Iter-4360dd15-0050-candidate-audit-originality-check

14 条候选的原始性核对结果（第 50 轮）

本轮目标：逐条核对 Crossref 抽出的 14 条候选，保留原始临床研究，并定位误纳入原因。

结论

Crossref 候选集仍然有效，但其中明确可判定为 非原始研究 的有 4 条：

| 序号 | DOI | 题名片段 | 分类 | 误纳入原因 |
|---|---|---|---|---|
| 35 | 10.1016/j.oret.2017.08.015 | Treatment of geographic atrophy with intravitreal sirolimus: the Age-Related Eye Disease Study 2 ancillary study | ancillary | 标题已明确写出 ancillary study，不是独立原始临床试验 |
| 42 | 10.1155/2019/7684352 | Low-dose sirolimus immunoregulation therapy in patients with active rheumatoid arthritis: a 24-week follow-up ... | follow-up | 标题明确是 24-week follow-up，属于后续随访分析 |
| 68 | 10.1212/CPJ.0000000000000514 | Everolimus for treatment-refractory seizures in TSC: extension of a randomized controlled trial | extension | 标题明确是 extension of a randomized controlled trial，不是原始入组试验 |
| 69 | 10.1111/epi.17099 | Adjunctive everolimus therapy ... results from the postextension phase of EXIST-3 | postextension | 标题明确是 postextension phase，属于延伸阶段结果 |

统计

- 候选总数：14
- 可判定为原始临床研究：10
- 明确非原始研究：4
- follow-up / extension / ancillary = 4

这轮的可复现核对方法

import requests, re
from collections import Counter
doi='10.1016/S2666-7568(23)00258-1'
msg=requests.get(f'https://api.crossref.org/works/{doi}', timeout=30,
                 headers={'User-Agent':'Mozilla/5.0'}).json()['message']
refs=msg['reference']
kw = re.compile(r'rapamycin|sirolimus|everolimus|temsirolimus|rapalog|mTOR inhibitor|mammalian target of rapamycin', re.I)
trialish_markers = re.compile(r'trial|randomi[sz]ed|placebo|phase\s*[1-4]|crossover|double[- ]blind|pilot|feasibility|futility|extension|proof[- ]of[- ]concept|study', re.I)
animal = re.compile(r'fruit fly|drosophila|mouse|mice|rat|rats|murine|animal|in vivo', re.I)
items=[]
for i,r in enumerate(refs,1):
    txt=' | '.join(str(r.get(k,'')) for k in ['article-title','journal-title','author','DOI','unstructured'])
    if kw.search(txt) and trialish_markers.search(txt) and not animal.search(txt):
        title=(r.get('article-title') or r.get('unstructured') or '').strip()
        items.append((i,title,r.get('year') or '',r.get('DOI') or ''))
for i,t,y,d in items:
    tl=t.lower()
    if any(x in tl for x in ['extension','follow-up','postextension','ancillary']):
        if 'ancillary' in tl:
            lab='ancillary'
        else:
            lab='follow-up/extension'
    else:
        lab='primary/interventional'
    print(i, lab, d, t)
print(Counter(
    'ancillary' if 'ancillary' in t.lower() else 'follow-up/extension' if any(x in t.lower() for x in ['extension','follow-up','postextension']) else 'primary/interventional'
    for _,t,_,_ in items
))

解释

这次核对说明：
- 仅用关键词 + trialish 标记，会把 extension / follow-up / ancillary 一并吸进来；
- 因而要想复现“综述真正纳入的原始临床研究”，必须先做 原始性排除，再谈是否还需要进一步按疾病域或干预目的筛选。