Related Experiment Videos
Structured-to-text ClinicalBERT embeddings with random Forest for heart disease prediction: a proof-of-concept study
1School of Advanced Sciences, Vellore Institute of Technology (VIT), Vellore, India.
Insights
This study introduces a novel ClinicalBERT framework to improve heart disease prediction by converting structured data into text. The model achieved 95.6% accuracy, offering enhanced cardiovascular risk assessment.
Area of Science:
- Medical Informatics
- Artificial Intelligence in Healthcare
- Cardiovascular Disease Research
Background:
- Heart disease is a leading global cause of mortality, necessitating advanced prediction systems.
- Traditional machine learning models for cardiovascular disease prediction often overlook contextual patient feature relationships.
Purpose of the Study:
- To develop and evaluate a structured-to-text ClinicalBERT framework for enhanced heart disease risk prediction.
- To leverage transformer-based embeddings for improved cardiovascular risk assessment.
Main Methods:
- Structured cardiovascular data from the UCI Statlog/Kaggle dataset were transformed into clinical text descriptions.
- ClinicalBERT processed text representations to generate contextual embeddings.
- A Random Forest classifier utilized these embeddings for heart disease prediction, with performance assessed via standard metrics.
Main Results:
- The ClinicalBERT + Random Forest framework achieved high performance: 95.6% accuracy, 88.89% precision, 95.30% recall, 91.30% F1-score, and 0.71 ROC-AUC.
- Contextual embeddings from ClinicalBERT demonstrated superior feature representation compared to conventional machine learning models.
- The framework showed feasibility in adapting ClinicalBERT to structured cardiovascular data.
Conclusions:
- The proposed ClinicalBERT framework offers a promising approach for accurate cardiovascular risk prediction.
- While a proof-of-concept, the study highlights the potential of text-based AI in analyzing structured clinical data.
- Future research should focus on larger datasets and external validation for broader clinical applicability.
Abstract:
Heart disease remains one of the leading causes of mortality worldwide, highlighting the need for accurate and early risk prediction systems. Traditional machine learning approaches for cardiovascular disease prediction primarily rely on structured clinical attributes and may not fully capture contextual relationships among patient features. To address this limitation, this study proposes a structured-to-text ClinicalBERT framework that transforms structured cardiovascular records into contextual clinical text representations and utilizes transformer-based embeddings for heart disease prediction. The study employs a publicly available UCI Statlog/Kaggle heart disease dataset containing 270 complete patient records. Structured cardiovascular attributes, including age, sex, chest pain type, blood pressure, cholesterol level, electrocardiogram results, and heart rate measurements, are converted into clinically meaningful textual descriptions. These text representations are processed using ClinicalBERT to generate contextual embeddings, which are subsequently used as input features for a Random Forest classifier. Model performance was evaluated using an 80:20 train-test split and assessed through Accuracy, Precision, Recall, F1-score, and ROC-AUC metrics. Experimental results demonstrate that the proposed ClinicalBERT + Random Forest framework achieved an accuracy of 95.6%, precision of 88.89%, recall of 95.30%, F1-score of 91.30%, and a ROC-AUC of 0.71 on the held-out test set. Comparative analysis with conventional machine learning models indicates that contextual embeddings generated by ClinicalBERT provide improved feature representation for cardiovascular risk prediction. The findings demonstrate the feasibility of adapting ClinicalBERT to structured cardiovascular data through contextual text generation. Although the proposed framework shows promising predictive performance, the study should be considered a proof-of-concept due to the limited dataset size and absence of external validation. Future work will focus on multicenter evaluation, explainable AI techniques, and broader clinical validation to enhance generalizability and real-world applicability.