Integrating clinical notes for enhanced mortality prediction in ICU
Predicting patient outcomes in Intensive Care Units (ICUs) is a critical task for im- proving patient management and treatment planning. Traditional predictive models primarily rely on structured time-series data (e.g., vital signs, lab results) to forecast outcomes like in-hospital mortality. Ho...
| Main Author: | |
|---|---|
| Other Authors: | |
| Format: | Final Year Project (FYP) |
| Language: | English |
| Published: |
Nanyang Technological University
2024
|
| Subjects: | |
| Online Access: | https://hdl.handle.net/10356/181084 |
| _version_ | 1826111558495764480 |
|---|---|
| author | Pang, Kelvin |
| author2 | Fan Xiuyi |
| author_facet | Fan Xiuyi Pang, Kelvin |
| author_sort | Pang, Kelvin |
| collection | NTU |
| description | Predicting patient outcomes in Intensive Care Units (ICUs) is a critical task for im-
proving patient management and treatment planning. Traditional predictive models
primarily rely on structured time-series data (e.g., vital signs, lab results) to forecast
outcomes like in-hospital mortality. However, these models often overlook the rich, un-
structured information available in clinical notes, which can provide important context
about a patient’s condition that is not reflected in structured data alone.
This study explores the viability of integrating unstructured clinical text data into mor-
tality prediction models using the STraTS (Self-Supervised Transformer for Sparse
and Irregularly Sampled Multivariate Clinical Time-Series) model. The primary ob-
jective is to assess whether text embeddings derived from clinical notes—processed
using Clinical Longformer and BioWordVec—can enhance predictive performance
in comparison to the baseline STraTS model, which uses structured time-series and
demographic data.
The predictive performance of three model variants, the baseline STraTS model, Clin-
ical Longformer, and BioWordVec models, will be evaluated. Models were evaluated
using ROC-AUC, PR-AUC, and (min(Re,Pr)). While the baseline model outperformed
both Clinical Longformer and BioWordVec, the Clinical Longformer demonstrated
potential in optimizing the precision-recall trade-off, a critical factor in mortality pre-
diction tasks involving imbalanced datasets.
The findings suggest that unstructured text data offers complementary value, while
structured data remains a more reliable predictor of mortality. |
| first_indexed | 2025-03-09T10:21:32Z |
| format | Final Year Project (FYP) |
| id | ntu-10356/181084 |
| institution | Nanyang Technological University |
| language | English |
| last_indexed | 2025-03-09T10:21:32Z |
| publishDate | 2024 |
| publisher | Nanyang Technological University |
| record_format | dspace |
| spelling | ntu-10356/1810842024-11-13T22:56:33Z Integrating clinical notes for enhanced mortality prediction in ICU Pang, Kelvin Fan Xiuyi Liu Siyuan College of Computing and Data Science xyfan@ntu.edu.sg, SYLiu@ntu.edu.sg Computer and Information Science Predicting patient outcomes in Intensive Care Units (ICUs) is a critical task for im- proving patient management and treatment planning. Traditional predictive models primarily rely on structured time-series data (e.g., vital signs, lab results) to forecast outcomes like in-hospital mortality. However, these models often overlook the rich, un- structured information available in clinical notes, which can provide important context about a patient’s condition that is not reflected in structured data alone. This study explores the viability of integrating unstructured clinical text data into mor- tality prediction models using the STraTS (Self-Supervised Transformer for Sparse and Irregularly Sampled Multivariate Clinical Time-Series) model. The primary ob- jective is to assess whether text embeddings derived from clinical notes—processed using Clinical Longformer and BioWordVec—can enhance predictive performance in comparison to the baseline STraTS model, which uses structured time-series and demographic data. The predictive performance of three model variants, the baseline STraTS model, Clin- ical Longformer, and BioWordVec models, will be evaluated. Models were evaluated using ROC-AUC, PR-AUC, and (min(Re,Pr)). While the baseline model outperformed both Clinical Longformer and BioWordVec, the Clinical Longformer demonstrated potential in optimizing the precision-recall trade-off, a critical factor in mortality pre- diction tasks involving imbalanced datasets. The findings suggest that unstructured text data offers complementary value, while structured data remains a more reliable predictor of mortality. Bachelor's degree 2024-11-13T22:56:33Z 2024-11-13T22:56:33Z 2024 Final Year Project (FYP) Pang, K. (2024). Integrating clinical notes for enhanced mortality prediction in ICU. Final Year Project (FYP), Nanyang Technological University, Singapore. https://hdl.handle.net/10356/181084 https://hdl.handle.net/10356/181084 en application/pdf Nanyang Technological University |
| spellingShingle | Computer and Information Science Pang, Kelvin Integrating clinical notes for enhanced mortality prediction in ICU |
| title | Integrating clinical notes for enhanced mortality prediction in ICU |
| title_full | Integrating clinical notes for enhanced mortality prediction in ICU |
| title_fullStr | Integrating clinical notes for enhanced mortality prediction in ICU |
| title_full_unstemmed | Integrating clinical notes for enhanced mortality prediction in ICU |
| title_short | Integrating clinical notes for enhanced mortality prediction in ICU |
| title_sort | integrating clinical notes for enhanced mortality prediction in icu |
| topic | Computer and Information Science |
| url | https://hdl.handle.net/10356/181084 |
| work_keys_str_mv | AT pangkelvin integratingclinicalnotesforenhancedmortalitypredictioninicu |