We present StrengthGaming, which uses scaling and shuffling technique to provide tempo variations to strength training while preserving the training volume and training goals. It affords game designers more flexibility in designing strength training-based exergames.
Training tempo describes the timing ratio of the three distinct phases of an exercise repetition according to muscle activities: 1)concentric: muscle shortening, 2)isometric: maintaining the same muscle length, and 3)eccentric: muscle lengthening (the above figure shows equal distribution for these three phases).
Training guidelines use workload and training volume, which is the total number of performed repetitions and time under tension, to quantify training. Our goal is to maintain training volume and training goals by keeping optimal ratio in training tempo while enabling designers to have tempo variation within a training session (and phases), to reduce repetitiveness and predictability in their game design. The technique we propose is described and shown in figure below:
Wearable inertial measurement unit (IMU) sensors support mobility and can track strength trainingexercises. We used the MYO’s IMU due to its integrated Bluetooth and our prior successful experience using its real-time API. We computed the orientation of the MYO device based on the accelerometer and gyroscope data in quaternion provided by MYO’s Unity SDK. The sensor readings are used to sense players’ progression throughout the bicep curl and are used to control the height of the bird in the game. As shown in figure below, our system gets sensor data, pitch angle of the MYO, over Bluetooth, and our game was developed using Unity 3D to support multiple platforms, including Android, iOS, MacOS, and Windows.
We conducted the experiment with 24 participants under four following conditions:
Tempo accuracy was calculated using the difference between the actual time performed byparticipants and the target time described by training guidelines.
We ran ANOVA test on the accuracy for integration of the two tempo and pair-wise comparison of each conditions. As shown in figure above, there was significant differences between all 3 conditions with visual guidance vs. the condition without any assistance (p < .01), improving accuracy from 64% to 71-74%. However, there was no statistical difference between the 3 conditions with visual guidance. Despite adding gamification and adapting dynamic tempo in our prototype, players wereable to achieve similar accuracy as the visual condition without any gameplay.
Participants’ ratings for entertainment level (1-7) is shown in figure (a) below. The average scores were 2.81 for None condition, 4.08 for Visual Guiance condition, 5.63 for Game - Fixed Tempo condition, and 5.96 for Game - Dynamic Tempo condition. We ran the Friedman test and the Bonferroni post-hoc test for pair-wise comparison, which are the recommended statistical tests for Likert data. The result shows significant difference between all pairs (p<.01). Both gaming conditions were rated significantly more entertaining than the non-gaming conditions, and that dynamic tempo was significantly more entertaining than fixed tempo.
In terms of preference ranking, the distribution of the top ranked choice is shown in figure (b) below. Dynamic tempo was most preferred by the most participants, with 47.9% vs. 37.5% for fixed tempoand 14.6% for visual guidance.