Immel Diana, Hilpert Bernhard, Schwarz Patricia, Hein Andreas, Gebhard Patrick, Barton Simon, Hurlemann René
Department of Psychiatry, School of Medicine and Health Sciences, Carl von Ossietzky Universität Oldenburg, Oldenburg, Germany.
Affective Computing Group, Deutsches Forschungszentrum für Künstliche Intelligenz GmbH, German Research Center for Artificial Intelligence, Kaiserslautern, Germany.
JMIR Form Res. 2025 Mar 26;9:e59527. doi: 10.2196/59527.
Depression is a serious mental health condition that can have a profound impact on the individual experiencing the disorder and those providing care. While psychotherapy and medication can be effective, there are gaps in current approaches, particularly in outpatient care. This phase is often associated with a high risk of relapse and readmission, and patients often report a lack of support. Socially interactive agents represent an innovative approach to the provision of assistance. Often powered by artificial intelligence, these virtual agents can interact socially and elicit humanlike emotions. In health care, they are used as virtual therapeutic assistants to fill gaps in outpatient aftercare.
We aimed to explore the expectations of patients with depression and health care professionals by conducting a qualitative survey. Our analysis focused on research questions related to the appearance and role of the assistant, the assistant-patient interaction (time of interaction, skills and abilities of the assistant, and modes of interaction) and the therapist-assistant interaction.
A 2-part qualitative study was conducted to explore the perspectives of the 2 groups (patients and care providers). In the first step, care providers (n=30) were recruited during a regional offline meeting. After a short presentation, they were given a link and were asked to complete a semistructured web-based questionnaire. Next, patients (n=20) were recruited from a clinic and were interviewed in a semistructured face-to-face interview.
The survey findings suggested that the assistant should be a multimodal communicator (voice, facial expressions, and gestures) and counteract negative self-evaluation. Most participants preferred a female assistant or wanted the option to choose the gender. In total, 24 (80%) health care professionals wanted a selectable option, while patients exhibited a marked preference for a female or diverse assistant. Regrading patient-assistant interaction, the assistant was seen as a proactive recipient of information, and the patient as a passive one. Gaps in aftercare could be filled by the unlimited availability of the assistant. However, patients should retain their autonomy to avoid dependency. The monitoring of health status was viewed positively by both groups. A biofeedback function was desired to detect early warning signs of disease. When appropriate to the situation, a sense of humor in the assistant was desirable. The desired skills of the assistant can be summarized as providing structure and emotional support, especially warmth and competence to build trust. Consistency was important for the caregiver to appear authentic. Regarding the assistant-care provider interaction, 3 key areas were identified: objective patient status measurement, emergency suicide prevention, and an information tool and decision support system for health care professionals.
Overall, the survey conducted provides innovative guidelines for the development of virtual therapeutic assistants to fill the gaps in patient aftercare.
抑郁症是一种严重的心理健康状况，会对患病个体及提供护理的人员产生深远影响。虽然心理治疗和药物治疗可能有效，但当前的治疗方法仍存在不足，尤其是在门诊护理方面。这一阶段通常与高复发率和再入院风险相关，患者常表示缺乏支持。社交互动智能体代表了一种提供援助的创新方法。这些虚拟智能体通常由人工智能驱动，能够进行社交互动并引发类似人类的情感。在医疗保健领域，它们被用作虚拟治疗助手以填补门诊后续护理的空白。
我们旨在通过开展一项定性调查来探索抑郁症患者和医疗保健专业人员的期望。我们的分析聚焦于与助手的外观和角色、助手与患者的互动（互动时间、助手的技能和能力以及互动方式）以及治疗师与助手的互动相关的研究问题。
进行了一项分为两部分的定性研究，以探索这两组人员（患者和护理提供者）的观点。第一步，在一次地区性线下会议上招募了护理提供者（n = 30）。经过简短介绍后，给他们提供了一个链接，并要求他们完成一份基于网络的半结构化问卷。接下来，从一家诊所招募了患者（n = 20），并对他们进行了半结构化的面对面访谈。
调查结果表明，助手应是一个多模态沟通者（语音、面部表情和手势），并对抗消极的自我评价。大多数参与者更喜欢女性助手，或者希望有选择性别的选项。总共有24名（80%）医疗保健专业人员希望有可选择的选项，而患者明显更喜欢女性或多样化的助手。关于患者与助手的互动，助手被视为信息的主动接收者，而患者则是被动接收者。助手随时可用可以填补后续护理的空白。然而，患者应保持自主性以避免产生依赖。两组人员都对健康状况监测持积极看法。希望有生物反馈功能来检测疾病的早期预警信号。在适当的情况下，希望助手具有幽默感。助手所需的技能可概括为提供条理和情感支持，尤其是温暖和能力以建立信任。一致性对于护理人员显得真实很重要。关于助手与护理提供者的互动，确定了三个关键领域：客观测量患者状态、紧急自杀预防以及为医疗保健专业人员提供信息工具和决策支持系统。
总体而言，所进行的调查为开发虚拟治疗助手以填补患者后续护理的空白提供了创新指南。
