Improving Everyday Task Performance Through Repeated Practice in Virtual Reality. (VKI)

There are very few effective interventions that promote functional independence in people with Alzheimer's disease (AD) and related dementias. This R21 project is the first step in the long-term goal of developing an effective, enjoyable, portable, and inexpensive non-immersive virtual reality (VR) training intervention for improving the performance of everyday tasks. The investigators' VR training approach is built upon the results of past studies that show 1) when people with AD repeatedly practice daily tasks they subsequently perform them more completely and without error; and 2) healthy people are able to transfer skills learned in VR-contexts to tasks in the real world. This R21 study will obtain preliminary data to inform a future randomized clinical trial through three aims: Aim 1) To test the hypothesis that individuals with mild-moderate AD will show improved performance on an everyday task after repeatedly practicing the task in a non-immersive VR setting; Aim 2) To explore usability and acceptability of the VR training as well as associations between individual differences variables (e.g., cognitive abilities, demographics) and training effects. To test Aim 1, 40 participants with mild to moderate AD will be recruited to complete daily VR Training sessions for one week. VR Training will include repeated practice of a single, everyday task in a non-immersive VR-context (VR Breakfast or VR Lunch; counterbalanced across participants). The primary outcome measure is performance of the real-life version of the trained task, which will be collected before and at two time points after training, compared to performance of an untrained, control task of comparable difficulty, and scored from video by coders blinded to training task/condition. To evaluate Aim 2, all participants and an informant will complete interviews and questionnaires and participants will complete tests of cognitive abilities. Usability and acceptability of the VR training will be evaluated and associations between participant variables and VR Training results will be explored. If the proposed hypothesis is supported and results show that training effects generalize from virtual to real tasks in the study sample, then VR training of custom and individualized tasks will be investigated in a future randomized, controlled clinical trial for maintaining and improving functional abilities in people with mild to moderate AD.

Functional disability, the loss of independence in everyday tasks like meal preparation, is a core diagnostic criterion for clinical dementia/Alzheimer's disease (AD) and contributes to the disease's exorbitant costs and caregiver burden. There is currently no cure for AD and available interventions have had limited success at reducing functional disability. Cognitive training leads to reliable improvements on the specific task(s) that are repeatedly practiced, but transfer to untrained, functional tasks is limited. Because of both limited transfer and rigid (procedural) learning in AD, novel language-based interventions for dementia have focused on repeated practice with a circumscribed and highly personalized vocabulary to maintain functional communication. Repeated practice of everyday tasks (leveraging intact procedural memory also has been shown to improve performance of trained tasks in people with AD but is not a feasible intervention, because the effort required to set up tasks, monitor performance, and provide feedback/supervision is far greater than having a caregiver simply complete the task for the patient. However, if repeated practice of specific everyday tasks could be implemented in a more cost-effective, efficient manner, then this straightforward intervention strategy could reduce functional disability and associated costs of care and caregiver burden. This R21 proposal will investigate the efficacy of a low-cost and portable computer training program that uses non-immersive virtual reality (VR) to enable participants with AD to independently practice meaningful everyday activities (e.g., meal preparation). The goal of this R21 study is to evaluate the extent to which training in a virtual context will generalize to real life tasks. If training effects generalize from virtual to real settings, then training of customized tasks for maintaining and improving functional abilities in older adults with AD will be investigated in a future randomized, controlled clinical trial. This proposal was informed by a cognitive activation model on which everyday activities are represented as schema hierarchies. Activation within the schema hierarchy may spread automatically from higher level schema (e.g., make lunch) to lower level schema (e.g., spread peanut butter) as well as from objects or environments that may "trigger" associated schema (e.g., a coffee mug may activate coffee-making schema). Deliberate cognitive control is essential to modulate activations, enable smooth transitions between schema subtasks, and inhibit inappropriate activation from objects in the environment. Without control mechanisms, interference from competing schema or objects in the environment may derail performance and lead to inefficiency and disorganization of task steps (i.e., commission errors like mis-sequencing, incorrect object selection, etc.). Another mechanism for performance difficulties involves premature decay of schema activation or degraded task schema/knowledge (i.e., omission errors). On this model, repeated practice serves to strengthen associations within a schema hierarchy and between task schema and target objects in the environment, reducing the burden on control mechanisms and the likelihood of premature decay. To evaluate the efficacy of VR Training the investigators will recruit 40 participants with AD to complete daily training sessions for one week. VR Training will include repeated practice of a single, everyday task in a non-immersive VR context (e.g., VR Breakfast) using a laptop with a touch-screen interface. Performance of a real-life version of the trained task (e.g., Real Breakfast) and a matched, untrained/control, real task (e.g., Real Lunch) before and at two time points post-VR Training will serve as the primary outcome measures. All procedures are designed to achieve the following aims: Aim 1: To test the hypothesis that individuals with mild-moderate AD will show improved performance on an everyday task after repeatedly practicing the task in VR. When people with AD repeatedly practice tasks in real life, they show improved performance of those tasks. Transfer from VR training to real life outcomes has not been systematically studied in AD, but healthy adults, a case report of a woman with moderate AD, and pilot data from the investigators' laboratory support transfer from VR to real tasks. Thus, the investigators predict that participants will show significantly higher task accomplishment, fewer errors and faster completion times from Baseline to Post-Testing on real trained tasks versus real untrained tasks. Aim 2 (Exploratory): To explore usability and acceptability of the VR Training and associations between individual differences variables (e.g., cognitive abilities) and training effects. These analyses will elucidate mechanisms and identify individuals most likely to benefit from VR Training. Results will demonstrate the efficacy and feasibility of VR Training and identify the patient characteristics most strongly associated with training benefits on real life tasks. Data will be used to inform a future randomized clinical trial using VR tasks that are customized to participants' daily routines.

