Title: Deep learning-based sensitive attribute identification and protection using Fernet

Authors: Lakshmi Sri Lasya Tatiraju; M. Sumathi

Addresses: School of Computing, SASTRA (Deemed to be University), Thanjavur, Tamil Nadu, India ' School of Computing, SASTRA (Deemed to be University), Thanjavur, Tamil Nadu, India

Abstract: Patient data, including private information, treatment plans, and disease specifics, is more sensitive in the medical industry. Applying protection to sensitive attributes instead of the entire attribute is more beneficial for authorised users and safeguards sensitive attributes. To achieve this trade-off, this work proposes to identify sensitive attributes. For sensitive attribute identification, machine learning algorithms like random forest, decision tree, and support vector machine are used along with deep learning algorithms like CNN, LSTM, BERT, etc. Results of the deep learning, machine learning, and fuzzy classification models are analysed to find the most effective method. LSTM with GloVe embedding stands out with the highest accuracy at 99.75%, closely followed by CNN at 89.48%, and then BERT performs well with 89% accuracy. The results demonstrated that the LSTM model outperforms the fuzzy and ML models. Subsequently, Fernet encrypts the found data to protect sensitive information.

Keywords: sensitive attributes; medical dataset; fuzzy classification; random forest; decision tree; support vector machine; SVM; convolution neural network; CNN; long short-term memory; LSTM; Fernet encryption.

DOI: 10.1504/IJIIDS.2026.155376

International Journal of Intelligent Information and Database Systems, 2026 Vol.18 No.3/4, pp.308 - 329

Received: 11 May 2024
Accepted: 07 Jan 2025
Published online: 30 Jul 2026 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article