Associations between music and dance relationships, rhythmic proficiency, and spatiotemporal movement modulation ability in adults with and without mild cognitive impairment

Abstract

Background: Personalized dance-based movement therapies may improve cognitive and motor function in individuals with mild cognitive impairment (MCI), a precursor to Alzheimer's disease. While age- and MCI-related deficits reduce individuals' abilities to perform dance-like rhythmic movement sequences (RMS)-spatial and temporal modifications to movement-it remains unclear how individuals' relationships to dance and music affect their ability to perform RMS.

Objective: Characterize associations between RMS performance and music or dance relationships, as well as the ability to perceive rhythm and meter (rhythmic proficiency) in adults with and without MCI.

Methods: We used wearable inertial sensors to evaluate the ability of 12 young adults (YA; age=23.9±4.2 yrs; 9F), 26 older adults without MCI (OA; age=68.1±8.5 yrs; 16F), and 18 adults with MCI (MCI; age=70.8±6.2 yrs; 10F) to accurately perform spatial, temporal, and spatiotemporal RMS. To quantify self-reported music and dance relationships and rhythmic proficiency, we developed Music (MRQ) and Dance Relationship Questionnaires (DRQ), and a rhythm assessment (RA), respectively. We correlated MRQ, DRQ, and RA scores against RMS performance for each group separately.

Results: The OA and YA groups exhibited better MRQ and RA scores than the MCI group (p<0.006). Better MRQ and RA scores were associated with better temporal RMS performance for only the YA and OA groups (r²=0.18-0.41; p<0.045). DRQ scores were not associated with RMS performance in any group.

Conclusions: Cognitive deficits in adults with MCI likely limit the extent to which music relationships or rhythmic proficiency improve the ability to perform temporal aspects of movements performed during dance-based therapies.

Keywords: Alzheimer’s disease; dance; gait analysis; mild cognitive impairment; music; rehabilitation; rhythm; therapy.

Figure 1:. Spatial modifications and corresponding targets used in spatial and spatiotemporal rhythmic movement sequences (RMS).

A) Spatial modifications. The left two columns correspond to modifications to swing-phase kinematics during movement, while the right two columns correspond to modifications to stance-phase kinematics. The bullets describe joint kinematics defining each spatial target variable for the corresponding modification. Deviations from these target values quantified RMS performance. The colored lines denote the hip (purple), knee (orange), and ankle (red) target values. Adapted, with permission, from Rosenberg et al., 2023 [8].

Figure 2:. Rhythmic stepping sequences used in temporal and spatiotemporal rhythmic movement sequences (RMS).

Each sequence consisted of 2–6 steps, synchronized to either a duple (2-count) or waltz (3-count) meter. Sequences were comprised of very-quick (& = half-beat per step), quick (q = one beat per step), and slow (S = two beats per step) steps. The numbers above each musical note reflect the beat count. A) Simple duple sequences were two-count rhythms with the strong beat on the downbeat and spanned 1–2 measures. B) Complex duple sequences were also two-count rhythms with a weak beat on the downbeat and spanned 2 measures. C) Waltz sequences were three-count rhythms spanning 1–2 measures. Re-used, with permission, from Rosenberg et al., 2023 [8].

Figure 3:. Calculation of error on spatial and temporal modifications comprising RMS.

The left column shows timeseries data. The middle column depicts how percent errors computed within spatial and/ or temporal modifications comprising a given RMS. The right column depicts how percent errors within multiple modifications combine to produce error for each RMS. Dashed lines denote spatial and temporal targets. Yellow dots denote the portion of the stride (e.g., swing vs. stance) where the joint angles were compared to target values. A) One example spatial modification (Attitude). Percent errors of peak joint angles relative to target values were computed for each joint separately (left), then averaged across strides and target variables to compute and overall modification error (middle). The middle column shows stride-averaged data from one participant from each of the YA (orange), OA (gray), and MCI (purple) groups. B) One example temporal modification (bottom; simple duple–quick-quick-Slow-Slow; qqSS). The target rhythm is shown as musical notation below the temporal timeseries. Temporal targets were defined by step sequences, shown by the q-q-S-S notation. Deviations in the timing of kinematic peaks relative to the prescribed temporal pattern constituted error. The middle column shows a histogram of percent errors, which were averaged across strides (dashed line). The bottom middle plot depicts a Fast Fourier Transform of the timeseries data, in which the dominant frequencies (peaks) are compared to target frequencies (dashed lines). Temporal modification performance was defined as the average of these percent errors. Finally, to compute RMS performance error, all spatial and/or temporal modifications constituting an RMS were averaged (within-RMS performance).

Figure 4:. Boxplots representing distributions of Music Relationship Questionnaire (MRQ), Dance Relationship Questionnaire (DRQ), and Rhythm Assessment (RA) scores.

A) MRQ, DRQ, and RA scores for the three participant groups (YA, OA, MCI). Higher MRQ and DRQ scores (max = 7) reflect stronger music and dance relationships, respectively. Higher RA scores (max = 10) reflect greater rhythmic proficiency. B) RMS performance error on each of spatial, temporal, and spatiotemporal RMS for the three participant groups. For both plots, dots represent individual participants. Higher composite scores indicate better performance on the MRQ, DRQ, and RA (upper green arrow). Lower RMS error indicates better performance on RMS (lower green arrow). Each spatial and temporal RMS, error was averaged across 9 RMS from their respective domains and spatiotemporal error was averaged across 4 RMS. For all boxplots, p-values denote significant differences according to independent-samples t-tests (α = 0.05).

Figure 5:. Linear regression testing for within-group associations between RMS performance and each of the MRQ, DRQ, and RA.

Each dot represents a single participant. Lines indicate within-group linear fits. Colors denote the groups (YA: orange, OA: gray, MCI: purple). Regression R-squared values, slopes, and p-values are shown on the corresponding plots for fits that were significantly different from zero (Wald tests; α = 0.05), with colors corresponding to groups. Each of the spatial and temporal RMS percent errors were averaged across the 9 respective RMS and spatiotemporal errors were averaged across 4 RMS. Lower RMS error indicates better performance (green arrow), on average, across RMS within the corresponding RMS class. A) MRQ vs. temporal (top) and spatiotemporal (bottom) RMS performance errors. Higher MRQ scores represent stronger music relationships. B) DRQ vs. spatial (top) and spatiotemporal (bottom) RMS performance errors. Higher DRQ scores represent stronger dance relationships. C) Comparisons of RA with percent errors on temporal (top) and spatiotemporal (bottom) RMS. Higher RA scores imply greater rhythmic proficiency. D) Comparisons of RA subcomponents—auditory rhythm reproduction (max score = 4) and auditory meter recognition (max score = 5)—to temporal (top) and spatiotemporal (bottom) RMS.