| dc.description.abstract | Limb amputation can cause severe functional disability for the performance of activities of daily living (ADLs). Amputee patients use prosthetic devices (PDs) to perform ADLs. PDs require a substantial amount of cognitive resources, and some users reject their devices due to poor usability. However, very few studies have investigated usability issues, and they mainly used subjective methods such as questionnaires. In addition, no prior studies classified cognitive workload of using PDs in early stage of the design process. To fill out these research gaps, the objectives of this work were to: 1) Provide an objective usability evaluation of prosthetic devices, 2) Develop a human performance modeling tool to assess the usability and cognitive workload of upper limb PDs (i.e., HPM-UP: Human Performance Model for Upper limb Prostheses) and (3) Validate the model with experimental data.
Chapters 1 and 2 provide a review of literature on usability evaluation of prosthetic devices and human performance modeling approaches. In Chapter 3, the computational formulations for each dimension of HPM-UP was developed based on the literature and existing theories: (1) adaptive learning curve formulation was used for calculating learnability; (2) based on the learnability formula, error rate was calculated using a natural exponential function, (3) memorability was calculated based on the ACT-R declarative module, (4) efficiency was formulated based on user task performance, (5) satisfaction was formulated based on the expectation confirmation theory, and (6) cognitive workload classification model was developed with the Naïve Bayes algorithm.
Chapters 4 and 5 focused on validating the HPM-UP. A human subject experiment (Experiment 1) was conducted with 30 able-bodied participants using three types of PDs. Hypotheses were formulated for each dimension of HPM-UP to test if there was any significant difference among the human-subject data, HPM-UP estimates, and the benchmark model estimates. A second human subject study (Experiment 2) was conducted with 20 able-bodied participants to validate the HPM-UP in a virtual environment. The findings of both experiments suggested that there were no significant differences between the human subject data and HPM-UP estimates. However, there were significant differences between human subject data and benchmark model. Also, there were significant differences between the HPM-UP estimates and the benchmark model outcomes.
HPM-UP can be run using a graphical user interface (GUI) and do not require hard-coding to run the model. It is a first comprehensive usability evaluation package developed in an R Shiny package format and released on GitHub, which can be used by other researchers, designers, or clinicians. The outcomes of this research are expected to be useful for both researchers and practitioners as this is the first computational modeling approach to assess the usability of prosthetic devices early in the design cycle. In addition, the results are expected to provide a basis to enhance the design of prosthetic devices to reduce cognitive workload and improve device usability. | |
