In view of the privacy leakage and data security problems faced by the sensitive data processing of users of the mental health assessment system, a Paillier homomorphic encryption technology is proposed to be applied to the psychological assessment system to ensure that the data remains encrypted during the processing process. First, based on the Paillier homomorphic encryption algorithm, this paper encrypts the psychological assessment data input by the user to ensure the security of the data during transmission and storage. Secondly, the encrypted data is input into the psychological assessment analysis model, and homomorphic operations are used to perform necessary statistical analysis and calculations to achieve the evaluation of the assessment results without decrypting the data. Then the calculation results are decrypted to obtain the final psychological assessment results, and subsequent analysis and interpretation are carried out. After using homomorphic encryption technology, the average accuracy of the mental health assessment reaches 94.68%. Homomorphic encryption technology can not only effectively protect the privacy of sensitive data but also maintain good data processing performance.