Walk test with smartphone sensor can classify fall risk in amputees

In lower-limb amputees, automated foot strikes from a six-minute walk test (6MWT) can be used to calculate step-based features for fall risk classification, according to a study published online Aug. 18 in PLOS Digital Health.

Pascale Juneau, from Ottawa Hospital Research Institute in Ontario, Canada, and colleagues evaluated fall risk classification using the random forest model with a recently developed automated foot strike detection approach. A total of 80 lower-limb amputees (27 fallers and 53 nonfallers) performed a 6MWT with a smartphone at the posterior pelvis. A novel Long Short-Term Memory approach was used to complete automated foot strike detection. Using manually labeled or automated foot strikes, step-based features were calculated.

The researchers found that for 64 of 80 participants, manually labeled foot strikes correctly classified fall risk (accuracy, 80 percent; sensitivity, 55.6 percent; specificity, 92.5 percent). Fifty-eight of 80 participants were correctly classified by automated foot strikes (accuracy, 72.5 percent; sensitivity, 55.6 percent; specificity, 81.1 percent). Equivalent fall classification results were seen with both approaches, but six more false positives were seen with automated foot strikes.

"This study demonstrated that automatically detected foot strikes from a single smartphone sensor location on the body can be used to calculate step-based features for lower limb amputees after completing a 6MWT, leading to preliminary fall risk classification," the authors write.