27.06.2024The Krauthammer Group Develops Explainable Deep Learning Model for Predicting Disease Activity in Joint Disorders

See Trottet et al., PLOS Digital Health 2024

A new study by the Krauthammer group presents DAS-Net, an explainable deep learning model for predicting disease activity in chronic inflammatory joint diseases. This model utilizes patient data to make predictions and identifies similar patients based on disease progression. The approach outperforms traditional models and enhances understanding through feature attribution, providing insights into key patient characteristics impacting disease.

Fig 10. t-SNE visualisation of patient representations.

Each point shows the t-SNE embedding of a representation of a patient at a given time. The subplots show the decomposition overlaid with the feature values (restricted to the embeddings with an available value for the feature). Furthermore, we highlighted a patient from the test set (larger filled dot) and her nearest neighbours (triangles) as computed by our algorithm. For each continuous feature we compute the average value in the entire cohort and within the subset of nearest neighbours. For categorical features, we computed the proportion of each category. We overlaid the plots with values representing different patient characteristics; (A) Diagnosis,