Objective The estimation of the state of health SOH of lithium-ion power batteries is an important research topic in the current field. Accurate SOH estimation is crucial for ensuring the normal operation of electric vehicles and maintaining the lifespan of lithium batteries. To address the challenges in lithium battery SOH prediction such as difficulty in selecting health features and low prediction accuracy a lithium battery SOH prediction model based on health factor extraction and the IDBO-BiLSTM model is proposed. Methods First health factors were extracted from NASA data and principal component analysis PCA was applied to reduce dimensionality and fuse features. Next the fused features were decomposed into multiple modal components using the complete ensemble empirical mode decomposition with adaptive noise CEEMDAN from which the effective modes were selected. Subsequently an improved dung beetle optimizer IDBO was used to optimize the parameters of the BiLSTM model to obtain the optimal BiLSTM model. Finally the extracted effective features were fed into the optimal BiLSTM model for SOH prediction. To demonstrate the applicability of the method comparative experiments were conducted using three battery datasets from NASA. Results The results show that the IDBO-BiLSTM-based prediction model performs well in SOH estimation achieving a mean squared error MSE below 4e-4 and a coefficient of determination R 2 above 0. 98 in all test cases.Conclusion The proposed model demonstrates superior performance and feasibility in improving the accuracy of lithium battery SOH prediction.