We appreciated the letter from Wohlgemut and colleagues regarding the TIC score that we recently published in Critical Care [1, 2]. They highlight the value of this score for the early detection of traumatic coagulopathy, and recognize its ease of use upon hospital admission [2]. However, they challenged several points in our discussion regarding their Bayesian network score [3]. Briefly, they highlight the flexibility in modelling continuous variables, as well as the improved discrimination and calibration of their model. They suggest that prediction of coagulopathy is possible even with missing variables, which the TIC score theoretically cannot do. Finally, they suggest it may no longer be necessary to compromise model performance to achieve a simpler, more user-friendly model, due to advances in user interface design and user experience. It should be noted that we have deliberately chosen to compare our results with the model described by Yet B et al. because the model had good performance metrics and provides a realistic view of the causality of trauma-induced coagulopathy [3].

Their method relies on a set of data including clinical observation, physiological parameters, radiological findings and laboratory values being implemented automatically in a software application, enabling clinical decision making. Unfortunately, this does not currently match the reality of contemporary hospital care [4]. Furthermore, they argue that the Bayesian network scoring system can handle missing variables and estimate them from pre-existing data. We have chosen to include only pre-hospital parameters in our model, as they are immediately available at the time of admission or even during the pre-hospital phase of care. In our model, no parameters are missing at admission, except in rare cases when capillary hemoglobin measurement is not available. The score we described can be easily calculated mentally, driving immediate decision-making for the trauma patient [1]. The author's final point regarding the strength of Bayesian network analysis also reveals its weakness: weak diffusion. To our knowledge, the model has not been published and is therefore not reproducible by other centers, with the added difficulty of understanding and executing this analysis for the physician unfamiliar with such a model. There is also the “black box” issue, of using a model that obscures the weight of each variable, and its associated under-utilization in the medical field [5]. A simpler—but still accurate—scoring system not only provides a better understanding of the presence of trauma-induced coagulopathy, but also enables rapid and easy implementation of corrective therapy.

In conclusion, models such as Bayesian networks are currently rarely used, even though they represent a highly promising tool for the future of medicine. Until these techniques are ready for prime time, we believe it is still useful to have simple, pragmatic and easy-to-use tools to help doctors anticipate patients' needs.

No datasets were generated or analysed during the current study.

1. Brac L, Levrat A, Vacheron C-H, Bouzat P, Delory T, David J-S. Development and validation of the tic score for early detection of traumatic coagulopathy upon hospital admission: a cohort study. Crit Care. 2024;28:168.

   Article PubMed PubMed Central Google Scholar

2. Wohlgemut JM, Pisirir E, Stoner RS, et al. Bayesian networks may allow better performance and usability than logistic regression. Crit Care. 2024;28:234.

   Article PubMed PubMed Central Google Scholar

3. Yet B, Perkins Z, Fenton N, Tai N, Marsh W. Not just data: a method for improving prediction with knowledge. J Biomed Inform. 2014;48:28–37.

   Article PubMed Google Scholar

4. Rajkomar A, Dean J, Kohane I. Machine learning in medicine. N Engl J Med. 2019;380:1347–58.

   Article PubMed Google Scholar

5. Kyrimi E, Dube K, Fenton N, et al. Bayesian networks in healthcare: what is preventing their adoption? Artif Intell Med. 2021;116: 102079.

   Article PubMed Google Scholar

Download references

We thanks Dr Kenji Inaba (Department of surgery, University of Southern California, Los Angeles, USA) for proofreading and correcting the English version of the article.

None.

Authors and Affiliations

1. Département d’Anesthésie Réanimation, Hospices Civils de Lyon, Centre Hospitalier Lyon Sud, Lyon, France

   Charles-Hervé Vacheron & Jean Stéphane David

2. CIRI-Centre International de Recherche en Infectiologie (Team PHE3ID), Univ Lyon, Inserm, U1111, CNRS, UMR5308, ENS Lyon, Université Claude Bernard Lyon 1, 46 Allée d’Italie, 69007, Lyon, France

