Developed by engineers on research team, an Adaptive Robotic Nursing Assistant (ARNA) is a mobile assistive robot that can navigate in cluttered hospital environments and perform chores as a nursing assistant. Some possible tasks are roles of patient sitter and walkers. From the robot’s perspective, these skills still require a high level of intelligence, autonomy, and semi-autonomy to accomplish them in collaboration with a nurse. The necessary intelligence helps efficient communication with humans, autonomous navigation in hospital environments, accurate object identification, and manipulation. For this purpose, the ARNA is equipped with many sensors including cameras, 3D RGB-D sensors, LIDAR, multi-modal skin sensors, and force-torque sensors. With frequent advances in science and technology related to the sensors, development and testing of the robots must be iterative in order to reflect best practices for quality and safety in health care. Testing the use of robots with nursing students in a safe, laboratory learning environment provides a useful foundation for the next series of tests in an actual hospital environment with registered nurses and patients and is the purpose of this study.
Research Questions
- What is the efficacy of ARNA’s sensor systems (navigation, object detection, voice recognition, force detection, manipulation, adaptive physical guidance and assisted teleoperation) in patient sitter and patient walker scenarios?
- Is the use of robots acceptable to nursing students and faculty in patient sitter and patient walker scenarios?
Methods:
Design: The study design is descriptive.
Theoretical Framework: The Technology Acceptance Model (Davis, 1989) served as the theoretical framework of the study. An integrative literature review demonstrated that TAM effectively predicts nurse’s acceptance of health care technology (Strudwick, 2015). Ifinedo (2016) found prediction was strengthened by the inclusion of demographics.
Sample: The sample will consist of students (n=48) enrolled in basic undergraduate and second-degree nursing programs. Students will work in pairs, with one student acting as patient and the second student acting as nurse in the 4 scenarios. Each scenario (sitter and walker) will be repeated 3 times on each date with different student pairs (e.g., sitter scenario scheduled for 1/2019 and 3/2019 with each date involving testing with 3 student pairs; walker scenario scheduled for 2/2019 and 4/2019 with each date involving 3 student pairs). Students will receive course credit for clinical/research hours from either NUR 330 undergraduate nursing research course or NUR 474 Culminating Capstone course for participating in the project. Nurse faculty supervising students in the laboratory (n=5) will also be enrolled in the study as participants.
The study has been approved by the university IRB. All students and faculty will complete informed consents.
Procedure for Sitter intervention: A tablet with a custom programmed Android app will be used to send commands to the robot to request 3 fetching tasks and assess the student’s temperature and blood pressure. Nursing students will either sit or lie on the bed and interact with the robot. As part of an effort to evaluate the task performance, the time to complete the task will be recorded. The student is allowed to issue emergency stop commands during robot operation. Such commands are given priority and allow the robot to return to a safe position. Initial data for this scenario will be collected in January 2019, adaptations will be made to the robot systems and sensors if needed, and the scenario will be repeated in March 2019.
Procedure for Walker intervention: A patient walker walks alongside a patient providing assistance. Tasks performed by a patient walker may include watching for the risk of falls, applying supporting forces for comfort and navigation, pushing wheelchairs, and bringing along medical equipment such as IV poles and respirators. A robot providing patient walker assistance needs capability to: (1) assess the risk of falls by detection of deviation from normal walking patterns, (2) provide sturdy support for the patient, and (3) adaptively respond to the contact load and stabilize the patient. Additionally, a robot may need to push a gurney or bed transporting patients from one place to another in the hospital. Initial data for this scenario will be collected in February 2019, adaptations will be made to the robot systems and sensors if needed, and the scenario will be repeated in April 2019.
Instruments: Data related to the efficacy of the robots will be collected from each sensor by the engineering co-authors, including time to complete each task and accuracy of task completion. Formative data will be discussed by the research team after the January and February testing and will inform the next iteration of the sensors and robots. Efficacy is defined as full functioning of each sensor in each case scenario, with improvements in time and accuracy for the final two scenarios.
Acceptability will be measured using the two subscales of the Technology Acceptance Model (Davis, 1989), Perceived Usefulness and Perceived Ease of Use. Strong reliability and validity data have been demonstrated (Davis, 1989). Each subscale consists of 6 items in a Likert scale format with 7 response options from likely to unlikely. The Technology Acceptance Model is the gold standard for investigation of technology adoption in health care.
EVALUATION
Descriptive data will be used to describe the effectiveness of the robot in each scenario, including the accuracy and the time completion for each task. In addition, qualitative data will be captured from the engineering co-authors in terms of sensor expected and observed performance. The rationale for adaptions of each sensor and the actual change will also be recorded. A full description of the sensors and systems used in the last testing will be captured. The sensor performances in the scenarios beween January-March and February-April will be compared by t-tests.
The acceptability data will be described separately for faculty and students. Mean scores will be reported for each sub-scale. The student data for the 4 scenarios will be combined for regression analysis, with demographic data and perceived ease of use as independent variables and perceived usefulness as the dependent variable.
Results:
Data collection for the study will occur January-April, 2019.
Conclusion:
Technology will continue to be used in health care organizations to enhance the effectiveness of the workforce and to ensure safe and effective health care for patients. Careful testing of the technology for efficacy and acceptability will provide evidence for effective purchasing and implementation decisions, and provide the evidence for change.