W-25: The Use of Voice Assistant Technology to Increase Engagement in Clinical Trials
Poster Presenter
Karin Beckstrom
Innovation Lab
ERT United States
Objectives
This poster will report on a project to evaluate the ability for voice assistance technology to collect direct input from participants during a clinical trial and to evaluate if alternative methods of interaction would enhance the patient reported outcome experience.
Method
An interactive voice questionnaire was created which allowed subjects to hear the questions and verbally respond for 5 quality of life questions using commercially available artificial intelligence technology. Subjects were then asked about their experience of completing the questionnaire.
Results
Over a period of several months, 12 volunteers listened to and responded to the voice questionnaire. When asked about the experience people called out several advantages to the use of voice assistance technology, as well as potential concerns to be addressed prior to use in a full clinical trial.
Advantages:
Easy to use: It is a different experience than a typical app on a cellphone. The ability to launch the questionnaire on demand by just asking for it made the experience feel easy.
Hands-free: People could listen and answer the questionnaire while being physically engaged in other tasks, such as getting dressed for work.
Inclusive: Enables participation from people who have dexterity or visual differences and find app use difficult.
Enjoyable: This was a novel experience for many of the respondents and they were openly delighted with task.
Concerns:
The technology is so new that people with significant accents had trouble being understood, which made completion more complicated.
The artificial intelligent assistant is always listening. A portion of the volunteers expressed concerns that the device was always in active listening mode and could be recording data outside of the questionnaire. The information that there is a switch to turn off the listening mode and only questionnaire responses are transmitted for the study will need to be called out and explained in an informed consent.
Data privacy: The data is initially stored in a secured Amazon Web Services Database but once transferred to the clinical database any rights Amazon has to use anonymized data for purposes of supporting and enhancing the service will need to be explained and listed in an informed consent.
Features in the questionnaire to address regulatory compliance, for example allowing answers to be changed before submission does increase the amount of time to complete the questionnaire. Further usability testing and updates are needed to minimize the impact.
Conclusion
Overall conclusion: Use of the interactive voice questionnaire was an enjoyable experience for participants and felt to be easy to use. Despite some specific concerns, Artificial intelligent personalities, like Amazon’s Alexa used in this project, have the potential to be used to collect patient reported outcomes and can be a mechanism for those with certain physical limitations to directly provide their outcome data.
Impact on clinical research: Initial enrollment in clinical trials and maintaining enrollment throughout the trial is a major challenge for bringing new drugs to market. Building tools that enable participants who previously were excluded due to the high burden of effort can be one of the solutions to expand enrollment. Including engaging artificial personalities can be an additional solution to keeping people engaged throughout the trial.
Overcoming challenges: There were a number of concerns raised in the project that need to be addressed, these ranged from general concerns about the implications of artificial intelligence technology, for example ‘Will it always be listening?’, to questions about data privacy that would need to be fully covered in the informed consent process. Issues were also experienced by those with strong accents, which could hinder global roll out in the near term, however as the underlying consumer technology gains traction globally, the associated machine learning will ‘get smarter’ at understanding accents.
Next Steps: This technology is in its infancy and the possibilities to educate, engage and enable participation are now being considered and tested. With results for proof of concepts and pilots showing promise, adding voice technology to a phase four or sub-study is the next step.
