Suominen Hanna, Lehtikunnas Tuija, Back Barbro, Karsten Helena, Salakoski Tapio, Salanterä Sanna
Turku Centre for Computer Science (TUCS), Joukahaisenkatu 3-5 B, 20520 Turku, Finland.
Int J Med Inform. 2007 Oct;76 Suppl 2:S293-301. doi: 10.1016/j.ijmedinf.2007.05.006. Epub 2007 Jun 28.
The present study discusses ethics in building and using applications based on natural language processing in electronic nursing documentation. Specifically, we first focus on the question of how patient confidentiality can be ensured in developing language technology for the nursing documentation domain. Then, we identify and theoretically analyze the ethical outcomes which arise when using natural language processing to support clinical judgement and decision-making. In total, we put forward and justify 10 claims related to ethics in applying language technology to nursing documents.
A review of recent scientific articles related to ethics in electronic patient records or in the utilization of large databases was conducted. Then, the results were compared with ethical guidelines for nurses and the Finnish legislation covering health care and processing of personal data. Finally, the practical experiences of the authors in applying the methods of natural language processing to nursing documents were appended.
Patient records supplemented with natural language processing capabilities may help nurses give better, more efficient and more individualized care for their patients. In addition, language technology may facilitate patients' possibility to receive truthful information about their health and improve the nature of narratives. Because of these benefits, research about the use of language technology in narratives should be encouraged. In contrast, privacy-sensitive health care documentation brings specific ethical concerns and difficulties to the natural language processing of nursing documents. Therefore, when developing natural language processing tools, patient confidentiality must be ensured. While using the tools, health care personnel should always be responsible for the clinical judgement and decision-making. One should also consider that the use of language technology in nursing narratives may threaten patients' rights by using documentation collected for other purposes.
Applying language technology to nursing documents may, on the one hand, contribute to the quality of care, but, on the other hand, threaten patient confidentiality. As an overall conclusion, natural language processing of nursing documents holds the promise of great benefits if the potential risks are taken into consideration.
本研究探讨在电子护理记录中构建和使用基于自然语言处理的应用程序时的伦理问题。具体而言，我们首先关注在为护理记录领域开发语言技术时如何确保患者隐私。然后，我们识别并从理论上分析在使用自然语言处理来支持临床判断和决策时出现的伦理结果。我们总共提出并论证了10条与将语言技术应用于护理记录的伦理相关的主张。
对近期与电子病历或大型数据库利用中的伦理相关的科学文章进行了综述。然后，将结果与护士伦理准则以及芬兰涵盖医疗保健和个人数据处理的立法进行了比较。最后，附上了作者在将自然语言处理方法应用于护理记录方面的实践经验。
具备自然语言处理能力的患者记录可能有助于护士为患者提供更好、更高效和更个性化的护理。此外，语言技术可能会增加患者获得关于其健康的真实信息的可能性，并改善叙述的性质。由于这些益处，应鼓励对语言技术在叙述中的使用进行研究。相比之下，对隐私敏感的医疗保健记录给护理记录的自然语言处理带来了特定的伦理问题和困难。因此，在开发自然语言处理工具时，必须确保患者隐私。在使用这些工具时，医护人员应始终对临床判断和决策负责。还应考虑到在护理叙述中使用语言技术可能会因使用为其他目的收集的记录而威胁患者权利。
一方面，将语言技术应用于护理记录可能有助于提高护理质量，但另一方面，可能会威胁患者隐私。总体而言，如果考虑到潜在风险，护理记录的自然语言处理有望带来巨大益处。
