Privacy-Aware AI, Health Report Interpretation, Large Language Models, Medical Summarization, Explainable AI

The rapid growth of electronic health records and diagnostic reports creates both an opportunity and a responsibility: we can make clinical information more accessible, but must also protect patient privacy. This paper presents a privacy-aware AI system that automatically detects and removes personally identifiable information before producing clinically relevant summaries and patient-facing recommendations using a large language model (LLM). Unlike many biomedical summarization approaches that primarily extract findings, our approach combines privacy-preserving preprocessing, structured clinical-value extraction, and user-friendly delivery. The system uses adaptive text processing to balance information retention with privacy protection and a structured extraction module to highlight key laboratory parameters and flag abnormal values. To broaden accessibility, the framework supports multilingual spoken explanations via a voice-enabled interface. A prototype demonstrates concise, comprehensible outputs while minimizing sensitive data exposure, illustrating the practicality of secure, explainable, and patient-centered healthcare AI. [1]–[3]