robot for health data acquisition among older adults: a pilot randomised controlled cross-over trial - touch screen smart table

Health care professionals (HCP)
Faced with increased demand for assessment of important health status measures, such as patients-
Measurement of results reported (PROM)
It takes time.
The aim of this study was to investigate the validity and acceptability of robot assistants using HCP, and to examine the hypothesis that robots can independently obtain ball data from older adults.
Design a randomized controlled cross-test
In a study, a social robot and a nurse distributed three ball questionnaires with 52 questions.
Setting up clinical clinics in community units
Living in the elderly.
Forty participantstwo community-
Elderly people living (mean age: 77. 1 years, SD: 5.
7 years, female 45%).
The main result of the measurement is the task time required for robot-patient and nurse-patient interaction.
The secondary result is the similarity of the data and the percentage of the robot's autonomous interaction.
Two values were obtained from the questionnaire (Robot and nurse
For three vulnerable indicators
Elastic.
Data similarity is determined by using Bland-comparing these index values
Kappa of Cohen (κ)
Correlation coefficient within class (ICC).
Acceptability was assessed using questionnaires.
Results The average interview time of the robot was 16. 57u2009min (SD=1. 53u2009min)
This is not longer than the interview time of the nurse (14. 92u2009min, SD=8. 47u2009min; p=0. 19).
Three bland.
Altman's plot shows a moderate to substantial agreement between vulnerability.
Points of existence and Resilience (κ=0. 61, 0. 50 and 0. 45, and ICC=0. 79, 0. 86 and 0.
66 respectively).
The robot independently completed 39 of 42 interviews (92. 8%).
Conclusion social robots can help HCPs effectively and acceptably by interviewing the elderly.
Background/objectives of health care professionals (HCP)
Faced with increased demand for assessment of important health status measures, such as patients-
Measurement of results reported (PROM)
It takes time.
The aim of this study was to investigate the validity and acceptability of robot assistants using HCP, and to examine the hypothesis that robots can independently obtain ball data from older adults.
Design a randomized controlled cross-test
In a study, a social robot and a nurse distributed three ball questionnaires with 52 questions.
Setting up clinical clinics in community units
Living in the elderly.
Forty participantstwo community-
Elderly people living (mean age: 77. 1 years, SD: 5.
7 years, female 45%).
The main result of the measurement is the task time required for robot-patient and nurse-patient interaction.
The secondary result is the similarity of the data and the percentage of the robot's autonomous interaction.
Two values were obtained from the questionnaire (Robot and nurse
For three vulnerable indicators
Elastic.
Data similarity is determined by using Bland-comparing these index values
Kappa of Cohen (κ)
Correlation coefficient within class (ICC).
Acceptability was assessed using questionnaires.
Results The average interview time of the robot was 16. 57u2009min (SD=1. 53u2009min)
This is not longer than the interview time of the nurse (14. 92u2009min, SD=8. 47u2009min; p=0. 19).
Three bland.
Altman's plot shows a moderate to substantial agreement between vulnerability.
Points of existence and Resilience (κ=0. 61, 0. 50 and 0. 45, and ICC=0. 79, 0. 86 and 0.
66 respectively).
The robot independently completed 39 of 42 interviews (92. 8%).
Conclusion social robots can help HCPs effectively and acceptably by interviewing the elderly.
Introduction a set of important medical data is composed of patient responses to the medical questionnaire, such as patient-
Measurement of results reported (PROM).
1-3 Dance data provide basic information about the patient's health status and the effectiveness of the care provided.
1 A survey of nearly 100 clinical trials published between 2007 and 2013 found that 27% of the trials used PROMs;
However, it is a time to interview an older patient for a dance --
Consumption management tasks of health care professionals (HCP)
Their time is often very limited.
The problem is further exacerbated by the increasing shortage of medical staff;
As a result, patients are often asked to provide their own data using a computer, tablet or smartphone.
Many patients, especially elderly patients, have difficulty using digital technology solutions due to lack of digital literature or disability (eg, low vision).
9 in the case of requiring elderly patients to fill out the form through the Internet
High response rate (74%)
Increase with age
Social robots can be seen as humanoid robots that one person can interact with like another person.
They are becoming potential support technologies for HCPs and their potential to participate in patient data collection is currently under investigation.
The application of social robots in the nursing of elderly patients has been widely investigated14-21;
However, as far as we know, their ability to conduct a health status questionnaire independently in a hospital setting has not yet been assessed.
Therefore, our research increases the scarcity of research on robots.
Assist in the investigation.
Our assumption is that if the HCP conducts the survey, there is no significant difference between the task time of the social robot that independently conducts the long-time ball Survey and the task time (
Current practice).
We have demonstrated the acceptability and effectiveness of social robots interacting with a group of older volunteers, but have not compared this to routine care.
In this study, we aim to test our assumptions with the community.
A specially designed robot participant interactive program is used on Pepper robots to live in the elderly.
This social robot has a friendly appearance and a height of 1.
2 m as the first choice for the elderly.
The speech recognition ability of 24 peppers is based on matching with a pre-programmed set of words.
The robot further combines the face recognition in the camera image with the direction of the voice signal, turning the head to the person talking.
