Subject

Determining contact forces in the knee joint during walking is relevant for the study of knee pathology and rehabilitation thereof. In particular, the knee joint loading, which the force at the joint over a period of time, is relevant. However, it is obviously completely impossible to directly measure such forces in the knee in vivo in a walking human. There exist however advanced musculoskeletal simulation models in biomechanics that try to predict muscle forces from the kinematic data and ground reaction force data captured in the gait lab. Such models can be validated on a few available datasets from subjects with a knee prosthesis in which a force sensor is embedded [Fregly et al, 2012].
The big downside of these models is that they only allow the knee joint loading to be estimated in the gait lab, as motion capture data as well as ground reaction force data is required. This makes that it becomes hard to estimate knee joint loading in the natural environment of the patient and it makes it a cumbersome and time-consuming procedure, as motion capture requires the accurate placement of a set of infrared reflective markers on a set of anatomical reference points on the body.
There is however limited work that tries to estimate knee joint loading from wearable sensors. Despite these sensors being completely external to the body, there are initial interesting results in the literature. In particular IMU sensors and sEMG sensors are being used for this goal. With the adoption of machine learning and deep-learning methods in biomechanics, together with reference datasets becoming publicly available, it now becomes feasible to train a machine learning method to predict the joint load based on the sEMG or IMU sensor data. It however remains a very challenging problem to train such a classifier as the sensor data is extremely subject dependent.
Your task is to replicate and validate an existing method for this task. You will investigate recent state-of-the-art methods for preprocessing the sensor data to remove subject variability, such that patient specific models can be learnt, while still benefiting from additional training data from extra subjects.

Framework of the Thesis

Fregly, B.J., Besier, T.F., Lloyd, D.G., Delp, S.L., Banks, S.A., Pandy, M.G., and D'Lima, D.D. (2012) Grand challenge competition to predict in vivo knee loads. Journal of Orthopaedic Research 30, 503-513.
Rane, L., Ding, Z., McGregor, A.H. et al. Deep Learning for Musculoskeletal Force Prediction. Ann Biomed Eng 47, 778–789 (2019). https://doi.org/10.1007/s10439-018-02190-0