Blind De-identification of Private Medical Data Using Byte Fragmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for processing private medical data with AI and machine learning models often violate privacy regulations, such as HIPAA and GDPR, as they expose sensitive information during data processing.
Innovation Solution
A method involving splitting each byte of data into two random portions, inserting random bits, and processing these portions separately by different entities using encrypted algorithms, ensuring that no private data is shared directly, thus maintaining privacy compliance with HIPAA, GDPR, and CCPA standards.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If private medical data is processed directly by AI and machine learning models, then the algorithms can operate on real data to provide accurate medical diagnostics and treatment, but private information is exposed violating HIPAA and GDPR regulations
Solution Approach 1:
The patent divides private medical data into multiple fragmented portions, where each fragment contains only a subset of the original information. These fragments are then distributed to different computing entities for processing. This segmentation ensures that no single entity possesses the complete private data, thereby preventing privacy violations while still enabling accurate medical diagnostics through collective processing of all fragments.
Solution Approach 2:
The patent introduces encrypted data as an intermediary form between the original private medical data and the AI processing system. The data is encrypted using homomorphic encryption or similar techniques that allow computations to be performed on the encrypted data without decrypting it. This intermediary encrypted form protects patient privacy while still enabling the AI models to process and analyze the medical information accurately.
2Object-affected harmful factors
If data is de-identified by removing or masking personal information, then privacy regulations are complied with, but the data loses some utility for accurate medical analysis
Solution Approach 1:
The patent performs preliminary encryption and fragmentation of the medical data before it is transmitted to or processed by external AI systems. This preliminary action ensures that the data is already in a protected state before leaving the controlled environment, maintaining both privacy and full data utility. The encryption scheme is designed to preserve all statistical and analytical properties needed for medical analysis while preventing identification of individual patients.
Solution Approach 2:
The patent transforms the data from its original readable format into encrypted mathematical representations that preserve the statistical relationships and patterns necessary for medical analysis. The encryption parameters are carefully chosen to maintain the distribution, correlations, and other analytical properties of the original medical data, ensuring that de-identified data retains full utility for diagnostic and research purposes.
3Object-affected harmful factors
If encrypted algorithms are used to process data, then privacy is maintained, but the computational complexity and processing time increase
Solution Approach 1:
The patent segments both the data and the computational workload across multiple entities. By dividing the processing task into smaller fragments that can be executed in parallel on different systems, the overall processing time is reduced despite the added complexity of encryption. Each entity processes only its assigned fragment, and the results are aggregated to produce the final diagnostic outcome, maintaining security while improving efficiency through parallel computation.
Data Source
AI summary
Disclosed is a system and method of de-identifying data. A method includes splitting, at a first entity, a byte of data of an original record into a first random portion and a second random portion, inserting first random bits into the first random portion to yield a first new byte and inserting second random bits into the second random portion to yield a second new byte. The method then includes transmitting the second new byte to a second entity, receiving, at the first entity, a first portion of an algorithm from the second entity and processing the first new byte by the first portion of the algorithm to yield a first partial result. The first partial result can be combined with a second partial result from the second entity processing the second new byte by a second portion of the algorithm.