From data to milestones: Predicting milestones to reduce mental load in physical rehabilitation

Details

Supervisors

Pierre-Antoine Absil ; Andries Pauwels

Faculty

Ecole polytechnique de Louvain

Degree label

Master [120] en science des données, orientation technologies de l'information, à finalité spécialisée: analyse de données

Abstract

This thesis presents a predictive model designed to improve recovery after surgery by estimating key milestones for patients undergoing total knee arthroplasty and total hip arthroplasty. The model uses patient provided data to generate personalized recovery ranges, predicting when patients may reach milestones such as walking without crutches, returning to work, stopping medication, resume driving, satisfaction, and feeling pain. Instead of providing a single estimate, the model offers an interval for each milestone, reflecting the inherent uncertainty in the recovery process. This approach helps patients and clinicians gain a more realistic understanding of recovery, reducing stress and mental load, and assisting with decision making. The model uses advanced machine learning techniques, including bootstrapping, to produce reliable prediction intervals for each milestone, aiming for a confidence level of at least 80%. Additionally, it introduces two new milestones: (1) X Years Younger, which calculates improvements in functional age using walking speed derived from steps per minute, and (2) a Complications prediction model that assesses the likelihood of post-surgery complications, offering early risk detection and assisting with decision taking. By transforming patient data into valuable recovery insights, this predictive model clarifies the recovery process, supports better outcomes, and provides a structured, supportive experience for both patients and healthcare providers. The model uses advanced machine learning techniques, including bootstrapping, to produce reliable prediction intervals for each milestone, aiming for a confidence level of at least 80%. Additionally, it introduces two new milestones: (1) X Years Younger, which calculates improvements in functional age using walking speed derived from steps per minute, and (2) a Complications prediction model that assesses the likelihood of post-surgery complications, offering early risk detection and assisting with decision taking. By transforming patient data into valuable recovery insights, this predictive model clarifies the recovery process, supports better outcomes, and provides a structured, supportive experience for both patients and healthcare providers.