Public articles linked to the same research event.
bioRxiv This work presents the Hierarchical Temporal Transformer (HTT), a two-level architecture in which level 1 encodes each report with BiomedBERT adapted by LoRA and level 2 is a temporal transformer that reads a patient's full report sequence using a continuous-time positional encoding built from the measured number of days between visits plus learnable cancer-type embeddings; on a controlled synthetic corpus of sequential radiology reports it reaches validation AUROC 0.942 versus 0.881 for a single-report baseline and transfers to held-out pancreatic cancer at 0.995 versus 0.949, and on 4,786 real pathology reports from TCGA-Reports spanning 14 cancer types it predicts tumor grade for three types withheld entirely from training with AUROC 1.000 on thyroid carcinoma, 0.960 on sarcoma and 0.
This work presents the Hierarchical Temporal Transformer (HTT), a two-level architecture in which level 1 encodes each report with BiomedBERT adapted by LoRA and level 2 is a temporal transformer that reads a patient's full report sequence using a continuous-time positional encoding built from the measured number of days between visits plus learnable cancer-type embeddings; on a controlled synthetic corpus of sequential radiology reports it reaches validation AUROC 0.942 versus 0.881 for a single-report baseline and transfers to held-out pancreatic cancer at 0.995 versus 0.949, and on 4,786 real pathology reports from TCGA-Reports spanning 14 cancer types it predicts tumor grade for three types withheld entirely from training with AUROC 1.000 on thyroid carcinoma, 0.960 on sarcoma and 0.
This work presents the Hierarchical Temporal Transformer (HTT), a two-level architecture in which level 1 encodes each report with BiomedBERT adapted by LoRA and level 2 is a temporal transformer that reads a patient's full report sequence using a continuous-time positional encoding built from the measured number of days between visits plus learnable cancer-type embeddings; on a controlled synthetic corpus of sequential radiology reports it reaches validation AUROC 0.942 versus 0.881 for a single-report baseline and transfers to held-out pancreatic cancer at 0.995 versus 0.949, and on 4,786 real pathology reports from TCGA-Reports spanning 14 cancer types it predicts tumor grade for three types withheld entirely from training with AUROC 1.000 on thyroid carcinoma, 0.960 on sarcoma and 0.
This work presents the Hierarchical Temporal Transformer (HTT), a two-level architecture in which level 1 encodes each report with BiomedBERT adapted by LoRA and level 2 is a temporal transformer that reads a patient's full report sequence using a continuous-time positional encoding built from the measured number of days between visits plus learnable cancer-type embeddings; on a controlled synthetic corpus of sequential radiology reports it reaches validation AUROC 0.942 versus 0.881 for a single-report baseline and transfers to held-out pancreatic cancer at 0.995 versus 0.949, and on 4,786 real pathology reports from TCGA-Reports spanning 14 cancer types it predicts tumor grade for three types withheld entirely from training with AUROC 1.000 on thyroid carcinoma, 0.960 on sarcoma and 0.