INS NYC 2024 Program

Poster	Poster Session 04 Program Schedule 02/15/2024 12:00 pm - 01:15 pm Room: Shubert Complex (Posters 1-60) Poster Session 04: Neuroimaging \| Neurostimulation/Neuromodulation \| Teleneuropsychology/Technology Final Abstract #44 Do We Need to Ask? Utility of Ecological Momentary Assessment for Contextualizing Sensor Data in a Smartphone Digital Phenotyping Study. Katherine Hackett, Temple University, Philadelphia, United States Shiyun Xu, University of Pennsylvania, Philadelphia, United States Moira McKniff, Temple University, Philadelphia, United States Sophia Holmqvist, Temple University, Philadelphia, United States Marina Kaplan, Temple University, Philadelphia, United States Lido Paglia, Temple University, Philadelphia, United States Ian Barnett, University of Pennsylvania, Philadelphia, United States Tania Giovannetti, Temple University, Philadelphia, United States Category: Assessment/Psychometrics/Methods (Adult) Keyword 1: aging disorders Keyword 2: ecological validity Keyword 3: technology Objective: Examine whether daily self-reports of day typicality and time spent near smartphone correspond with objective passive digital phenotyping data, and evaluate the impact of using self-report data to inform analysis of passive data (e.g., excluding atypical days). Participants and Methods: 25 participants (M age=70.5+4.6; M education=17.3+1.8; 72% female; 76% non-Hispanic White) with normal cognition or mild cognitive impairment downloaded an open-source smartphone application (mindLAMP) to capture global positioning system (GPS) trajectories for a 4-week study period. Baseline neuropsychological tests and self-report measures of depression, cognitive decline and functional disability were collected as external validators. Participants also completed a nightly Ecological Momentary Assessment (EMA) survey including two 4-point scales to provide context about the prior day’s passive data (1. How typical was your day today on a scale of 1 (very typical) to 4 (very atypical)?; 2. What is the approximate percent of time you spent away from your phone today [response options ranged from 0-25% to 75-100%]). Daily responses were dichotomized to classify days as typical vs. atypical and frequent smartphone use (<50% away from phone) vs. infrequent smartphone use (>50% away from phone). GPS data were processed into monthly average and standard deviation mobility features for each participant according to methods presented previously, and also stratified by EMA data. GPS features for typical vs. atypical and frequent vs. infrequent smartphone use days were compared using paired samples t-tests. Correlation coefficients between GPS features and baseline validators were compared when including versus excluding atypical and infrequent smartphone use days using Fisher’s z-tests. Results: Across the sample, 56% of days were rated as very typical or typical, and 77.5% of days were spent with frequent smartphone use. Four out of 13 GPS features were significantly different on typical vs. atypical days, with more home time, higher circadian routine indices, and less distance travelled from home on days rated as typical (2.5<\|t(21)\|<3.9, p<.05). None of the average GPS features differed on frequent vs. infrequent smartphone use days. When including all days, there were 15 significant correlations between GPS features and external validators showing more activity and more day-to-day variability in GPS were associated with better global cognition, less self-reported functional decline, and less depression at baseline (.40<\|r\|<.51, p<.05). When excluding atypical days, the above correlations were no longer significant; however, the strength of the correlation coefficients was significantly different for only 3 of the 15 relationships. When excluding infrequent smartphone use days, there remained 14 significant correlations with no difference in strength. Conclusions: Including EMA questions about the prior day may offer useful information to contextualize sensor data but may not be critical for validation of the clinical utility of these data. GPS features corresponded to self-reported typicality of day, whereas percent of time spent away from phone did not appear to impact passive GPS data. Excluding days rated as atypical limited the number of statistically significant results in our relatively small sample, but did not impact the strength of most of the associations. Findings should be tested in larger samples with more diverse smartphone use behaviors.

Poster

02/15/2024
12:00 pm - 01:15 pm
Room: Shubert Complex (Posters 1-60)

Poster Session 04: Neuroimaging | Neurostimulation/Neuromodulation | Teleneuropsychology/Technology

Final Abstract #44

Do We Need to Ask? Utility of Ecological Momentary Assessment for Contextualizing Sensor Data in a Smartphone Digital Phenotyping Study.

Katherine Hackett, Temple University, Philadelphia, United States
Shiyun Xu, University of Pennsylvania, Philadelphia, United States
Moira McKniff, Temple University, Philadelphia, United States
Sophia Holmqvist, Temple University, Philadelphia, United States
Marina Kaplan, Temple University, Philadelphia, United States
Lido Paglia, Temple University, Philadelphia, United States
Ian Barnett, University of Pennsylvania, Philadelphia, United States
Tania Giovannetti, Temple University, Philadelphia, United States

Category: Assessment/Psychometrics/Methods (Adult)

Keyword 1: aging disorders
Keyword 2: ecological validity
Keyword 3: technology

Objective:

Examine whether daily self-reports of day typicality and time spent near smartphone correspond with objective passive digital phenotyping data, and evaluate the impact of using self-report data to inform analysis of passive data (e.g., excluding atypical days).

Participants and Methods:

25 participants (M age=70.5+4.6; M education=17.3+1.8; 72% female; 76% non-Hispanic White) with normal cognition or mild cognitive impairment downloaded an open-source smartphone application (mindLAMP) to capture global positioning system (GPS) trajectories for a 4-week study period. Baseline neuropsychological tests and self-report measures of depression, cognitive decline and functional disability were collected as external validators. Participants also completed a nightly Ecological Momentary Assessment (EMA) survey including two 4-point scales to provide context about the prior day’s passive data (1. How typical was your day today on a scale of 1 (very typical) to 4 (very atypical)?; 2. What is the approximate percent of time you spent away from your phone today [response options ranged from 0-25% to 75-100%]). Daily responses were dichotomized to classify days as typical vs. atypical and frequent smartphone use (<50% away from phone) vs. infrequent smartphone use (>50% away from phone). GPS data were processed into monthly average and standard deviation mobility features for each participant according to methods presented previously, and also stratified by EMA data. GPS features for typical vs. atypical and frequent vs. infrequent smartphone use days were compared using paired samples t-tests. Correlation coefficients between GPS features and baseline validators were compared when including versus excluding atypical and infrequent smartphone use days using Fisher’s z-tests.

Results:

Across the sample, 56% of days were rated as very typical or typical, and 77.5% of days were spent with frequent smartphone use. Four out of 13 GPS features were significantly different on typical vs. atypical days, with more home time, higher circadian routine indices, and less distance travelled from home on days rated as typical (2.5<|t(21)|<3.9, p<.05). None of the average GPS features differed on frequent vs. infrequent smartphone use days. When including all days, there were 15 significant correlations between GPS features and external validators showing more activity and more day-to-day variability in GPS were associated with better global cognition, less self-reported functional decline, and less depression at baseline (.40<|r|<.51, p<.05). When excluding atypical days, the above correlations were no longer significant; however, the strength of the correlation coefficients was significantly different for only 3 of the 15 relationships. When excluding infrequent smartphone use days, there remained 14 significant correlations with no difference in strength.

Conclusions:

Including EMA questions about the prior day may offer useful information to contextualize sensor data but may not be critical for validation of the clinical utility of these data. GPS features corresponded to self-reported typicality of day, whereas percent of time spent away from phone did not appear to impact passive GPS data. Excluding days rated as atypical limited the number of statistically significant results in our relatively small sample, but did not impact the strength of most of the associations. Findings should be tested in larger samples with more diverse smartphone use behaviors.