Development of an Autodiagnostic Adaptive Precision Trainer for Decision Making (ADAPT-DM)

Abstract

The Autodiagnostic Adaptive Precision Trainer for Decision Making (ADAPT-DM) is a framework for adaptive training of decision making skills. The training challenge is that decision making behavior is mostly unobservable with traditional behavioral measures, which generally only give access to outcome performance. This article describes the ADAPT-DM framework, which utilizes physiological sensors, specifically electroencephalography and eye tracking, to detect indicators of implicit cognitive processing relevant to decision making and accomplish the granularity required to pinpoint and process level issues. Using these advanced measures, the trainee's performance on these cognitive processes can be assessed in real time and used to drive smart adaptations that individualize training. As a proof of concept, the ADAPT-DM framework was conceptually applied to the contact evaluation task in submarine navigation. Simulated data from 75 students, grouped into three levels of expertise (novice, intermediate, and expert), were used for principal component analysis to identify skill dimensions that reflect proficiency levels. Then ADAPT-DM's composite diagnosis was performed, which uses an expertise model that integrates automated expert modeling for automated student evaluation machine learning models with eye tracking and electroencephalography data to assess which proficiency level the simulated students actions were most similar to. Based on additional assessments, the diagnostic engine is able to determine whether the student (a) performs to criterion, in which case training could be accelerated, (b) is in an optimal learning state, or (c) is in a nonoptimal learning state for which remediation or mitigation are needed. Using root cause analysis techniques, the ADAPT-DM process level measures then allow instructors to pinpoint where in the decision making process breakdowns occur, so that optimal training adaptations can be implemented.

Open PDF

Document Details

Document Type
Technical Report
Publication Date
Jun 01, 2010
Accession Number
ADA530322

Entities

People

  • Angela Carpenter
  • Kelly Hale
  • Meredith Carroll
  • Sven Fuchs

Organizations

  • Design Interactive (United States)

Tags

Communities of Interest

  • Autonomy
  • C4I
  • Human Systems
  • Space

DTIC Thesaurus Topics

  • Adaptive Training
  • Cognition
  • Cognitive Science
  • Cognitive Systems Engineering
  • Cognitive Workload
  • Computational Science
  • Computers
  • Data Mining
  • Failure Mode And Effect Analysis
  • Human Systems Integration
  • Human-Computer Interaction
  • Machine Learning
  • Psychology
  • Situational Awareness
  • Students
  • Supervised Machine Learning
  • Test And Evaluation

Readers

  • Adaptive Control and Estimation with Uncertainty in Dynamic Systems.
  • Instructional Design and Training Evaluation.
  • Team-Based Human-Centered Cognitive Task Decision Making and Information Performance.

Technology Areas

  • AI & ML