In radiology report there is one section named as Impression. In this section, the radiologist summarizes the findings. But the process of writing the impression statements is time-consuming and highly repetitive. Model is trying to automate the generation of radiology impressions using neural sequence-to-sequence learning and the pointer generator model. It is text summarization problem, where the source sequence is the radiology findings and the target sequence the impression statement. Existing models neural sequence-to-sequence learning and the pointer generator model fails to encorporate background text along with findings. This model is build on top of this models to make use of background text as well.A board-certified radiologist indicated that 67% of sampled system summaries are at least as good as the reference summaries written by well-trained radiologists.
Input variables : Radiology Report findings, Clinical text
Output Variables : Impression statements
| Statistical | : | Somers D | Accuracy | Precision and Recall | Confusion Matrix | F1 Score | Roc and Auc | Prevalence | Detection Rate | Balanced Accuracy | Cohen's Kappa | Concordance | Gini Coefficent | KS Statistic | Youden's J Index |
| Business | : | Bed Occupancy Rate | Medical Equipment Utilization | Optimized Hospital Resource Utilization |
| Infrastructure | : | Log Bytes | Logging/User/IAMPolicy | Logging/User/VPN | CPU Utilization | Memory Usage | Error Count | Prediction Count | Prediction Latencies | Private Endpoint Prediction Latencies | Private Endpoint Response Count |
Visit Model : github.com
Additional links : nlp.stanford.edu
| Model Category | : | Public |
| Date Published | : | February, 2019 |
| Healthcare Domain | : | Provider |
| Code | : | github.com |
| Utilization Management |
| Hospital Resource Utilization |