Skip to main content
Advertisement
Browse Subject Areas
?

Click through the PLOS taxonomy to find articles in your field.

For more information about PLOS Subject Areas, click here.

< Back to Article

Fig 1.

The structure of the model proposed in this study.

It is an end-to-end pipeline integrating Transformer-based text embedding, multi-task learning, and transfer learning modules.

More »

Fig 1 Expand

Fig 2.

The data preprocessing process in this study (This figure details the specialized clinical text normalization workflow, including medical concept recognition, temporal marker preservation, and ontology-based noise reduction to prepare EHR data for Transformer processing).

More »

Fig 2 Expand

Fig 3.

The principle of Transformer-based text embedding process in this study.

More »

Fig 3 Expand

Fig 4.

The multi-task learning framework in this study.

More »

Fig 4 Expand

Fig 5.

The transfer learning framework in this study (This figure explains the two-phase optimization with clinical domain adapters and confusion loss, showing how pretrained ClinicalBERT weights are adapted to target institutions while mitigating catastrophic forgetting through curriculum learning).

More »

Fig 5 Expand

Table 1.

Logistic regression performance across datasets.

More »

Table 1 Expand

Table 2.

Random forest performance across datasets.

More »

Table 2 Expand

Table 3.

CNN performance across datasets.

More »

Table 3 Expand

Table 4.

Robustness evaluation of our model across datasets.

More »

Table 4 Expand

Fig 6.

Comparative analysis reveals fundamental differences in how each baseline handles dataset variations.

More »

Fig 6 Expand

Table 5.

Logistic regression temporal performance.

More »

Table 5 Expand

Table 6.

Random forest temporal performance.

More »

Table 6 Expand

Table 7.

CNN temporal performance.

More »

Table 7 Expand

Table 8.

Model performance across different temporal periods.

More »

Table 8 Expand

Fig 7.

Comparative temporal analysis reveals fundamental architectural differences in handling concept drift.

More »

Fig 7 Expand

Table 9.

Training vs. validation performance on MIMIC-III dataset.

More »

Table 9 Expand

Table 10.

Ablation study results on MIMIC-III dataset.

More »

Table 10 Expand