   Charles-Hervé Vacheron

3. Department of Intensive Care, Annecy‐Genevois Hospital, Annecy, France

   Louis Brac & Albrice Levrat

4. Research on Healthcare Performance RESHAPE, INSERM U1290, Université Claude Bernard Lyon 1, Lyon, France

   Jean Stéphane David

1. Charles-Hervé VacheronView author publications

   You can also search for this author in PubMed Google Scholar

2. Louis BracView author publications

   You can also search for this author in PubMed Google Scholar

3. Albrice LevratView author publications

   You can also search for this author in PubMed Google Scholar

4. Jean Stéphane DavidView author publications

   You can also search for this author in PubMed Google Scholar

Contributions

Charles-Hervé VACHERON: This author helped in writing the first draft of the manuscript, reviewing the manuscript; Jean Stephane DAVID, Albrice LEVRAT & Louis BRAC: These authors helped in reviewing the manuscript.

Corresponding author

Correspondence to Louis Brac.

Ethics approval

Not applicable.

Consent for publication

Not applicable.

Competing interests

JSD did lectures and consulting for LFB (Les Ullis, France).

Publisher's Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Open Access This article is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which permits any non-commercial use, sharing, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if you modified the licensed material. You do not have permission under this licence to share adapted material derived from this article or parts of it. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by-nc-nd/4.0/.

Reprints and permissions

Cite this article

Vacheron, CH., Brac, L., Levrat, A. et al. Scoring system: use and not use from the future to present. Crit Care 28, 303 (2024). https://doi.org/10.1186/s13054-024-05081-0

Download citation

• Received: 23 July 2024

• Accepted: 27 August 2024

• Published: 13 September 2024

• DOI: https://doi.org/10.1186/s13054-024-05081-0

Share this article

Anyone you share the following link with will be able to read this content:

Sorry, a shareable link is not currently available for this article.

Provided by the Springer Nature SharedIt content-sharing initiative

我们感谢 Wohlgemut 及其同事关于我们最近在 Critical Care 上发表的 TIC 评分的来信 [1, 2]。他们强调了该评分对于早期发现创伤性凝血病的价值，并认识到其在入院时易于使用[2]。然而，他们对我们关于贝叶斯网络评分的讨论中的几个观点提出了质疑 [3]。简而言之，他们强调了连续变量建模的灵活性，以及​​模型的辨别力和校准的改进。他们认为，即使缺少变量，也可以预测凝血障碍，而 TIC 评分理论上无法做到这一点。最后，他们建议，由于用户界面设计和用户体验的进步，可能不再需要为了实现更简单、更用户友好的模型而牺牲模型性能。应该指出的是，我们特意选择将我们的结果与 Yet B 等人描述的模型进行比较。因为该模型具有良好的性能指标，并提供了创伤引起的凝血病因果关系的现实观点 [3]。

他们的方法依赖于一组数据，包括临床观察、生理参数、放射学结果和实验室值，这些数据在软件应用程序中自动实现，从而能够做出临床决策。不幸的是，这目前与当代医院护理的现实不符[4]。此外，他们认为贝叶斯网络评分系统可以处理缺失的变量并根据预先存在的数据估计它们。我们选择在模型中仅包含院前参数，因为它们在入院时甚至在院前护理阶段即可立即获得。在我们的模型中，入院时没有丢失参数，除非在极少数情况下无法测量毛细血管血红蛋白。我们描述的分数可以很容易地在心理上计算出来，从而推动创伤患者立即做出决策[1]。作者关于贝叶斯网络分析的优点的最后一点也暴露了它的弱点：弱扩散。据我们所知，该模型尚未发表，因此其他中心无法复制，对于不熟悉此类模型的医生来说，理解和执行此分析会增加难度。还有一个“黑匣子”问题，即使用的模型掩盖了每个变量的权重，及其在医学领域相关的利用不足[5]。更简单但仍然准确的评分系统不仅可以更好地了解创伤引起的凝血病的存在，而且可以快速、轻松地实施纠正治疗。

总之，贝叶斯网络等模型目前很少使用，尽管它们代表了未来医学的一个非常有前途的工具。在这些技术成熟之前，我们相信拥有简单、实用且易于使用的工具来帮助医生预测患者的需求仍然很有用。

