Probabilistic approach to longitudinal response prediction: application to radiomics from brain cancer imaging
This work addresses longitudinal response prediction for brain cancer patients using radiomics, offering an incremental improvement by incorporating uncertainty and dimensionality control.
The study tackled the problem of predicting disease progression over time from medical imaging by developing a probabilistic model that integrates baseline and follow-up radiomic features, showing competitive performance against state-of-the-art methods while handling uncertainty and reducing data requirements.
Longitudinal imaging analysis tracks disease progression and treatment response over time, providing dynamic insights into treatment efficacy and disease evolution. Radiomic features extracted from medical imaging can support the study of disease progression and facilitate longitudinal prediction of clinical outcomes. This study presents a probabilistic model for longitudinal response prediction, integrating baseline features with intermediate follow-ups. The probabilistic nature of the model naturally allows to handle the instrinsic uncertainty of the longitudinal prediction of disease progression. We evaluate the proposed model against state-of-the-art disease progression models in both a synthetic scenario and using a brain cancer dataset. Results demonstrate that the approach is competitive against existing methods while uniquely accounting for uncertainty and controlling the growth of problem dimensionality, eliminating the need for data from intermediate follow-ups.