Trusted sharing and access control model of medical big data based on privacy computing

This paper focuses on the balance between data privacy protection and availability in the medical big data environment. By constructing a specific model, the performance of encryption, decryption, access control decision-making and other operations is evaluated, and the availability of data in disease diagnosis and prediction before and after processing is compared. The experimental results show that the feature integrity of the data before and after processing is 98% and 96% respectively, and the prediction accuracy is reduced from 85% to 83%. The availability of data in different formats also remains at a high level, and the availability of structured medical record data is 96%. In terms of performance overhead, with the increase of data volume, the time overhead of encryption and decryption increases approximately linearly. For example, the encryption time of Paillier encryption algorithm for 100MB data is 120 milliseconds, which increases to 580 milliseconds when it is 500MB, the access control decision-making time is stable within 50 milliseconds, and the storage space of encrypted data is increased by about 20% taking 1GB as an example. The research shows that the model can effectively protect data privacy, while maintaining the availability of data for medical applications, and the performance overhead is affordable, which provides strong support for medical research and applications.