Prediction of Muscle Power in Elderly Using Functional Screening Data

The importance of identifying functional decline in older adults in order to put in place an intervention program has led to the introduction of many functional screening tests. Some of these tests include One Leg Stance (OLS), Timed Up and Go (TUG), grip strength and Sit to Stand (STS). As a part of this work, we propose a linear-regression-based muscle-power estimation model for older people. Our analysis includes carefully choosing the parameters that have potential impact over muscle power and then regress over these parameters to estimate the model weights.
$A$
total of98 participants (24 females, 74 males) aged above 65 years with mean age
$(70.3\pm 5.4)$
years were included for the functional screening test and different questionnaires. Data was collected in Jodhpur, India. We used the Takai et al. (2009) and smith et al. (2010) models to derive estimates of muscle power from the
$STS$
, and then determined the potential impact of the other observed features towards these power estimates. We applied Bootstrap-aggregation and 5 Fold Cross-Validation to validate the model and check its accuracy. The best performance was achieved for the Smith model
$(R^{2}=0.70)$
than the Takai model
$\left(R^2=0.59\right)$
. The proposed model was able to estimate muscle power using the Smith estimation, which could help to identify functional decline, loss of independence and physical frailty in older people. It could also help detect patients suffering from Sarcopenia (loss of skeletal muscle mass).