Liquid desiccant dehumidifiers have received widespread attention due to their ability to utilize low-grade thermal energy and high dehumidification efficiency. However， the prediction accuracy of their mass transfer performance still needs to be improved. This article built up an experimental platform of a single-channel internally cooled liquid desiccant dehumidifier to study the effects of different parameters on its mass transfer efficiency. Meanwhile， an artificial neural network （ANN） model was established with MATLAB to predict the mass transfer efficiency of the dehumidifier. The ANN model was verified and validated with the above experimental data. The results indicated that the a mean absolute relative difference （MARD） between the predicted Sh of the ANN model and the experimental Sh was 4.07%. Compared with existing empirical formulas， the ANN model established in this paper had higher prediction accuracy. In addition， this article also used the ANN model to study the trend of Sh under different parameter changes， thereby investigating the impact of different parameters on the dehumidification performance.