Predicting cognitive scores from wearable-based digital physiological features using machine learning: data from a clinical trial in mild cognitive impairment.

BACKGROUND: Continuous assessment and remote monitoring of cognitive function in individuals with mild cognitive impairment (MCI) enables tracking therapeutic effects and modifying treatment to achieve better clinical outcomes. While standardized neuropsychological tests are inconvenient for this purpose, wearable sensor technology collecting physiological and behavioral data looks promising to provide proxy measures of cognitive function. The objective of this study was to evaluate the predictive ability of digital physiological features, based on sensor data from wrist-worn wearables, in determining neuropsychological test scores in individuals with MCI. METHODS: We used the dataset collected from a 10-week single-arm clinical trial in older adults (50-70 years old) diagnosed with amnestic MCI (N = 30) who received a digitally delivered multidomain therapeutic intervention. Cognitive performance was assessed before and after the intervention using the Neuropsychological Test Battery (NTB) from which composite scores were calculated (executive function, processing speed, immediate memory, delayed memory and global cognition). The Empatica E4, a wrist-wearable medical-grade device, was used to collect physiological data including blood volume pulse, electrodermal activity, and skin temperature. We processed sensors' data and extracted a range of physiological features. We used interpolated NTB scores for 10-day intervals to test predictability of scores over short periods and to leverage the maximum of wearable data available. In addition, we used individually centered data which represents deviations from personal baselines. Supervised machine learning was used to train models predicting NTB scores from digital physiological features and demographics. Performance was evaluated using "leave-one-subject-out" and "leave-one-interval-out" cross-validation. RESULTS: The final sample included 96 aggregated data intervals from 17 individuals. In total, 106 digital physiological features were extracted. We found that physiological features, especially measures of heart rate variability, correlated most strongly to the executive function compared to other cognitive composites. The model predicted the actual executive function scores with correlation r = 0.69 and intra-individual changes in executive function scores with r = 0.61. CONCLUSIONS: Our findings demonstrated that wearable-based physiological measures, primarily HRV, have potential to be used for the continuous assessments of cognitive function in individuals with MCI.

Enhancing perceptual and attentional skills requires common demands between the action video games and transfer tasks.

Despite increasing evidence that shows action video game play improves perceptual and cognitive skills, the mechanisms of transfer are not well-understood. In line with previous work, we suggest that transfer is dependent upon common demands between the game and transfer task. In the current study, participants played one of four action games with varying speed, visual, and attentional demands for 20 h. We examined whether training enhanced performance for attentional blink, selective attention, attending to multiple items, visual search and auditory detection. Non-gamers who played the game (Modern Combat) with the highest demands showed transfer to tasks of attentional blink and attending to multiple items. The game (MGS Touch) with fewer attentional demands also decreased attentional blink, but to a lesser degree. Other games failed to show transfer, despite having many action game characteristics but at a reduced intensity. The results support the common demands hypothesis.

Are videogame training gains specific or general?

Many recent studies using healthy adults document enhancements in perception and cognition from playing commercial action videogames (AVGs). Playing action games (e.g., Call of Duty, Medal of Honor) is associated with improved bottom-up lower-level information processing skills like visual-perceptual and attentional processes. One proposal states a general improvement in the ability to interpret and gather statistical information to predict future actions which then leads to better performance across different perceptual/attentional tasks. Another proposal claims all the tasks are separately trained in the AVGs because the AVGs and laboratory tasks contain similar demands. We review studies of action and non-AVGs to show support for the latter proposal. To explain transfer in AVGs, we argue that the perceptual and attention tasks share common demands with the trained videogames (e.g., multiple object tracking (MOT), rapid attentional switches, and peripheral vision). In non-AVGs, several studies also demonstrate specific, limited transfer. One instance of specific transfer is the specific enhancement to mental rotation after training in games with a spatial emphasis (e.g., Tetris). In contrast, the evidence for transfer is equivocal where the game and task do not share common demands (e.g., executive functioning). Thus, the "common demands" hypothesis of transfer not only characterizes transfer effects in AVGs, but also non-action games. Furthermore, such a theory provides specific predictions, which can help in the selection of games to train human cognition as well as in the design of videogames purposed for human cognitive and perceptual enhancement. Finally this hypothesis is consistent with the cognitive training literature where most post-training gains are for tasks similar to the training rather than general, non-specific improvements.

Enhancing cognition with video games: a multiple game training study.

