JMIR Publications

ABSTRACT

Background:

Studies have shown a relationship between worse glycemic control and lower cognitive scores in youth with Type 1 Diabetes (T1D). However, most studies only assessed long-term glucose control (e.g., years-decades) and cognition at a single time point. Understanding this relationship at a higher temporal resolution (e.g., minutes-hours) and in naturalistic settings has potential clinical implications. Newer technology (e.g., continuous glucose monitoring [CGM] and ecological momentary assessment [EMA]) provides a unique opportunity to explore the specific glucose dynamics that influence dynamic cognition; that is, cognitive functions that fluctuate in the short-term and are influenced by environmental factors.

Objective:

Before we can assess this relationship, we first need to determine the feasibility of measuring cognition in youth in daily life and the practicality of integrating EMAs of cognition and continuously measured glucose—the purpose of this study. Specifically, we assessed the acceptability of measuring dynamic cognition using a smartphone application (app), the youngest age at which it was feasible to use a smartphone app, and adherence to cognitive testing in daily life in youth with and without T1D. Further, we illustrated integration of dynamic cognitive performance scores from the smartphone app and glucose from CGM in youth with T1D.

Methods:

The study was conducted in three phases including one in-lab study and two remote studies. For all phases, youth were asked to complete cognitive tests on the Ambulatory Research in Cognition (ARC) smartphone app that measured processing speed, associative memory, and working memory. For the in-lab study, youth completed app testing four times during their session. For the remote phases, youth were asked to complete cognitive tests five times per day for either 10 or 14 consecutive days in daily life. Youth were asked to rate their impressions of the app. Youth with T1D wore a CGM.

Results:

N=74 youth (N=53 Control; N=21 T1D) aged 4-16 participated. Youth generally reported liking/understanding the ARC app tasks in a lab and remote setting. Age was significantly correlated with task performance such that younger youth performed more poorly; around age 9 is when youth started to perform well. Youth had high mean testing adherence in daily life completing between 76.3%-80.2% of tasks. EMAs of cognition and glucose data were successfully integrated at two distinct temporal resolutions for each participant with T1D—for each day of participation and for each individual cognitive test.

Conclusions:

Data from this pilot and feasibility study support the use of an app to measure dynamic cognitive function in youth ≥9 years old in naturalistic settings and the feasibility of integrating EMAs of cognition and glucose. Next steps include using EMA in a fully powered study to determine the relationship between short-term glycemic control and cognition in youth with T1D.