Modeling Dialogue in Conversational Cognitive Health Screening Interviews

Proceedings of the 12th Conference on Language Resources and Evaluation (LREC 2020), pages 1167–1177 Marseille, 11–16 May 2020 c⃝ European Language Resources Association (ELRA), licensed under CC-BY-NC 1167 Modeling Dialogue in Conversational Cognitive Health Screening Interviews Shahla Farzana, Mina Valizadeh, Natalie Parde Department of Computer Science University of Illinois at Chicago 851 S. Morgan St., Chicago, IL 60607 {sfarza3, mvaliz2, parde}@uic.edu Abstract Automating straightforward clinical tasks can reduce workload for healthcare professionals, increase accessibility for geographically- isolated patients, and alleviate some of the economic burdens associated with healthcare. A variety of preliminary screening procedures are potentially suitable for automation, and one such domain that has remained underexplored to date is that of structured clinical interviews. A task-specific dialogue agent is needed to automate the collection of conversational speech for further (either manual or automated) analysis, and to build such an agent, a dialogue manager must be trained to respond to patient utterances in a manner similar to a human interviewer. To facilitate the development of such an agent, we propose an annotation schema for assigning dialogue act labels to utterances in patient-interviewer conversations collected as part of a clinically-validated cognitive health screening task. We build a labeled corpus using the schema, and show that it is characterized by high inter-annotator agreement. We establish a benchmark dialogue act classification model for the corpus, thereby providing a proof of concept for the proposed annotation schema. The resulting dialogue act corpus is the first such corpus specifically designed to facilitate automated cognitive health screening, and lays the groundwork for future exploration in this area. Keywords: dialogue act classification, dialogue modeling, cognitive health screening 1. Introduction Recent advancements in artificial intelligence have opened new pathways for improving patient and clinician health- care experiences, with technologies including but not lim- ited to predictive disease modeling, ambient healthcare monitoring, and clinical record-keeping assistance. At the same time, shifting population demographics have ush- ered in new and pressing healthcare concerns. Dementia is one such increasingly critical concern as median popu- lation ages around the globe continue to rise (Tom et al., 2015). Although cures for dementia remain out of reach, researchers believe that early diagnosis can mitigate its ef- fects (Prince et al., 2011). Diagnosis typically requires cog- nitive tests or in-person screening interviews, which can be costly, resource-intensive, and stressful for patients. A conversational agent capable of conducting screening inter- views to elicit the information necessary to perform a pre- liminary assessment of cognitive health could pose an inex- pensive, flexible, low-stress alternative that could simulta- neously increase patient accessibility and reduce clinician workload. Training such an agent to interpret natural language in- put and select suitable follow-up responses cannot be done without appropriate, clinically-relevant dialogue modeling data. Currently, there exists no spoken or text-based cor- pus containing utterances annotated with the dialogue in- tentions and associated characteristics necessary to facili- tate cognitive health screening. In this work, we set out to fill that void. Our contributions are as follows: 1. We establish a dialogue act …

Content truncated to 500 words