<p>The technical debt of software refers to a set of technical problems and defects that are present in the software and need to be fixed to ensure proper and acceptable performance of the software. These problems sometimes arise for reasons such as the time pressure the development team is under to ensure timely delivery, lack of proper planning, and failure to use standard and optimized methodologies. Accepted Technical Debt or Self-admitted Technical Debt (SATD) is a type of technical debt that has been intentionally implemented by developers in an inappropriate and inferior way, but will be carried over to future decisions and projects. One of the most important sources for identifying accepted technical debt is the source code or comments in the software source code. So far, methods such as keyword search, pattern search, natural language processing and deep learning have been used. This paper proposes a prediction model called QuantumPSO-LSTM together with the Bidirectional Encoder Representations from Transformers (BERT) word embedding method to identify and classify source code annotations. In the proposed model, due to the use of text data, the Long short-term memory (LSTM) is able to overcome the long-term dependency problem that exists in Recurrent Neural Networks (RNN). Moreover, QuantumPSO, a swarm intelligence algorithm based on quantum behavior, optimizes the hyperparameters of the model and avoids preprocessing. The results show that the proposed model could achieve an improvement of 10% and 13% compared to LSTM and Convolutional Neural Network (CNN) networks, respectively.</p>