BACKGROUND: Previous evidence points to a causal link between playing action video games and enhanced cognition and perception. However, benefits of playing other video games are under-investigated. We examined whether playing non-action games also improves cognition. Hence, we compared transfer effects of an action and other non-action types that required different cognitive demands. METHODOLOGY/PRINCIPAL FINDINGS: We instructed 5 groups of non-gamer participants to play one game each on a mobile device (iPhone/iPod Touch) for one hour a day/five days a week over four weeks (20 hours). Games included action, spatial memory, match-3, hidden- object, and an agent-based life simulation. Participants performed four behavioral tasks before and after video game training to assess for transfer effects. Tasks included an attentional blink task, a spatial memory and visual search dual task, a visual filter memory task to assess for multiple object tracking and cognitive control, as well as a complex verbal span task. Action game playing eliminated attentional blink and improved cognitive control and multiple-object tracking. Match-3, spatial memory and hidden object games improved visual search performance while the latter two also improved spatial working memory. Complex verbal span improved after match-3 and action game training. CONCLUSION/SIGNIFICANCE: Cognitive improvements were not limited to action game training alone and different games enhanced different aspects of cognition. We conclude that training specific cognitive abilities frequently in a video game improves performance in tasks that share common underlying demands. Overall, these results suggest that many video game-related cognitive improvements may not be due to training of general broad cognitive systems such as executive attentional control, but instead due to frequent utilization of specific cognitive processes during game play. Thus, many video game training related improvements to cognition may be attributed to near-transfer effects.

The effects of acute stress on human prefrontal working memory systems.

We examined the relationship between acute stress and prefrontal-cortex (PFC) based working memory (WM) systems using behavioral (Experiment 1) and functional magnetic resonance imaging (fMRI; Experiment 2) paradigms. Subjects performed a delayed-response item-recognition task, with alternating blocks of high and low WM demand trials. During scanning, participants performed this task under three stress conditions: cold stress (induced by cold-water hand-immersion), a room temperature water control (induced by tepid-water hand-immersion), and no-water control (no hand-immersion). Performance was affected by WM demand, but not stress. Cold stress elicited greater salivary cortisol readings in behavioral subjects, and greater PFC signal change in fMRI subjects, than control conditions. These results suggest that, under stress, increases in PFC activity may be necessary to mediate cognitive processes that maintain behavioral organization.

Visual working memory for global, object, and part-based information.

We investigated visual working memory for novel objects and parts of novel objects. After a delay period, participants showed strikingly more accurate performance recognizing a single whole object than the parts of that object. This bias to remember whole objects, rather than parts, persisted even when the division between parts was clearly defined and the parts were disconnected from each other so that, in order to remember the single whole object, the participants needed to mentally combine the parts. In addition, the bias was confirmed when the parts were divided by color. These experiments indicated that holistic perceptual-grouping biases are automatically used to organize storage in visual working memory. In addition, our results suggested that the bias was impervious to top-down consciously directed control, because when task demands were manipulated through instruction and catch trials, the participants still recognized whole objects more quickly and more accurately than their parts. This bias persisted even when the whole objects were novel and the parts were familiar. We propose that visual working memory representations depend primarily on the global configural properties of whole objects, rather than part-based representations, even when the parts themselves can be clearly perceived as individual objects. This global configural bias beneficially reduces memory load on a capacity-limited system operating in a complex visual environment, because fewer distinct items must be remembered.

Evidence for multiple manipulation processes in prefrontal cortex.

The prefrontal cortex (PFC) is known to subserve working memory (WM) processes. Brain imaging studies of WM using delayed response tasks (DRTs) have shown memory-load-dependent activation increases in dorsal prefrontal cortex (PFC) regions. These activation increases are believed to reflect manipulation of to-be-remembered information in the service of memory-consolidation. This speculation has been based on observations of similar activation increases in tasks that overtly require manipulation by instructing participants to reorder to-be-remembered list items. In this study, we tested the assumption of functional equivalence between these two types of WM tasks. Participants performed a DRT under two conditions with memory loads ranging from 3 to 6 letters. In an "item-order" condition, participants were required to remember letters in the order in which they were presented. In a "reordering" condition, participants were required to remember the letters in alphabetical order. Load-related activation increases were observed during the encoding and maintenance periods of the order maintenance condition, whereas load-related activation decreases were observed in the same periods of the reordering condition. These results suggest that (1) the neural substrates associated with long-list retention and those associated with reordering are not equivalent, (2) cognitive processes associated with long-list retention may be more closely approximated by item-order maintenance than by reordering, and (3) multiple forms of WM manipulation are dissociable on the basis of fMRI data.