Application of Machine Learning-NLP Approach with Fully Homomorphic Encryption Techniques in Medical PII Data¶
Authors: Uchenna Ndolo; Hoda El-Sayed; Md Kamruzzaman Sarker
Year: 2025
Venue: 2025 6th International Conference on Artificial Intelligence, Robotics and Control (AIRC)
type: application collection: ppml status: auto
Links: DOI
Summary¶
Auto-extracted summary (unreviewed)
Applies fully homomorphic encryption to classical ML classifiers (SGD classifier, XGBoost, decision tree) for classifying medical dialogues that may contain personally identifiable information, and benchmarks FHE inference against plaintext models. FHE models match plaintext accuracy (100% for SGD and XGBoost, 91.7% for the decision tree) but incur large increases in training and especially prediction time (e.g. 20,460 s for the FHE decision tree), highlighting the privacy/latency trade-off.
Abstract¶
Integrating machine learning (ML) into healthcare systems has revolutionized diagnostic capabilities and personalized treatment plans. However, using sensitive medical data introduces significant privacy concerns, necessitating secure computational frameworks. Privacy-preserving machine learning (PPML) addresses these challenges by enabling data processing without exposing sensitive information. This study explores the application of Fully Homomorphic Encryption (FHE) with machine learning models for classifying medical dialogues containing potentially sensitive personally identifiable information (PII). Benchmarking FHE-based inference models against the traditional plaintext approaches. This study demonstrates that FHE can achieve comparable ensuring robust data privacy. However, enhanced security comes at the cost of increased computational time and memory usage. The study highlights a critical trade-off between maintaining high-speed inference and preserving privacy, providing valuable insights into medical data. These results guide decision-makers in evaluating the deployment of secure, FHE-based solutions in production environments. This study explores integrating privacy-preserving techniques with machine learning and outlines future research directions.
Source: pdf
Extracted details¶
| Field | Value | Source | Confidence |
|---|---|---|---|
| contribution | FHE-based privacy-preserving classification of medical dialogues containing PII | llm | 0.70 |
| contribution | benchmark of FHE vs plaintext inference for SGD, XGBoost and decision-tree classifiers | llm | 0.70 |
| contribution | analysis of accuracy vs computation-time/memory trade-off for FHE deployment | llm | 0.70 |
| schemes | FHE (scheme not named in excerpt) | llm | 0.70 |
| models | SGD classifier | llm | 0.70 |
| models | XGBoost | llm | 0.70 |
| models | decision tree | llm | 0.70 |
| datasets | medical dialogue PII dataset (unnamed) | llm | 0.70 |
| tasks | binary text classification | llm | 0.70 |
| tasks | medical PII detection | llm | 0.70 |
| tasks | healthcare NLP | llm | 0.70 |
| interaction | non-interactive FHE | llm | 0.70 |
| limitations | significant increase in training and prediction time under FHE | llm | 0.70 |
| limitations | increased memory usage under FHE | llm | 0.70 |
| llm relevance | low | llm | 0.70 |
| key result | accuracy: 100% (SGD+FHE), 100% (XGB+FHE), 91.7% (DT+FHE) (validation set, binary classification under FHE) | llm | 0.70 |
| key result | prediction time: 270.39 s (SGD+FHE), 0.410 s (XGB+FHE), 20459.98 s (DT+FHE) (FHE inference) | llm | 0.70 |
| key result | training time: 44.40 s (SGD+FHE), 5.05 s (XGB+FHE), 91.20 s (DT+FHE) (FHE setting) | llm | 0.70 |
Keyword mentions¶
Counts of vocabulary terms in the full text (first page of occurrence in brackets).
- scheme: Partially homomorphic encryption 5 [p.2], Leveled / somewhat HE 5 [p.2]
- model: Decision tree 11 [p.2], Gradient boosting / XGBoost 5 [p.3], Transformer 1 [p.5], GAN 1 [p.3]
- task: Medical / healthcare 26 [p.1], Genomics 2 [p.1], Finance 2 [p.1]
- topic: Homomorphic encryption 49 [p.1], Privacy-preserving machine learning 4 [p.1], Private / secure inference 3 [p.3]
Files¶
| Path (under papers/) | Collection | Pages | Size | SHA-256 | Note |
|---|---|---|---|---|---|
references/ppml/applications/ndolo-2025_ml-nlp-approach-with-fhe-techniques-in-medical.pdf |
ppml/applications | 6 | 2.2 MB | c1f1f4d525e9 |
was Ndolo et al. - 2025 - Application of Machine Learning-NLP Approach with Fully Homomorphic Encryption Techniques in Medical.pdf |
Database key: application-of-machine-learning-nlp-approach-with-fully-homomorphic · created 2026-11-05 · updated 2026-11-05 · edit with python3 tools/paperdb.py set application-of-machine-learning-nlp-approach-with-fully-homomorphic COLUMN VALUE