《毕业后医学教育中的资源可归因科研损失（RARL-R）：一种量化可补救科研机会差距的反事实框架》
张泽书
《毕业后医学教育中的资源可归因科研损失（RARL-R）：一种量化可补救科研机会差距的反事实框架》
Resource-Attributable Research Loss (RARL-R) in Postgraduate Medical Training: A Counterfactual Framework for Quantifying Remediable Research Opportunity Gaps
Abstract
Research participation is an important component of postgraduate medical training, yet trainees often differ substantially in access to mentorship, protected research time, funding, data, infrastructure, methodological support, and collaboration. Observed scholarly output may therefore reflect not only individual characteristics but also the research-resource environment in which training occurs. We introduce Resource-Attributable Research Loss (RARL-R), a counterfactual framework for quantifying potentially remediable research-opportunity gaps in postgraduate medical education. For trainee i, Rᵢ denotes the observed research-resource profile, R* a prespecified feasible adequacy benchmark, and Rᵢ† = Rᵢ ∨ R* a no-downgrade remediation profile. With μᵢ(r) = E[Yᵢ(r) | Xᵢ], the primary estimand is the resource-attributable expected output gap: Δᵢᴿ = [ μᵢ(Rᵢ†) − μᵢ(Rᵢ) ]₊ When the scholarly outcome has a meaningful zero and ratio-scale interpretation, this gap may additionally be expressed as equivalent research time: τᵢᴿ = T [ 1 − μᵢ(Rᵢ) / μᵢ(Rᵢ†) ]₊ RARL-R does not estimate innate research ability, prescribe how much a trainee should have produced, or provide an adjusted ranking. Instead, it defines an individualized conditional expected counterfactual contrast between observed research resources and a feasible remediation of specified resource deficits. We develop the conceptual and mathematical formulation of RARL-R, specify its measurement and identification requirements, and evaluate estimator recoverability using Monte Carlo simulation. Across 200 replications, a correctly specified g-computation estimator achieved mean RMSEs of 0.209, 0.132, and 0.092 equivalent research years at sample sizes of n = 200, 500, and 1000, respectively. Corresponding mean correlations between true and estimated RARL-R values were 0.938, 0.976, and 0.988. A deliberately crude cohort-reference comparator showed substantially poorer performance. RARL-R provides an opportunity-sensitive framework for studying how modifiable features of postgraduate medical research environments may contribute to differences in expected scholarly development. The simulation demonstrates internal recoverability under a controlled, correctly specified setting, not empirical validity in real trainees. Future evaluation requires prospective multicentre longitudinal validation, prespecified context-sensitive benchmarks, and careful assessment of confounding, positivity, measurement error, model misspecification, and interference.Key words
postgraduate medical education/ research opportunity/ scholarly development/ educational equity/ counterfactual inference/ g-computation/ research environment/ equivalent research time引用本文复制引用
张泽书.《毕业后医学教育中的资源可归因科研损失（RARL-R）：一种量化可补救科研机会差距的反事实框架》[EB/OL].(2026-11-09)[2026-11-10].https://sinoxiv.napstic.cn/article/26468086.学科分类
医学研究方法基金
重点研发计划 脑卒中后遗症患者个体化立体定向经颅磁刺激神经智能调控设备研发( 2026BEG02008 )