Privacy safe text tools detect health misinformation with strong accuracy
Privacy-Preserving Text Inference for Health-Misinformation Classification and Semantic Retrieval
Cryptography and SecurityMachine Learning
Summary
Health misinformation can be harmful, but checking user text for wrong info risks exposing private medical details. The authors created a way to analyze text securely using special encryption so user data stays hidden during processing. Their method combines several advanced language tools and kept accuracy the same whether encryption was on or off. They also made searching similar sentences faster and less data-heavy while keeping it private. This shows it’s possible to fight health misinformation without revealing sensitive user information.
What this means in practice
- •For healthcare data teams: Protect patient privacy while analyzing medical text for misinformation using encrypted AI models hosted on servers.
- •For document search developers: Build privacy-preserving semantic search that returns relevant health-related texts without exposing user queries or indexed data.
Authors
Jasleen Kaur, Chloé Cachet, Koray Karabina, Jean Belo Klamti, Philippe Lamontagne
Abstract
Health-misinformation analysis presents a dual challenge: misleading claims can affect public well-being, while systems designed to detect them may expose sensitive user-submitted text. This motivates privacy-preserving inference for public-health applications in which user queries may contain confidential medical information and trained models and reference databases are maintained on organizational servers. We design a privacy-preserving text inference framework based on CKKS homomorphic encryption for health-misinformation classification and semantic retrieval. The classification pipeline fuses BioBERT embeddings, PCA-reduced MPNet embeddings, and SVD-reduced TF-IDF features, while the retrieval pipeline combines MPNet embeddings with ISCC-SCT semantic fingerprints through late fusion. We evaluate predictive performance and benchmark plaintext versus ciphertext-based execution. The hybrid classifier achieved an accuracy of 0.971 and a macro-F1 score of 0.950, with identical reported metrics under plaintext and ciphertext-based inference. The hybrid late-fusion similarity model achieved a Pearson correlation of 0.8691 with human sentence-pair similarity scores on STS-B. For ciphertext-based retrieval, we evaluated naive and optimized implementations. The optimized implementation uses CKKS single instruction, multiple data (SIMD) packing to compute and return multiple similarity scores together, reducing single-query latency over 7,128 indexed items from 222.028 to 5.168 seconds and communication overhead from 6.11 GB to 2.13 MB. These findings demonstrate the technical feasibility of privacy-preserving text inference for health-misinformation applications and characterize its predictive utility and efficiency trade-offs.