Reliability of Smartphone-Based Vibration Threshold Measurements

TL;DR

This work addresses measuring vibrotactile perception thresholds ($VPT$) with a smartphone platform to enable accessible, longitudinal tracking of sensory function. It compares a staircase-based method and a decay-based approach against clinical tuning forks (CTF at 128 Hz and RSTF at 64 Hz) across hand and foot locations in healthy adults. Study 1 demonstrates the smartphone staircase yields excellent reliability, comparable to or exceeding RSTF, while the decay method and CTF show more variability; Study 2 shows unguided home-use yields VPTs similar to guided testing, with identifiable but correctable placement errors. The results support potential for at-home monitoring of peripheral neuropathies and inform design refinements for reliable, user-friendly mobile sensory testing tools, with future work including broader populations and clinical validation.

Abstract

Smartphone-based measurement platforms can collect data on human sensory function in an accessible manner. We developed a smartphone app that measures vibration perception thresholds by commanding vibrations with varying amplitudes and recording user responses via (1) a staircase method that adjusts a variable stimulus, and (2) a decay method that measures the time a user feels a decaying stimulus. We conducted two studies with healthy adults to assess the reliability and usability of the app when the smartphone was applied to the hand and foot. The staircase mode had good reliability for repeated measurements, both with and without the support of an in-person experimenter. The app has the potential to be used at home in unguided scenarios.

Figures (5)

• Figure 1: Smartphone setup pertaining to the studies described in Section \ref{['sec:repeated']} and Section \ref{['sec:guidance']} for both the hand (A and C) and foot (B and D). The phone is placed on a pillow, and users wear noise cancelling headphones playing white noise. Detailed placement of the hand (C) and foot (D) placements are also displayed.
• Figure 2: Repeated threshold measurements for each tool on Day 1. Individual participants ($n=10$) are shown in shades of pink, and group level data is shown in black. For the smartphone methods (A and B), mean and standard deviation are shown. For the tuning fork methods (C and D), median and quantiles (25th and 75th) are shown. Unlike the other tools, the smartphone decay method (B) had significant variability between each of the trials on a single day for both the hand and foot positions. Only Day 1 data is plotted in this figure given space limitations and lack of major differences between Days 1 and 2. Statistical results for both Days 1 and 2 are provided in Table \ref{['Table: stats']}. Although we tested 11 participants, we only present data from 10 participants. One participant's results were not used in the analyses as explained in Section \ref{['subsec:study 1 stat methods']}
• Figure 3: Guided vs. unguided VPTs at the hand (left) and foot (right). Experimenter guidance did not significantly impact measured VPTs at the hand (p = 0.327) nor at the foot (p = 0.166). Individual subjects are shown in shades of pink ($n = 15$ for hand, $n = 13$ for foot) and mean and standard deviation are shown in black.
• Figure 4: Flowchart depicting the errors made by participants during the unguided condition. Participant errors predominately consisted of incorrect toe or finger placement, either placing only a portion of the finger or toe on the phone or placing the finger or toe too high on the phone (noted in the darker-colored squares). Some participants may have made multiple mistakes (for instance, having only the toe pad, while also placing it too high on the phone), while some participants may have made mistakes only for a specific trial.
• Figure 5: The two most common errors made by participants in the unguided condition were (a) placing only the finger pad on the phone (instead of the finger pad and distal interphalangeal (DIP) joint) and (b) placing the toe too high on the phone (interphalangeal (IP) joint not aligned with bottom-center edge of phone). As a reminder, the instructed placements are found in Figure \ref{['setup']}.