Allaart Corinne G, van Houwelingen Sanne, Hilkens Pieter He, van Halteren Aart, Biesma Douwe H, Dijksman Lea, van der Nat Paul B
Department of Computer Science, Faculty of Science, Vrije Universiteit Amsterdam, Amsterdam, Netherlands.
Department of Value Improvement, St. Antonius Hospital, Nieuwegein, Netherlands.
JMIR Hum Factors. 2025 Jan 22;12:e56521. doi: 10.2196/56521.
Patients with cerebrovascular accident (CVA) should be involved in setting their rehabilitation goals. A personalized prediction of CVA outcomes would allow care professionals to better inform patients and informal caregivers. Several accurate prediction models have been created, but acceptance and proper implementation of the models are prerequisites for model adoption.
This study aimed to assess the added value of a prediction model for long-term outcomes of rehabilitation after CVA and evaluate how it can best be displayed, implemented, and integrated into the care process.
We designed a mock-up version, including visualizations, based on our recently developed prediction model. We conducted focus groups with CVA patients and informal caregivers, and separate focus groups with health care professionals (HCPs). Their opinions on the current information management and the model were analyzed using a thematic analysis approach. Lastly, a Measurement Instrument for Determinants of Innovations (MIDI) questionnaire was used to collect insights into the prediction model and visualizations with HCPs.
The analysis of 6 focus groups, with 9 patients, 4 informal caregivers, and 8 HCPs, resulted in 10 themes in 3 categories: evaluation of the current care process (information absorption, expectations of rehabilitation, prediction of outcomes, and decision aid), content of the prediction model (reliability, relevance, and influence on the care process), and accessibility of the model (ease of understanding, model type preference, and moment of use). We extracted recommendations for the prediction model and visualizations. The results of the questionnaire survey (9 responses, 56% response rate) underscored the themes of the focus groups.
There is a need for the use of a prediction model to assess CVA outcomes, as indicated by the general approval of participants in both the focus groups and the questionnaire survey. We recommend that the prediction model be geared toward HCPs, as they can provide the context necessary for patients and informal caregivers. Good reliability and relevance of the prediction model will be essential for its wide adoption.
脑血管意外（CVA）患者应参与设定其康复目标。对CVA结局进行个性化预测将使护理专业人员能够更好地告知患者及非正式照护者。已经创建了几种准确的预测模型，但模型的接受度和正确实施是模型被采用的前提条件。
本研究旨在评估CVA后康复长期结局预测模型的附加价值，并评估如何以最佳方式展示、实施该模型并将其整合到护理过程中。
我们基于最近开发的预测模型设计了一个包括可视化的模拟版本。我们与CVA患者和非正式照护者进行了焦点小组讨论，并与医疗保健专业人员（HCPs）进行了单独的焦点小组讨论。使用主题分析方法分析了他们对当前信息管理和模型的意见。最后，使用创新决定因素测量工具（MIDI）问卷收集HCPs对预测模型和可视化的见解。
对6个焦点小组（9名患者、4名非正式照护者和8名HCPs）的分析产生了3个类别中的10个主题：当前护理过程的评估（信息吸收、康复期望、结局预测和决策辅助）、预测模型的内容（可靠性、相关性和对护理过程的影响）以及模型的可及性（易于理解、模型类型偏好和使用时机）。我们提取了针对预测模型和可视化的建议。问卷调查结果（9份回复，回复率56%）强调了焦点小组的主题。
焦点小组和问卷调查的参与者普遍认可，表明需要使用预测模型来评估CVA结局。我们建议预测模型应面向HCPs，因为他们可以为患者和非正式照护者提供必要的背景信息。预测模型良好的可靠性和相关性对于其广泛采用至关重要。
