Digitalized Cognitive Assessment mediated by a Virtual Caregiver

Matteo Luperto, Marta Romeo, Francesca Lunardini, Nicola Basilico, Ray Jones, Angelo Cangelosi, Simona Ferrante, N. Alberto BorgheseIJCAI 2018 · DOI

TL;DR

This paper introduces a novel framework for at-home cognitive assessment of Mild Cognitive Impairment (MCI) using digitalized tests and a Virtual Caregiver (VC). The system aims to facilitate early detection of MCI by providing a user-friendly, multimodal experience that mimics traditional caregiver interaction, bridging the gap between home-based monitoring and clinical evaluation.

What Problem Does This Paper Address?

The aging global population presents a growing challenge in healthcare costs, particularly concerning age-related cognitive decline. Mild Cognitive Impairment (MCI) is a critical intermediate phase before dementia, and early detection is crucial for timely intervention. Currently, neuropsychological assessments for MCI are typically performed in clinical settings by a human clinician, which can be resource-intensive and may not facilitate continuous monitoring. This paper addresses the need for accessible, at-home cognitive assessment tools that can support early detection and monitoring of MCI, thereby reducing the burden on healthcare systems and improving patient outcomes.

Methods

The proposed framework consists of two main components:

  • Digitalized Cognitive Tests: These are web-based versions of established paper-and-pencil neuropsychological tests, specifically adapted for a tablet interface (Samsung Galaxy Tab A6 with S-Pen) to enhance user experience. The tests include digital versions of the Trail Making Test (TMT-A and TMT-B) and the Bells Test. The digital implementation aims to maintain structural similarity to the original tests while optimizing for the touch interface and reducing distractions. The system automatically collects and analyzes test results.
  • Virtual Caregiver (VC): This is an AI agent that acts as a multimodal interface, guiding the user through the cognitive tests. The VC interacts with the user via speech and a graphical interface, providing instructions, overseeing test execution, and detecting errors. It utilizes cloud-based speech recognition (Google Cloud Platform Speech API) and speech synthesis (Acapela Voice As a Service). The VC's role is to mimic the multifaceted support provided by a human caregiver during assessments.

The interaction between the VC and the digital tests is managed through a system that exchanges messages to ensure synchronized operation, allowing either component to detect changes in state and update accordingly.

Datasets

Not discussed in the paper. The paper focuses on the framework and methodology rather than specific datasets for training or evaluation.

Key Insights and Ideas

  • Bridging the Home-Clinical Gap: The system effectively bridges the gap between at-home cognitive monitoring and formal clinical assessment by providing a structured, yet user-friendly, digital assessment experience.
  • Multimodal Interaction for Accessibility: The use of a multimodal interface (speech and graphical) is crucial for making cognitive assessments accessible to older adults, who may have sensory or motor impairments that hinder the use of traditional interfaces.
  • Mimicking Human Caregiver Role: The Virtual Caregiver's design, which aims to replicate the instructional and supportive role of a human caregiver, is key to improving user engagement and test acceptability.
  • Early Detection as a Primary Goal: The system's primary objective is not to replace clinical assessment but to serve as a tool for early cognitive monitoring, identifying patterns indicative of MCI to trigger timely clinical evaluation.

Key Results

The paper describes a demonstration scenario and the flowchart of the interaction between the VC and the digital tests. While specific quantitative results on accuracy or performance metrics are not detailed, the paper emphasizes the successful implementation of a framework that can:

  • Provide digitalized versions of established cognitive tests.
  • Guide users through these tests using a multimodal Virtual Caregiver.
  • Collect and analyze test data automatically.
  • Facilitate a more accessible and potentially continuous form of cognitive monitoring at home.

Limitations and Future Work

Limitations:

  • The paper acknowledges that the system does not intend to replace clinical assessment performed by a clinician.
  • The described demonstration is a proof-of-concept, and extensive clinical validation is likely required.

Future Work:

  • The authors suggest that the framework is a building block for a wider multi-actor platform (Movecare project) that aims to provide assistance, activities, and transparent monitoring to elders at home.
  • The system's potential for early cognitive monitoring and identifying patterns correlated with MCI is highlighted as a direction for future development and validation.

Summary updated

BibTeX
@inproceedings{luperto2018digitalized,
  author = {Matteo Luperto and Marta Romeo and Francesca Lunardini and Nicola Basilico and Ray Jones and Angelo Cangelosi and Simona Ferrante and N. Alberto Borghese},
  title = {{Digitalized Cognitive Assessment mediated by a Virtual Caregiver}},
  booktitle = {International Joint Conference on Artificial Intelligence},
  year = {2018},
  doi = {10.24963/ijcai.2018/856},
}
Figure