Skip to content

Private pathological assessment via machine learning and homomorphic encryption

Authors: Ahmad Al Badawi; Mohd Faizal Bin Yusof
Year: 2024
Venue: BioData Mining

type: application collection: ppml, sok-references status: auto

Links: DOI · link

Summary

Auto-extracted summary (unreviewed)

The paper builds a framework for private pathological assessment by running SVM inference on CKKS-encrypted medical records, together with a feature-extraction technique that turns medical images into feature vectors. Evaluated on several medical datasets, the encrypted SVM reaches the same classification accuracy as plaintext SVM at 128-bit security, with each secure inference taking only a few seconds on general-purpose CPUs.

Abstract

Purpose: The objective of this research is to explore the applicability of machine learning and fully homomorphic encryption (FHE) in the private pathological assessment, with a focus on the inference phase of support vector machines (SVM) for the classification of confidential medical data. Methods: A framework is introduced that utilizes the Cheon-Kim-Kim-Song (CKKS) FHE scheme, facilitating the execution of SVM inference on encrypted datasets. This framework ensures the privacy of patient data and negates the necessity of decryption during the analytical process. Additionally, an efficient feature extraction technique is presented for the transformation of medical imagery into vectorial representations. Results: The system’s evaluation across various datasets substantiates its practicality and efficacy. The proposed method delivers classification accuracy and performance on par with traditional, non-encrypted SVM inference, while upholding a 128-bit security level against established cryptographic attacks targeting the CKKS scheme. The secure inference process is executed within a temporal span of mere seconds. Conclusion: The findings of this study underscore the viability of FHE in enhancing the security and efficiency of bioinformatics analyses, potentially benefiting fields such as cardiology, oncology, and medical imagery. The implications of this research are significant for the future of privacy-preserving machine learning, promoting progress in diagnostic procedures, tailored medical treatments, and clinical investigations.
Source: pdf

Extracted details

Field Value Source Confidence
contribution CKKS-based framework for SVM inference on encrypted medical data llm 0.65
contribution feature-extraction technique converting medical images into vectors for encrypted SVM llm 0.65
contribution evaluation on multiple medical datasets showing accuracy on par with plaintext SVM llm 0.65
contribution open-source implementation for general-purpose CPUs llm 0.65
schemes CKKS llm 0.65
models SVM llm 0.65
datasets heart disease (Cleveland Clinic) llm 0.65
tasks medical diagnosis llm 0.65
tasks pathological assessment llm 0.65
tasks medical image classification llm 0.65
tasks binary classification llm 0.65
hardware general-purpose CPU llm 0.65
security level 128-bit llm 0.65
threat model untrusted cloud server, client-server llm 0.65
interaction non-interactive FHE llm 0.65
limitations FHE incurs computational overhead and ciphertext expansion llm 0.65
limitations non-polynomial functions must be approximated numerically, introducing potential inaccuracy llm 0.65
llm relevance low llm 0.65
key result inference latency: a few seconds (CKKS SVM inference, 128-bit security, CPU) llm 0.65
key result accuracy: on par with plaintext SVM (encrypted SVM inference across evaluated medical datasets) llm 0.65

SoK reference [10]

A. Al Badawi and M. Faizal Bin Yusof, “Private pathological assessment via machine learning and homomorphic encryption,” BioData Mining, vol. 17, no. 1, p. 33, 2024.

Listed on the reference page of the site.

Keyword mentions

Counts of vocabulary terms in the full text (first page of occurrence in brackets).

  • scheme: CKKS 50 [p.1], MPC 7 [p.2], Paillier 3 [p.20], Differential privacy 3 [p.2], Multiparty HE 2 [p.24], Garbled circuits 2 [p.21]
  • library: OpenFHE 6 [p.11]
  • model: SVM 105 [p.1], Autoencoder 8 [p.14], CNN 7 [p.4], RNN / LSTM 3 [p.4], Logistic regression 3 [p.4], Linear regression 2 [p.4], AlexNet 1 [p.24], GAN 1 [p.25]
  • dataset: Breast cancer (Wisconsin) 11 [p.3], Heart disease 11 [p.2], MedMNIST 11 [p.3], MNIST 6 [p.3], Diabetes (Pima) 6 [p.4], Chest X-ray 4 [p.12], UCI repository 3 [p.24], CIFAR-10 2 [p.20], ECG 2 [p.12]
  • hardware: GPU 7 [p.5], FPGA 2 [p.22], ASIC 1 [p.22], CUDA 1 [p.25]
  • technique: Sigmoid 24 [p.3], Chebyshev 7 [p.10], Polynomial approximation 6 [p.10], Convolution 5 [p.4], Bootstrapping 4 [p.6], ReLU 2 [p.14], SIMD packing 2 [p.7], Fine-tuning 2 [p.11], Federated learning 2 [p.25], Semi-honest model 2 [p.8]
  • assumption: 128-bit security 4 [p.3], LWE 3 [p.6], RLWE 2 [p.6], Lattice-based 2 [p.23]
  • task: Medical / healthcare 52 [p.1], Genomics 10 [p.4], Image classification 2 [p.18], Text classification 2 [p.10]
  • topic: Homomorphic encryption 59 [p.1], Open source 6 [p.4], Benchmark 5 [p.12], Privacy-preserving machine learning 3 [p.1]

Files

Path (under papers/) Collection Pages Size SHA-256 Note
references/ppml/applications/al-badawi-2024_private-pathological-assessment-via-ml-and-he.pdf ppml/applications 25 1.4 MB 44aac4d431f7 byte-identical to references/sok-references/010_private-pathological-assessment-via-machine-learning-and-hom.pdf; was Badawi and Yusof - 2024 - Private pathological assessment via machine learning and homomorphic encryption.pdf
references/sok-references/010_private-pathological-assessment-via-machine-learning-and-hom.pdf sok-references 25 1.4 MB 44aac4d431f7

Database key: private-pathological-assessment-via-machine-learning-and-homomorphic · created 2026-11-05 · updated 2026-11-05 · edit with python3 tools/paperdb.py set private-pathological-assessment-via-machine-learning-and-homomorphic COLUMN VALUE