Towards Transparent Clinical NLP: Lightweight Justification Generation with Privacy Guarantees
Year: 2025
Type: conference-paper
Abstract: In modern healthcare, artificial intelligence has become a key player in improving diagnostic accuracy and clinical efficiency. However, its integration must be responsible—balancing innovation with patients’ right to privacy and transparency. In this work, we introduce a lightweight but secure natural language processing (NLP) framework that not only explains clinical decisions but also ensures the safety of patient data. Built on a distilled transformer architecture and supported by ChatGPT-based rationale generation, our system includes authenticated encryption and fine-tuned models to deliver aspect-level justifications for medical text. We rigorously evaluate our system against privacy frameworks like HIPAA and GDPR and demonstrate its utility on both synthetic and real-world datasets. Through performance analysis and threat modeling, we show how explainability and privacy can coexist in practical clinical settings.
Source: Procedia Computer Science
Language: English
Subfield: Artificial Intelligence
Field: Computer Science
Domain: Physical Sciences
Open Access status: diamond
OpenAlex ID: W4416640551