive care integrating {focus_mid} improve {pro_out} and {prim_out} over {horizon} relative to {comp_mid}?",
        ],
        "patient_experience": [
            f"Among {pop_mid}, how does care delivery incorporating {focus_mid} compare with {comp_mid} in improving patient-reported {pro_out} and decision quality over {horizon}?",
            f"In {pop_mid} (including {prio_mid}), what treatment attributes, lived-experience priorities, and trade-offs between {prim_out} and {safe_out} govern preference for {focus_mid} versus {comp_mid}?",
            f"For {pop_mid}, does structured shared decision-making and symptom monitoring around {focus_mid} reduce treatment regret and improve {pro_out} over {horizon} compared with {comp_mid}?",
            f"Across diverse cultural and socioeconomic groups of {pop_mid}, how does {focus_mid} impact treatment burden, quality of life, and {pro_out} over {horizon} relative to {comp_mid}?",
            f"In {pop_mid}, do patient-reported outcomes ({pro_out}) over {horizon} favor {focus_mid} or {comp_mid} when accounting for daily treatment burden and {safe_out}?",
        ],
        "implementation": [
            f"Across real-world {setting_mid} settings caring for {pop_mid}, which implementation strategies for {focus_mid} maximize guideline fidelity, equity across {prio_mid}, and {prim_out} over {horizon}?",
            f"In routine clinical practice for {pop_mid}, does a multifaceted implementation bundle for {focus_mid} outperform {comp_mid} in adoption, sustainment, and {prim_out} over {horizon}?",
            f"What health-system, clinician, and patient-level barriers limit real-world uptake of {focus_mid} in {pop_mid}, and do targeted implementation interventions improve {prim_out} over {horizon}?",
            f"Among {setting_mid} practices managing {pop_mid}, how does structured clinical pathway integration of {focus_mid} compare with {comp_mid} for achieving equitable {prim_out} over {horizon}?",
            f"In scaling evidence-based care for {pop_mid}, does audit-feedback and decision support for {focus_mid} close care gaps across {prio_mid} and improve {prim_out} over {horizon}?",
        ],
        "health_services": [
            f"In health systems managing {pop_mid}, how does organizing care delivery around {focus_mid} compare with {comp_mid} on {prim_out}, resource utilization, and care coordination over {horizon}?",
            f"For {pop_mid} treated across {setting_mid} settings, does a redesigned service model incorporating {focus_mid} reduce avoidable hospitalizations and improve {prim_out} over {horizon} versus {comp_mid}?",
            f"Among {pop_mid} (especially {prio_mid}), what is the cost-effectiveness, capacity impact, and clinical effect on {prim_out} of deploying {focus_mid} compared with {comp_mid} over {horizon}?",
            f"Does multidisciplinary team triage and longitudinal care navigation for {focus_mid} improve {prim_out} and reduce {safe_out} over {horizon} compared with {comp_mid} in {pop_mid}?",
            f"Across regional healthcare networks serving {pop_mid}, how does decentralized or integrated delivery of {focus_mid} compare against {comp_mid} for {prim_out} over {horizon}?",
        ],
        "public_health": [
            f"At the population level among {pop_mid} (including {prio_mid}), does scaling {focus_mid} compared with {comp_mid} reduce health disparities and improve population {prim_out} over {horizon}?",
            f"Across high-burden communities of {pop_mid}, what is the population-level effectiveness and equity impact of public health interventions based on {focus_mid} versus {comp_mid} over {horizon}?",
            f"In population surveillance of {pop_mid}, how do social determinants and policy implementation of {focus_mid} modify {prim_out} and disease burden over {horizon} compared with {comp_mid}?",
            f"For public health programs serving {pop_mid}, does community-wide deployment of {focus_mid} achieve sustainable improvements in {prim_out} over {horizon} relative to {comp_mid}?",
            f"Among underserved subpopulations ({prio_mid}) within {pop_mid}, can structural prevention and care linkage via {focus_mid} close outcome gaps in {prim_out} over {horizon}?",
        ],
        "measurement": [
            f"In clinical research and care for {pop_mid}, what is the construct validity, responsiveness, and minimal clinically important difference of measurement via {focus_mid} for {prim_out} over {horizon}?",
            f"Among {pop_mid}, is {focus_mid} a validated surrogate or core outcome measure capable of reliably predicting {prim_out} over {horizon} compared with {comp_mid}?",
            f"Across multi-center studies of {pop_mid}, how can assay or instrument standardization for {focus_mid} eliminate inter-laboratory variability and improve {prim_out} assessment over {horizon}?",
            f"For evaluating interventions in {pop_mid}, does quantitative longitudinal tracking of {focus_mid} capture clinically meaningful change in {prim_out} and {pro_out} over {horizon} better than {comp_mid}?",
            f"In {pop_mid} (including {prio_mid}), does outcome measurement via {focus_mid} demonstrate cross-population measurement invariance and prognostic validity for {prim_out} over {horizon}?",
        ],
        "methodology": [
            f"In clinical studies of {pop_mid} addressing {focus_mid}, which study design and causal-inference framework best resolves confounding, time-zero bias, and heterogeneity when comparing {int_mid} against {comp_mid} over {horizon}?",
            f"For evaluating {focus_mid} versus {comp_mid} in {pop_mid}, how should estimands, intercurrent events, and competing risks be specified to yield unbiased estimates of {prim_out} over {horizon}?",
            f"In synthesizing heterogeneous evidence on {focus_mid} in {pop_mid}, can individual participant data meta-analysis or target-trial emulation reconcile conflicting findings on {prim_out} over {horizon}?",
            f"What adaptive platform trial architecture or covariate-adjusted randomization scheme optimizes statistical power across {prio_mid} when testing {focus_mid} versus {comp_mid} over {horizon}?",
            f"When validating predictive or causal claims regarding {focus_mid} in {pop_mid}, what external transportability and bias-quantification standards are required to establish {prim_out} over {horizon}?",
        ],
    }

    statement = templates[qtype][variant]

    plain_variants = [
        f"For people with {cond_mid}, how does {int_mid} compare with {comp_mid} for improving {prim_out} and minimizing {safe_out} over {horizon}?",
        f"In patients affected by {cond_mid}, what is the clearest medical evidence on whether {int_mid} works better than {comp_mid} over {horizon}?",
        f"Can {int_mid} safely improve {prim_out} and quality of life ({pro_out}) over {horizon} compared with {comp_mid} for people with {cond_mid}?",
        f"Among individuals with {cond_mid} (including {prio_mid}), does {int_mid} offer meaningful long-term benefits over {comp_mid} across {horizon}?",
        f"What are the real-world benefits and risks of {int_mid} versus {comp_mid} over {horizon} for patients managing {cond_mid}?",
    ]
    plain_stmt = plain_variants[variant]

    return title, statement, clean_statement, plain_stmt