Date of Award


Document Type

Masters Thesis

Degree Name


Organizational Unit

Daniel Felix Ritchie School of Engineering and Computer Science, Computer Science

First Advisor

Matthew J. Rutherford, Ph.D.

Second Advisor

Laleh Mehran, M.F.A.

Third Advisor

Rinku Dewri


Cross validation, Deep learning, Motion quality, Neural network, Quality of motion, Total hip arthroplasty


Artificial Intelligence (AI) is acquiring more recognition than ever by researchers and machine learning practitioners. AI has found significance in many applications like biomedical research for cancer diagnosis using image analysis, pharmaceutical research, and, diagnosis and prognosis of diseases based on knowledge about patients' previous conditions. Due to the increased computational power of modern computers implementing AI, there has been an increase in the feasibility of performing more complex research.

Within the field of orthopedic biomechanics, this research considers complex time-series dataset of the "sit-to-stand" motion of 48 Total Hip Arthroplasty (THA) patients that was collected by the Human Dynamics Laboratory at the University of Denver. The research focuses on predicting the motion quality of the THA patients by analyzing the loads acting on muscles and joints during one motion cycle. We have classified the motion quality into two classes: "Fair" and "Poor", based on muscle forces, and have predicted the motion quality using joint angles.

We address different types of Machine Learning techniques: Artificial Neural Networks (LSTM - long short-term memory, CNN - convolutional neural network, and merged CNN-LSTM) and data science approach (principal component analysis and parallel factor analysis), that utilize remodeled datasets: heatmaps and 3-dimensional vectors. These techniques have been demonstrated efficient for the classification and prediction of the motion quality.

The research proposes time-based optimization by predicting the motion quality at an initial stage of musculoskeletal model simulation, thereby, saving time and efforts required to perform multiple model simulations to generate a complete musculoskeletal modeling dataset. The research has provided efficient techniques for modeling neural networks and predicting post-operative musculoskeletal inference. We observed the accuracy of 83.33% for the prediction of the motion quality under the merged LSTM and CNN network, and autoencoder followed by feedforward neural network. The research work not only helps in realizing AI as an important tool for biomedical research but also introduces various techniques that can be utilized and incorporated by engineers and AI practitioners while working on a multi-variate time-series wide shaped data set with high variance.

Publication Statement

Copyright is held by the author. User is responsible for all copyright compliance.

Rights Holder

Shaswat Sharma


Received from ProQuest

File Format




File Size

157 p.


Artificial intelligence, Computer science, Biomechanics