Secure Capsule Computing for Privacy-Preserving Distributed AI

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Access to diverse, high-fidelity, privacy-protected data is a significant barrier for developing robust and generalizable artificial intelligence applications, particularly in healthcare, due to regulatory, legal, and ethical requirements for maintaining patient information privacy, leading to proof-of-concept algorithms that lack real-world applicability.

Innovation Solution

A system utilizing secure capsule computing frameworks that integrate algorithms with data storage structures in a privacy-preserving manner, enabling federated training and validation across multiple data sources while maintaining data privacy, through techniques like differential privacy and homomorphic encryption.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If data is shared across multiple sources to train AI algorithms, then algorithm robustness and generalizability are improved, but data privacy and security are compromised

Engineering Contradiction:

Improvealgorithm robustnessVSAvoiddata privacy risk

Core Design Contradiction:

ReliabilityVSObject-affected harmful factors

Solution Approach 1:

The patent introduces secure multi-party computation (SMC) and homomorphic encryption as intermediary technologies that enable AI algorithm training on distributed data without direct data sharing. These intermediaries allow computations to be performed on encrypted data, producing results that reveal no information about the underlying private data, thus resolving the contradiction between data utilization and privacy protection

Inventive Principle:

Principle #24Intermediary (Mediator)

Solution Approach 2:

The patent segments the AI training process into distributed computations performed separately at each data holder's location using federated learning. Each participant trains local model copies on their private data, and only model updates (not raw data) are shared and aggregated. This segmentation enables algorithm robustness through diverse data sources while maintaining data privacy by keeping sensitive information localized

Inventive Principle:

Principle #1Segmentation

2Object-affected harmful factors

If data is kept private and secure, then patient information protection is improved, but access to diverse data for AI development is limited

Engineering Contradiction:

Improvepatient information protectionVSAvoiddata accessibility

Core Design Contradiction:

Object-affected harmful factorsVSAdaptability or versatility

Solution Approach 1:

The patent replaces the mechanical approach of physical data sharing and transfer with cryptographic and computational methods. Instead of moving data between systems, the invention uses homomorphic encryption and secure multi-party computation to perform computations directly on encrypted data, substituting data movement with mathematical transformations that preserve privacy while enabling access

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

3Productivity

If data is shared without privacy protection, then algorithm development speed is improved, but risk of data leakage increases

Engineering Contradiction:

Improvealgorithm development speedVSAvoiddata leakage risk

Core Design Contradiction:

ProductivityVSObject-generated harmful factors

Solution Approach 1:

The patent applies preliminary cryptographic transformations (encryption, hashing, and other privacy-preserving techniques) to data before any computational operations. By pre-processing data with privacy protections built-in, the system enables rapid algorithm development on protected data, eliminating the need for slow iterative security reviews and manual privacy compliance checks

Inventive Principle:

Principle #10Preliminary action

4Object-affected harmful factors

If federated learning is used to train models on distributed data, then data privacy is preserved, but communication overhead and system complexity increase

Engineering Contradiction:

Improvedata privacy preservationVSAvoidsystem complexity

Core Design Contradiction:

Object-affected harmful factorsVSDevice complexity

Solution Approach 1:

The patent combines multiple privacy-preserving techniques (federated learning, homomorphic encryption, secure multi-party computation) into a unified framework. By merging these approaches, the system manages the inherent complexity through integrated architecture design, where each technique compensates for the weaknesses of others, achieving robust privacy protection without requiring separate complex systems for each method

Inventive Principle:

Principle #5Merging (Combining)

Data Source

PatentUS20260037836A1Distributed privacy-preserving computing on protected data

Publication Date: 2026.02.05 RGT UNIV OF CALIFORNIA

AI summary

The present disclosure relates to techniques for developing artificial intelligence algorithms by distributing analytics to multiple sources of privacy protected, harmonized data. Particularly, aspects are directed to a computer implemented method that includes receiving an algorithm and input data requirements associated with the algorithm, identifying data assets as being available from a data host based on the input data requirements, curating the data assets within a data storage structure that is within infrastructure of the data host, and integrating the algorithm into a secure capsule computing framework. The secure capsule computing framework serves the algorithm to the data assets within the data storage structure in a secure manner that preserves privacy of the data assets and the algorithm. The computer implemented method further includes running the data assets through the algorithm to obtain an inference.