With the growing expansion of remote healthcare services, the management and access to patient and physician information through electronic health records (EHRs) have become increasingly feasible. One of the most significant advantages of such systems is the ability to match patients with physicians based on patients’ specific needs, which not only reduces waiting times but also improves the overall quality of medical services. However, privacy concerns related to the storage and transmission of sensitive medical data on servers pose serious challenges, including the risk of patient data leakage or exposure of physician information to malicious attacks. Consequently, the development of a secure and efficient matching system that ensures data confidentiality has become a critical necessity.

To address these challenges, the use of cryptographic tools for protecting data prior to uploading it to cloud servers or other storage platforms has been proposed as an effective solution. Nevertheless, conventional encryption methods impose limitations on service optimization due to their inability to support search and operations over encrypted data. Therefore, this paper proposes a novel mechanism for secure and efficient patient-physician matching, enabling multi-keyword search over encrypted data in accordance with patient-specific requirements.

The proposed scheme employs an innovative technique based on inner product computation to facilitate similarity measurement and matching between two vectors, which not only enhances efficiency but also improves the accuracy of the results. Another key advantage of the proposed design is its fast search capability, achieved through a computational complexity of O(n), which contributes to better user experience for physicians and higher patient satisfaction. Security analysis confirms that the proposed scheme effectively preserves the privacy of both patient and physician data without disclosing any sensitive information. Compared to the model presented in [1] with a time complexity of 𝑂(𝑛𝑚𝑘), the proposed scheme significantly reduces the computational complexity to 𝑂(𝑛) while achieving higher matching accuracy and more efficient search speed. Furthermore, by supporting weighted vectors, multi-keyword queries, and employing ASPE, it offers enhanced security against KPA and CPA attacks.