当前研究期间没有生成或分析数据集。

1. 
   Brac L、Levrat A、Vacheron CH、Bouzat P、Delory T、David JS。用于入院时早期检测创伤性凝血病的抽动评分的开发和验证：一项队列研究。危重护理。 2024 年；28：168。

   
   文章 PubMed PubMed Central Google Scholar

2. 
   Wohlgemut JM、Pisirir E、Stoner RS 等人。贝叶斯网络可能比逻辑回归具有更好的性能和可用性。危重护理。 2024 年；28：234。

   
   文章 PubMed PubMed Central Google Scholar

3. 
   然而，B、Perkins Z、Fenton N、Tai N、Marsh W。不仅仅是数据：一种利用知识改进预测的方法。 J 生物医学信息。 2014；48：28-37。

   
   文章 PubMed 谷歌学术

4. 
   Rajkomar A、Dean J、Kohane I。医学中的机器学习。 N 英格兰医学杂志。 2019；380：1347–58。

   
   文章 PubMed 谷歌学术

5. 
   Kyrimi E、Dube K、Fenton N 等人。医疗保健中的贝叶斯网络：是什么阻碍了其采用？阿蒂夫智能医学。 2021；116：102079。

   
   文章 PubMed 谷歌学术

 下载参考资料

我们感谢 Kenji Inaba 博士（美国洛杉矶南加州大学外科）对本文英文版本的校对和纠正。

 没有任何。

 作者和单位

1. 
   法国里昂里昂南医院中心里昂民间临终关怀医院麻醉和复苏科

   
   查尔斯·埃尔韦·瓦什隆 (Charles-Hervé Vacheron) 和让·斯特凡·大卫 (Jean Stéphane David)

2. 
   CIRI 国际传染病研究中心（PHE3ID 团队）、里昂大学、Inserm、U1111、CNRS、UMR5308、ENS Lyon、Université Claude Bernard Lyon 1, 46 Allée d'Italie, 69007, Lyon, France

    查尔斯·埃尔维·瓦什隆

3. 
   法国安纳西安纳西热内瓦医院重症监护室

   
   路易斯·布拉克 & 阿尔布里斯·莱夫拉特

4. 
   医疗保健绩效重塑研究，INSERM U1290，克劳德伯纳德里昂第一大学，法国里昂

    让·斯特凡·大卫

1. 
   Charles-Hervé Vacheron查看作者出版物

   
   您也可以在PubMed中搜索该作者 谷歌学术

2. 
   Louis Brac查看作者出版物

   
   您也可以在PubMed中搜索该作者 谷歌学术

3. 
   Albrice Levrat查看作者出版物

   
   您也可以在PubMed中搜索该作者 谷歌学术

4. 
   让·斯特凡·大卫 (Jean Stéphane David)查看作者出版物

   
   您也可以在PubMed中搜索该作者 谷歌学术

 贡献

Charles-Hervé VACHERON：该作者帮助撰写了手稿的初稿，审阅了手稿； Jean Stephane DAVID、Albrice LEVRAT 和 Louis BRAC：这些作者帮助审阅了手稿。

 通讯作者

路易斯·布拉克的通讯。

 道德批准

 不适用。

 同意发表

 不适用。

 利益竞争

JSD 为 LFB（法国 Les Ullis）进行讲座和咨询。

 出版商备注

施普林格·自然对于已出版的地图和机构隶属关系中的管辖权主张保持中立。

开放获取本文根据 Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License 获得许可，该许可允许以任何媒介或格式进行任何非商业使用、共享、分发和复制，只要您给予原作者适当的署名（ s) 和来源，提供知识共享许可证的链接，并指出您是否修改了许可材料。根据本许可，您无权共享源自本文或其部分内容的改编材料。本文中的图像或其他第三方材料包含在文章的知识共享许可中，除非材料的出处中另有说明。如果文章的知识共享许可中未包含材料，并且您的预期用途不受法律法规允许或超出了允许的用途，则您需要直接获得版权所有者的许可。要查看此许可证的副本，请访问 http://creativecommons.org/licenses/by-nc-nd/4.0/。

 转载和许可

 引用这篇文章

 下载引文

• 
  收件日期： 2024 年 7 月 23 日

• 
  接受日期： 2024 年 8 月 27 日

• 
  发布日期： 2024 年 9 月 13 日

 分享这篇文章

您与之分享以下链接的任何人都可以阅读此内容：

抱歉，本文目前没有可共享的链接。

由 Springer Nature SharedIt 内容共享计划提供