1. Computer Agents And Patient Memory for Medication Information
ABSTRACT
Computer Agents And Patient Memory for Medication Information
To guide development of a computer agent (CA)-based adviser system that presents patient-centered language to older adults (e.g., medication instructions in portal environments or smartphone apps), we evaluated 360 older and younger adults’ responses to medication information delivered by a set of CAs. We assessed patient memory for medication information, their affective responses to the information, and their perception of the CA’s teaching effectiveness and expressiveness. Each participant saw CAs varying in appearance. To investigate the impact of affective cues on patients, we varied CA message framing, with effects described either as gains of taking or losses of not taking the medication. Our results corroborate the idea that CAs can improve older adults' learning in part by engendering social responses.
Renato F. L. Azevedo1, Daniel G. Morrow, PhD1, Kuangxiao Gu1, Thomas S. Huang, PhD1, Mark Hasegawa-Johnson, PhD1, James Graumlich, MD2, Victor Sadauskas4, Tarek Sakakini1, Suma Bhat, PhD1, Ann Willemsen- Dunlap, PhD2,3, Donald J. Halpin3
1University of Illinois at Urbana-Champaign, Champaign, IL; 2University of Illinois College of Medicine at Peoria, Peoria, IL; 3Jump Simulation Center, Peoria, IL; 4Stanford University, Stanford, California. 3
RESULTS 1
Affective Responses by Age and Framing
INTRODUCTION
This prescription is for a medicine called Amiodarone. Amiodarone acts on your heart rhythm. If you take Amiodarone as directed by your doctor, your heart rhythm will be steady and regular. You will also have a heartbeat that is strong and steady. Your heart will pump enough blood so that you feel more energetic and can breathe more easily when you take Amiodarone. [Gain Frame]
Goal: Improve memory for medication information (especially by older adults) by leveraging progress in computer agent (CA) communication (Azevedo et al., 2018).
Reasons for medication nonadherence are manifold, including poor provider communication and inadequate health literacy, especially for older adults (Brown & Bussell, 2011; Fischer et al., 2010). CAs may improve communication by consistently emulating best practices of face-to-face communication with patients (Bickmore et al., 2009; Bickmore et al., 2010; Zhou et al., 2014). For example, nonverbal cues (facial expressions) as well as words convey positive and negative affect associated with benefits of taking (gain frame) or risks of not taking medication (loss frame), respectively. Older adults are especially sensitive to affective aspects of communication (Carstensen et al., 1999).
We investigated the following questions related to the impact of CAs on older and younger adults’ memory for and responses to the CA-based messages:
Gain frames  positive affective response Loss frames  negative affective response
OA more positive than YA (p<.01)
No difference for negative affect (p>.10, ns)
This prescription is for a medicine called Amiodarone. Amiodarone acts on your heart rhythm. If you don’t take Amiodarone as directed by your doctor, your heart rhythm will be unsteady and irregular. You will also have a heartbeat that is weak and unsteady. Your heart may not be able to pump enough blood so that you might feel faint and short of breath if you do not take Amiodarone. [Loss Frame]
Are there age differences in affective response to CA-based communication?
Are there age differences in message memory and CA preferences? Do these effects depend on CA age?
Are gain-framed messages better remembered than loss-framed messages? 2
METHOD
Using the CrazyTalk software (Reallusion, 2016), we developed naturalistic CAs that varied in gender and age) to investigate whether participants prefer CAs that match them on these characteristics (Baylor, 2011).
Gain > Loss messages (p<.001)
OA > YA (p <.01)
CAs rated more positively for gain versus loss messages (p<.001)
Participant Age X CA Age interaction trend (p<.10)
OA but not YA rated older CAs higher (p<.05) 4
Figure 1. Conversational Agents (CAs) of varying age and gender.
CONCLUSIONS
In an online (Mechanical Turk) study, we evaluated older (OA) and younger adults’ (YA) responses to medication information delivered by the CAs.
To investigate the impact of affective cues on patients, we varied CA message framing, with effects described as gains of taking or losses of not taking the medication.
Message memory measured by questions about medication purpose and effects.
Agent teaching effectiveness and expressiveness measured by Agent Persona Inventory (Ryu & Baylor, 2005).
Sample Demographics
Age:
YA (n=240)
m=36.4 years (19-59)
OA (n=120)
m=65.5 years (60-77)
Gender :
63.3% female
Education: YA
m= 15.4 years OA
m= 15.2 years
(t-test; p >.10, ns)
Help to read health materials (measure of health literacy; Chew et al. 2004):
Always = 2.5%
Always = 0%
Often = 5.8%
Often = 3%
Sometimes = 22.9%
Sometimes = 8%
Occasionally = 18.8%
Occasionally = 13%
Never = 50%
Never = 77%
(χ2; p >.05, ns)
Self-reported health: (7 pt scale - very unhealthy/very healthy)
m= 5.00
m= (t-test; p >.10, ns)
Compared to loss-framed messages, gain messages were responded to more positively and better remembered. CAs were even rated more positively when delivering gain messages.
Older adults responded positively to CA-based medication information:
Responded to CAs with more positive affect, consistent with age increase in emotional self-regulation (Carstensen et al., 1999).
Rated CAs more positively, especially the older CAs. This may reflect a 'matching effect', with people preferring more similar CAs (Baylor, 2011).
Better remembered the messages.
Findings add to work of others (e.g., Bickmore et al., 2010; Strassmann & Kramer, 2017; Or et al., 2011) suggesting that CAs can support self-care among older adults.
Funding: This work was supported by the Jump Applied Research for Community Health through Engineering and Simulation (ARCHES) program, UIUC/OSF Hospital, Peoria IL.
Human Factor Lab
Poster