Session: DAC-04-2: Data-Driven Design

Paper Number: 145053

145053 - A Data-Driven Recommendation Framework for Optimal Medical Walker Designs 

The rapidly advancing fields of statistical modeling and machine learning have significantly enhanced data-driven design and optimization. This work focuses on leveraging these design algorithms to optimize a medical walker, an integral part of gait rehabilitation and physiological therapy of the lower extremities. In particular, we focus on optimizing a 2-wheeled walker. However, our optimization framework can be applied to any assistive mobility device. 

More than 4 million people in the United States use a walker every day. Walkers enhance independent mobility, promote physical activity, elevate emotional well-being, and reduce falls and hospital visits. However, more than 41,000 walker-related injuries occur every year. Therefore, taking steps towards optimizing a walker may have a significant impact on public health and life expectancy, enabling people to work longer and economically prosper on both a micro and macro scale. To our knowledge, there is little data-driven research on optimizing the fundamental aspects of a walker, hence the creation of this paper.

Walkers with minimal weight, maximum stability, and maximum durability are generally desirable. However, excessive weight reductions can result in lower stability and structural integrity. We strive for the optimal compromise between these three conflicting objectives.

To achieve the desirable qualities of a walker, we first create a parametric model of a walker. We generate a dataset of over 5,000 parametric walker designs by uniformly sampling the vast 16-dimensional parametric design space. Each parameter corresponds to a physical defining characteristic of a walker, such as tube diameter, height, material, etc. We perform simulations using Finite Element Analysis (FEA) and calculations on each design to assess mass, structural integrity, and stability. These performance values include displacement vectors for the given load case, stress coefficients, mass, and other physical properties. We also introduce a novel method of systematically calculating the static stability index of a walker. 

Then, we train a predictive Machine-Learning (ML) model utilizing a stacked-ensemble approach shown to outperform traditional models. For a fully defined parametric walker design, our ML model can predict performance objectives and identify trade-offs between them. This ML model aids in transforming the discrete data points (i.e. the 5,000 parametric models and their performance values) to a continuous design space, where every possible valid design vector is mapped to a predicted location in the performance space. 

Lastly, we use Multi-Objective Counterfactuals for Design (MCD), a genetic-based optimization algorithm, to explore the diverse design space and search for high-performing designs by querying the ML performance predictor. The optimizer should not consider geometrically invalid designs, so we enforce strict constraints on the exploration space accordingly. MCD Sampling allows us to reduce the thousands of valid optimized designs to a small set of non-dominated ones.

MCD optimization allows highly customizable constraint enforcement. Users can set constraints on output design performance values such as mass, stability, and structural integrity. This way, the user can prioritize optimizing certain performance aspects over others. Additionally, users can constrain the optimizer’s exploration space. Users can define range constraints on all 16 parameters to generate designs catered to their needs. This allows for custom optimization in almost any scenario.

This work presents potential walker designs that demonstrate up to a 30% mass reduction while increasing structural stability and integrity relative to popular commercially available walkers. We also generate and present designs for various custom optimization scenarios, such as optimization for those of tall stature, pediatric patients, and those requiring additional stability. This work aims to demonstrate the advantages of previously unseen applications of data-driven design on mobility devices and to take a step toward the improved development of assistive mobility devices.

Presenting Author: Advaith Narayanan Leigh High School

Presenting Author Biography: Advaith Narayanan is a junior at Leigh High School, San Jose CA. He is also concurrently taking college level courses at a community College. His research interests lie at the interface of digital computation and engineering (i.e. CAD, simulation, etc). He is a USA Physics Olympiad 2024 Semifinalist. He is a recipient of IEEE Technical Excellence Award and a Recognition Award from The Office of Naval Research, US Navy and Marine Corps at Synopsys Silicon Valley Science and Technology Championship. He is a passionate hackathon enthusiast.

Authors:

Advaith Narayanan Leigh High School