Meuwly Didier, Ramos Daniel, Haraksim Rudolf
Netherlands Forensic Institute, Laan van Ypenburg 6, 2497GB The Hague, The Netherlands; University of Twente, Drienerlolaan 5, 7522NB Enschede, The Netherlands.
ATVS - Biometric Recognition Group, Escuela Politecnica Superior, Universidad Autonoma de Madrid, C/Francisco Tomas y Valiente 11, 28049 Madrid, Spain.
Forensic Sci Int. 2017 Jul;276:142-153. doi: 10.1016/j.forsciint.2016.03.048. Epub 2016 Apr 26.
This Guideline proposes a protocol for the validation of forensic evaluation methods at the source level, using the Likelihood Ratio framework as defined within the Bayes' inference model. In the context of the inference of identity of source, the Likelihood Ratio is used to evaluate the strength of the evidence for a trace specimen, e.g. a fingermark, and a reference specimen, e.g. a fingerprint, to originate from common or different sources. Some theoretical aspects of probabilities necessary for this Guideline were discussed prior to its elaboration, which started after a workshop of forensic researchers and practitioners involved in this topic. In the workshop, the following questions were addressed: "which aspects of a forensic evaluation scenario need to be validated?", "what is the role of the LR as part of a decision process?" and "how to deal with uncertainty in the LR calculation?". The questions: "what to validate?" focuses on the validation methods and criteria and "how to validate?" deals with the implementation of the validation protocol. Answers to these questions were deemed necessary with several objectives. First, concepts typical for validation standards [1], such as performance characteristics, performance metrics and validation criteria, will be adapted or applied by analogy to the LR framework. Second, a validation strategy will be defined. Third, validation methods will be described. Finally, a validation protocol and an example of validation report will be proposed, which can be applied to the forensic fields developing and validating LR methods for the evaluation of the strength of evidence at source level under the following propositions.
本指南提出了一种在源层面验证法医鉴定方法的方案，采用贝叶斯推理模型中定义的似然比框架。在源同一性推断的背景下，似然比用于评估痕迹样本（如指纹）和参考样本（如捺印指纹）源自同一或不同来源的证据强度。在制定本指南之前，讨论了其所需概率的一些理论方面，该指南的制定始于一个涉及该主题的法医研究人员和从业人员研讨会之后。在研讨会上，讨论了以下问题：“法医鉴定场景的哪些方面需要验证？”“似然比作为决策过程一部分的作用是什么？”以及“如何处理似然比计算中的不确定性？”。“验证什么？”这一问题侧重于验证方法和标准，而“如何验证？”涉及验证方案的实施。出于几个目标，认为有必要回答这些问题。首先，验证标准[1]中的典型概念，如性能特征、性能指标和验证标准，将通过类比适用于似然比框架或应用于该框架。其次，将定义一种验证策略。第三，将描述验证方法。最后，将提出一个验证方案和一份验证报告示例，可应用于在以下命题下开发和验证用于评估源层面证据强度的似然比方法的法医领域。