We measured the percentage of task completion without HCP intervention, the data obtained through HCP and the protocol between the robot
The survey was conducted and the task duration and acceptability of these data collection methods were compared.
The industrial design experiment is designed as a non-
Cross-blind random control
Experiments on data acquisition by robot comparison (
Participating robots: RP)and nurse (
Nurse-participant: NP)
Interaction with older participants.
Each participant answered three questionnaires for robot management at one meeting and three questionnaires for nurse management at another meeting. This within-subject cross-
Excessive design was selected to minimize variance unrelated to the change signal and to better detect differences in appreciation between HCP and the robot. A 2-
In order to minimize the learning effect, a weekly Flushing period was used between the two sessions. The 2-
The time of the week is a compromise between a longer flush, in this way
Excessive effects may be further reduced, and an increase in the concurrent incidence of these elderly subjects may limit comparability.
Researchers randomly assigned participants to two study groups using their signatures
In the latest situation, one group met nurses in the first meeting, the robot in the second meeting, and the other group encountered the opposite order of the interviewer.
In order to avoid the influence of study and boredom, this balance is adopted.
Participants were recruited through newspaper advertisements or local senior adult organizations from November 2017 to January 2018.
The inclusion criteria are as follows: over 70 years old, Dutch speaking, independent living, no cognitive impairment.
The focus of interaction design is the patient's self.
Assess their current weaknesses
The ability to cope with diseases and the ability to respond.
These assessments use a summary of topics (TOPICS-SF)
Personal Happiness Index (PWI)
26 and the resilience table are 27, respectively.
During the RP meeting, the nurse welcomed the participants and accompanied them to the examination room with the robot (figure 1).
The nurse and the participants were sitting opposite the robot, and the nurse explained that she had a new robot assistant that helped her with her administrative tasks by verbally managing the questionnaire.
The participant received an instruction card explaining his conversation options, which also shows a large font for ease of reading on the robot screen (
Online Supplementary Figure 1).
This allows participants to consider options that are not related to memory function and select the most appropriate answer.
After a short training conversation for the participants, the nurses instructed them how to command the robot to start the actual RP interaction and then leave the room for the participants to be alone with the robot.
Under the start command of the participant, the robot began to receive a questionnaire asking for confirmation of each answer it registered.
At the end of the interview, the robot thanked the participants.
This procedure is further detailed in Appendix 1 of the online supplement.
Supplementary information [bmjqs-2018-008977supp001. pdf]
Download the robot of the new tabDownload figureOpen powerpointFigure month to interview the pepper.
The NP interactive program is similar to the RP program, except that the nurse interviewed the participants, presented the questions and answer options on the paper form, and recorded the given answers.
Results The main outcome measure was the time required to complete the questionnaire in RP and NP interactions.
The secondary result measure is the data similarity, the percentage of RP interaction completed independently (
HCP without intervention).
We also evaluated participants' opinions on acceptability of clinical interviews using robotic technology.
Sample size dependent on t-using G * power pairtest (two tailed)
In the subjects, the sample size of 29 36 people was calculated to detect 0.
5 Impact on efficiency (time)
The ball is complete when the power is set to 0.
90, and use the alpha of 0. 10.
Data analysis robots record RP answers electronically, while NP answers are recorded on paper.
All data is stored in the Castor data management system.
30 analyze the data with SPSS statistical software (V. 22; IBM)
And Microsoft Excel (Office 365;
Microsoft, Redmond, Washington, USA).
The percentage of autonomous completion is determined as the number of RP interactions that do not interrupt events, as a percentage of the total number of RP interactions.
The interrupt event is defined as any HCP intervention required for the robot to further continue the interview, for example, due to a robot failure.
The length of the task for each interview is calculated as the time difference between the first and last answer.
In RP interaction, the time is recorded electronically by the robot, and for NP interaction, the time is calculated by the interview record.
The data similarity of the three indicators is calculated.
Index of vulnerability (FI)
From 18-Subject matter-
SF that does not include any missing values, 31.
FI is used to classify participants as weak, weak, or robust, where weak is equivalent to two to five deficits reported in the topic --
SF questionnaire (
FI to 0. 1–0. 25).
32 The overall PWI is calculated based on the average score of the PWI problem 2-8, which is converted to a value on a scale from 0 to 100, and the higher the score reflects a better resultbeing.
PWI is divided into three categories: low, medium and high, cutting
Practically defined as the total mean sd for all index values of PWInurse and PWIrobot.
Resilience Index generated by resilience table (RI)
Between 25 and 100 points, the higher the score, the higher the resilience.
Using gender conversion RIs-
Specific specification values, then divided into low, medium, or high
Category of resilience.
Thus, six indicators were obtained for each participant: FInurse, PWInurse, RInurse, FIrobot, PWIrobot, and RIrobot.
Following the methods of Bland and Altman, 33 34 analyzed the consistency between the two measures of each index using the scatter plot of the sample, where S (x,y)=((
Robot)
/2. index nurse-index robot).
Correlation coefficient within class (ICC)
For continuous measures (the indexes)
Using two in SPSS-
A hybrid model that is absolutely consistent between robot and nurse measurements was analyzed.
In addition, Cohen's kappa (κ)
Calculated ordinal metrics for analyzing interactions
The evaluator agreement between RP and NP.
The feature of kappa