Accurately estimating the state of health (SOH) and predicting the remaining useful life (RUL) of lithium‐ion batteries have become crucial challenges due to the complex aging mechanisms. This paper proposes a data‐driven method for SOH estimation and RUL prediction based on a partial differential thermal voltammetry (DTV) curve and long short‐term memory (LSTM) network. The Gaussian filter method is applied to eliminate measurement noise and obtain a smooth DTV curve. A novel health feature (HF) based on equally spaced sampling points on the DTV curve within partial voltage intervals is proposed for estimating SOH. Then, highly correlated HFs are selected as inputs to the proposed dual LSTM models for estimating SOH and predicting RUL. The aging datasets of three batteries from NASA Prognostics Center of Excellence are utilized to demonstrate the effectiveness and robustness of the proposed method for estimating SOH and RUL. The root mean square error (RMSE) for estimating SOH across the three batteries is less than 1.03%, and the RMSE for predicting RUL is less than 3.5 cycles. The validation results indicate that the proposed method provides an accurate and robust estimation of SOH and prediction of RUL.