Technology-based balance performance assessment can eliminate floor and ceiling effects

Many clinical measurement tools for balance have ceiling effects. Technology-based assessments using virtual reality systems such as the Computer-Assisted Rehabilitation Environment (CAREN) may provide a way to develop objective, quantitative measures that scale from low to high levels of difficulty. Our objective was to: (1) develop a performance assessment tool (PAT) for the CAREN; (2) quantify the reliability of the tool; (3) validate the scores against clinical balance measures; and (4) compare the scores from a population with balance impairments to those from able-bodied individuals in a cross-sectional validation study. Three games were developed on the CAREN and tested on 49 participants (36 able-bodied and 13 with impaired mobility). For each module, the corresponding measures were transformed into scores using a series of functions such that ceiling and flooring effects would be minimized. The results showed an association between scores and age, an overlap in scores from impaired high-performance individuals and able-bodied low performance individuals, and a correlation of PAT scores with other clinical tests. Several of the limitations of current clinical tools, including floor and ceiling effects, were overcome by the PAT, suggesting that the PAT can be used to monitor the effect of rehabilitation and training.

3) The Module S3, named "The Bus" (S3 Fig), consists of a game where participants standing inside a virtual bus are required to maintain balance after sudden shifts in the position of the platform.The platform will rapidly shift in one of 8 different directions (N, S, W, E, NW, NE, SW, SE) with different accelerations.Four different levels of challenge were defined by the acceleration of the platform movement, quantified by the time required for the platform to reach a displacement of 20 cm upon delivery of the perturbation.The times associated with each level were of 2400 ms, 1600 ms, 800 ms, and 400 ms.Participants are required to maintain balance during the movement of the bus without using any additional support.The task is repeated for all directions, moving away from the center and then back, for 4 levels of difficulty, each based on the acceleration of the platform described above.

PAT scoring structure
Separate scores were calculated for each leg (when available) within each module based on the module's corresponding outcome measures.For "S1: The Blocks", separate scores were calculated for each foot depending on which foot was required to be lifted during each task.For "S2: The Targets", separate scores were also calculated for each foot, depending on the stepping foot.In addition, separate scores were also calculated for each level of difficulty.For "S3: The Bus", separate scores were calculated for each level.The structure for the definition of all sub scores and their relation to each module as well as to the final score is presented in Supplementary Table S1.S1 Level 2 -S3_2 Level 3 -S3_3 Level 4 -S3_4

Supplementary Table
Conversion functions between outcome measures and task scores were defined separately for each combination of task and level within each game.For these functions, the following rules were set: a) all score values range between 0 and 100, b) a score value of 0 is associated with the lowest possible performance for a task, c) a score value of 100 is associated with the highest possible performance for a task, d) a score value of 70 is associated with the 95th percentile for the corresponding measure, e) the conversion function is defined only for positive values of the measures, f) the function for converting measures into scores is defined by a polynomial, g) the conversion function approaches the score of zero asymptotically.
Given the choice of measures, for all task measures in all three modules a flawless performance is associated with the measure value of zero and a score of 100.Conversely, for the lowest performance measure values, the following definitions were used: a) a score of zero was defined for any value of mvCOP higher than 0.5 m/s, b) a score of zero was defined for any value of peakLift higher than 20 cm, c) a score of zero was defined for a value of ratioStepping of 100%.
The resulting transformation functions were of the form: For "S1: The Blocks", separate measures were obtained for each leg.A single scoring function (Eq. 1) was defined using the best-performing leg, i.e., using the lower of the two values for the measures of mvCOP obtained from each participant.If the obstacles are not cleared then a score of zero will be assigned.
For "S2: The Targets", separate scoring functions were defined for each combination of direction and stepping type (i.e., single foot, double foot), but only the data collected from the lowest level (i.e., level 1) was used in the definition of the scoring function.All but the measures from the diagonal directions (±45˚) were excluded because 1) the direction for 0˚ allowed participants to use either foot, limiting the ability of the measure to characterize the performance for each foot; and 2) the direction of ±90˚ requires a movement that largely limits the movement of the knee; as such, it is not a good measure of general performance.Similar to "S1: The Blocks", separate measures were obtained for each leg, and the final score was calculated using the measure from the best-performing leg, i.e., the lowest of the two values for the measure of peakLift.The final score for "S2: The Targets" was calculated as the average of the scores across levels and stepping types.If a target is not reached, then a score of zero will be assigned.
For "S3: The Bus", a single scoring function was defined for the measures obtained from the lowest level (i.e., level 1).The final score for "S3: The Bus" was calculated as the average of the scores across levels.
The resulting coefficients for each of the scoring function are presented in Table S2.
Finally, to calculate the combined PAT score, we averaged the individual scores from each module.