To solve the problem of accurately predicting remaining life of lithium battery, this paper proposes an indirect prediction method based on improved grey wolf optimization least-squares support vector machine (IGWO-LSSVM). Three indirect health factors characterizing battery performance degradation are derived from discharge characteristic curves. To enhance prediction accuracy, the study incorporates a tent chaotic map, a nonlinear decreasing convergence factor, and a Levi flight strategy into the grey wolf algorithm. Combined with the LSSVM model, the lithium battery life prediction model with global optimization is formed. The proposed method is verified using the NASA data set and compared with GWO-LSSVM, PSO-ELM and BP algorithms. Experimental results show that the improved algorithm proposed in this paper outperforms other methods in terms of prediction accuracy.