Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Smartphone Vibrometric Force Estimation for Grip Related Strength Measurements

Colin Barry, Edward Jay Wang

TL;DR

This work investigates a smartphone-only method to estimate grip-related force using Vibrometric Force Estimation (VFE), leveraging a phone's vibration motor and IMU. By profiling a Google Pixel 4 with synchronized ground-truth force data, the authors train ridge regression models to predict absolute and relative grip force from high-frequency IMU signals. In a 15-fold hold-one-out evaluation, absolute force MAE is 1.88 lbs and relative force MAE is about 10%, demonstrating feasibility of a pinch-style proxy for HGS on a mobile device. The approach enables scalable, low-burden functional health measurements after per-model profiling, with potential clinical and rehabilitation applications while highlighting areas for future refinement and broader smartphone coverage.

Abstract

Hand grip strength is a widely used clinical biomarker linked to mobility, frailty, surgical outcomes, and overall health. This work explores a novel, phone only approach for estimating grip related force using a smartphone's built in vibration motor and inertial measurement unit. When the phone vibrates, applied finger force modulates the amplitude of high frequency accelerometer and gyroscope signals through Vibrometric Force Estimation. We profiled a Google Pixel 4 using synchronized IMU data and ground truth force measurements across varied force trajectories, then trained ridge regression models for both absolute and relative force prediction. In 15 fold hold one out validation, absolute force estimation achieved a mean absolute error of 1.88 lbs, while relative force estimation achieved a mean error of 10.1%. Although the method captures pinch type force rather than standardized full hand HGS, the results demonstrate the feasibility of smartphone based strength assessment using only on device sensors. This approach may enable large scale, low burden functional health measurements once profiling is completed for major smartphone models.

Smartphone Vibrometric Force Estimation for Grip Related Strength Measurements

TL;DR

This work investigates a smartphone-only method to estimate grip-related force using Vibrometric Force Estimation (VFE), leveraging a phone's vibration motor and IMU. By profiling a Google Pixel 4 with synchronized ground-truth force data, the authors train ridge regression models to predict absolute and relative grip force from high-frequency IMU signals. In a 15-fold hold-one-out evaluation, absolute force MAE is 1.88 lbs and relative force MAE is about 10%, demonstrating feasibility of a pinch-style proxy for HGS on a mobile device. The approach enables scalable, low-burden functional health measurements after per-model profiling, with potential clinical and rehabilitation applications while highlighting areas for future refinement and broader smartphone coverage.

Abstract

Hand grip strength is a widely used clinical biomarker linked to mobility, frailty, surgical outcomes, and overall health. This work explores a novel, phone only approach for estimating grip related force using a smartphone's built in vibration motor and inertial measurement unit. When the phone vibrates, applied finger force modulates the amplitude of high frequency accelerometer and gyroscope signals through Vibrometric Force Estimation. We profiled a Google Pixel 4 using synchronized IMU data and ground truth force measurements across varied force trajectories, then trained ridge regression models for both absolute and relative force prediction. In 15 fold hold one out validation, absolute force estimation achieved a mean absolute error of 1.88 lbs, while relative force estimation achieved a mean error of 10.1%. Although the method captures pinch type force rather than standardized full hand HGS, the results demonstrate the feasibility of smartphone based strength assessment using only on device sensors. This approach may enable large scale, low burden functional health measurements once profiling is completed for major smartphone models.

Paper Structure

This paper contains 16 sections, 3 figures.

Table of Contents

  1. Introduction
  2. Related Work
  3. Methods
  4. Force Estimation Profiling
  5. Data Analysis
  6. Signal Processing
  7. Temporal Alignment
  8. Artifact Correction
  9. Feature Engineering
  10. Model Training and Evaluation
  11. Results
  12. Applied Force Measurement Validation
  13. Discussion
  14. Conclusions
  15. Acknowledgements
  16. ...and 1 more sections

Figures (3)

  • Figure 1: Overview of Smartphone Hang Grip Strength Measurement. Part a) A hand squeezing a smartphone while the phone is vibrating. Part b) Smartphone accelerometer signal amplitude during vibration correlates with applied force intensity.
  • Figure 2: Regression plots of model performance for absolute applied force (left) and relative applied force (right).
  • Figure 3: A measurement session sample with representative performance for absolute applied force (top) and relative applied force (bottom) estimation.