Current Virtual Mental Health Assistants (VMHAs) provide counseling and
suggestive care. They refrain from patient diagnostic assistance because they
lack training in safety-constrained and specialized clinical process knowledge.
In this work, we define Proknow as an ordered set of information that maps to
evidence-based guidelines or categories of conceptual understanding to experts
in a domain. We also introduce a new dataset of diagnostic conversations guided
by safety constraints and Proknow that healthcare professionals use. We develop
a method for natural language question generation (NLG) that collects
diagnostic information from the patient interactively. We demonstrate the
limitations of using state-of-the-art large-scale language models (LMs) on this
dataset. Our algorithm models the process knowledge through explicitly modeling
safety, knowledge capture, and explainability. LMs augmented with ProKnow
guided method generated 89% safer questions in the depression and anxiety
domain. The Explainability of the generated question is assessed by computing
similarity with concepts in depression and anxiety knowledge bases. Overall,
irrespective of the type of LMs augmented with our ProKnow, we achieved an
average 82% improvement over simple pre-trained LMs on safety, explainability,
and process-guided question generation. We qualitatively and quantitatively
evaluate the efficacy of the proposed ProKnow-guided methods by introducing
three new evaluation metrics for safety, explainability, and process knowledge
adherence.

通过使用 Proknow 和具有引导作用的自然语言问答生成方法，我们基于 449 个抑郁症和焦虑症患者的诊断谈话开发出一种具有安全约束和专业知识的诊断对话数据集。我们证明了在这个数据集上使用最先进的大规模语言模型的局限性，通过明确建模安全、知识捕获和可解释性，我们的算法提高了 82％以上的生成问题的安全性、解释性和过程指导。

ProKnow: 面向安全约束和可解释性的过程知识用于心理健康诊断辅助问答生成

ProKnow: Process Knowledge for Safety Constrained and Explainable  Question Generation for Mental Health Diagnostic Assistance

AI systems have been widely adopted across various domains in the real world.
However, in high-value, sensitive, or safety-critical applications such as
self-management for personalized health or food recommendation with a specific
purpose (e.g., allergy-aware recipe recommendations), their adoption is
unlikely. Firstly, the AI system needs to follow guidelines or well-defined
processes set by experts; the data alone will not be adequate. For example, to
diagnose the severity of depression, mental healthcare providers use Patient
Health Questionnaire (PHQ-9). So if an AI system were to be used for diagnosis,
the medical guideline implied by the PHQ-9 needs to be used. Likewise, a
nutritionist's knowledge and steps would need to be used for an AI system that
guides a diabetic patient in developing a food plan. Second, the BlackBox
nature typical of many current AI systems will not work; the user of an AI
system will need to be able to give user-understandable explanations,
explanations constructed using concepts that humans can understand and are
familiar with. This is the key to eliciting confidence and trust in the AI
system. For such applications, in addition to data and domain knowledge, the AI
systems need to have access to and use the Process Knowledge, an ordered set of
steps that the AI system needs to use or adhere to.

AI 系统需要遵循专家制定的指导方针和流程，使用过程知识和领域知识，并提供用户可理解的解释，以在高价值、敏感或安全关键的应用程序中获得信任和信心.