Alzheimer Disease, Mild Dementia, Mild Cognitive Impairment, Dementia, Dementia, Vascular, Dementia Alzheimers, Dementia, Mild, Dementia, Mixed, Dementia of Alzheimer Type

The intervention is task specific. Participants are given an intervention for a specific task, and outcomes include the specific task for which they received the intervention (i.e., "trained task), as well as a control task for which they do not receive the intervention.

This is a within-participant study with two arms. All participants receive the Virtual Training with one target task and they are tested (after one day and one month) on the target task (arm 1) and as well as on a control task(arm 2). Assignment to the specific task that is trained is randomized across participants.

This is a within-participant study with two arms. All participants receive the Virtual Training with one target task and they are tested (after one day and one month) on the target task (arm 1) and as well as on a control task(arm 2). Assignment to the specific task that is trained is randomized across participants.

VR Training will include repeated practice of a single, everyday task in a non-immersive VR context (e.g., VR Breakfast) using a laptop with a touch-screen interface for four days over the course of one week.

The number of task steps accomplished out of 16 possible steps; also may be reported as the number of task steps omitted (omission errors)

The number of task steps accomplished out of 16 possible steps; also may be reported as the number of task steps omitted (omission errors)

The number total overt commission errors; may be characterized as a raw total or as an error rate measure (by dividing by total number of steps completed) to control for variability in the number of steps accomplished.

The number total overt commission errors; may be characterized as a raw total or as an error rate measure (by dividing by total number of steps completed) to control for variability in the number of steps accomplished.

The number total subtle errors coded; may be characterized as a raw total or as a micro-error rate measure (by dividing by total number of steps completed) to control for variability in the number of steps accomplished.

The number total subtle errors coded; may be characterized as a raw total or as a micro-error rate measure (by dividing by total number of steps completed) to control for variability in the number of steps accomplished.

The total time (in seconds) required to complete the task; may be characterized as a raw time or as a rate of completion measure (by dividing by total time by number of steps completed) to control for variability in the number of steps accomplished.

The total time (in seconds) required to complete the task; may be characterized as a raw time or as a rate of completion measure (by dividing by total time by number of steps completed) to control for variability in the number of steps accomplished.

Participant is asked open-ended questions about the burden the Virtual Training had on their daily life and how likely they would be to adopt the virtual training in their day-to-day life.

The System Usability Scale includes 10 items evaluated on a 1 (strongly disagree) - 5 (strongly agree) point scale. After correction (-1 for odd items and -5 for even, reverse-scored items), the total is multiplied by 2.5 to yield a total score ranging from 0 (poor usability) to 100 (excellent usability). The investigators also added a single seven-point adjective rating question regarding the usability of the Training (1 = worst imaginable to 7 = best imaginable) at the end of the questionnaire to facilitate interpretation.

Inclusion/Exclusion Criteria: 65 years or older; fluency in the English language; availability of an informant reporter who has knowledge of the participant's daily functioning; no lifetime history of severe psychiatric disorder (e.g., schizophrenia, bipolar disorder), nervous system infections or disorders (e.g., epilepsy, brain tumor, large-vessel stroke, major head trauma) other than Alzheimer's disease; no current metabolic or systemic disorders (e.g., B12 deficiency, renal failure, cancer); no current major depression or moderate-severe depression symptoms; no current moderate - severe, uncontrolled anxiety symptoms; no severe sensory deficits that would preclude visual detection or identification of common everyday objects used in the study or the inability to hear the task directions (e.g., blindness, total hearing loss); no severe motor weakness that would preclude the use of everyday objects or the VR Training computer touch screen (e.g., severe deformities or paralysis of both upper extremities) ; intact estimated general intellectual functioning (i.e., no history of intellectual disability); available to participate in the one-month follow-up session after the VR Training (i.e., no surgery, travel, etc. scheduled over the next month); diagnosis of mild to moderate Alzheimer's disease within the past year, including confirmation of mild to moderate dementia on Mini Mental Status Exam (score approximately 25 or lower), significant functional difficulties reported by informant report, and cognitive impairment on demographically adjusted (age, education, sex, and race) cognitive test scores at baseline. An informant (N = 40) also will be recruited for each participant with dementia. Informants are people who know the participant well and interact with the participant on a daily basis. Informants will be asked to report on the participants' daily functioning and the extent to which the informant is burdened by the participants. Informants also will be asked to report on changes in medical or mental status during the study period. Informant eligibility criteria is listed below: 18 years of age or older fluency in the English language available and willing to complete study questionnaires in person or by phone has daily contact with the participant reports that he/she is knowledgeable of the participant's daily functioning

The investigators will present preliminary and final data at scientific conferences and lectures at research centers, clinical settings, and senior centers. The study results will be publish. The Virtual Kitchen that is used for the training in this proposal is freely available to scientists who are interested in using it. Currently, investigators simply need to contact members of the study team to attain the program. The investigators will make the final training version of the Virtual Kitchen available for download on a website after procedures are fully manualized. After data collection is complete and the results of the aims are published, then the investigators intend to make the entire de-identified data set available following guidelines available through the Center for Open Science.