The paper focuses on enhancing the forecasting of India's GDP growth using a modified LSTM (Long Short-Term Memory) approach. It emphasizes the critical role of GDP growth forecasting in analyzing economic outlook, policy creation, and business decision-making. The study highlights the importance of various factors such as domestic consumption, investment levels, government spending, global trade dynamics, inflation, and monetary policy in forecasting GDP growth. The research also discusses the challenges and uncertainties in GDP growth forecasting, including natural disasters, geopolitical conflicts, and policy changes. The proposed modified LSTM model addresses these challenges by incorporating outlier treatment, missing data handling, relevant factor selection, and class imbalance mitigation. The model uses the Modified 'z' score method for outlier detection, the Expectation-Maximization (EM) algorithm for missing data, Lasso Regression with lagged variables for feature selection, and Time Series SMOTE for class imbalance. The model was trained and validated using historical GDP growth data, achieving an accuracy of 94.58%, a Mean Absolute Percentage Error (MAPE) of 1.26%, and a Root Mean Squared Error (RMSE) of 1.05%. The study also examines the impact of bank crises on Indian GDP growth, using an LSTM model to predict GDP growth under different scenarios. The results show a statistically significant negative impact of bank crises on GDP growth, with severe crises and deteriorating economic indicators leading to a 0.1-point decline in GDP growth. In conclusion, the modified LSTM model effectively captures the complex patterns in GDP time series data, providing accurate and reliable forecasts. The findings have implications for policymakers, economists, and financial analysts in making informed decisions and developing strategies to mitigate the adverse effects of economic shocks.The paper focuses on enhancing the forecasting of India's GDP growth using a modified LSTM (Long Short-Term Memory) approach. It emphasizes the critical role of GDP growth forecasting in analyzing economic outlook, policy creation, and business decision-making. The study highlights the importance of various factors such as domestic consumption, investment levels, government spending, global trade dynamics, inflation, and monetary policy in forecasting GDP growth. The research also discusses the challenges and uncertainties in GDP growth forecasting, including natural disasters, geopolitical conflicts, and policy changes. The proposed modified LSTM model addresses these challenges by incorporating outlier treatment, missing data handling, relevant factor selection, and class imbalance mitigation. The model uses the Modified 'z' score method for outlier detection, the Expectation-Maximization (EM) algorithm for missing data, Lasso Regression with lagged variables for feature selection, and Time Series SMOTE for class imbalance. The model was trained and validated using historical GDP growth data, achieving an accuracy of 94.58%, a Mean Absolute Percentage Error (MAPE) of 1.26%, and a Root Mean Squared Error (RMSE) of 1.05%. The study also examines the impact of bank crises on Indian GDP growth, using an LSTM model to predict GDP growth under different scenarios. The results show a statistically significant negative impact of bank crises on GDP growth, with severe crises and deteriorating economic indicators leading to a 0.1-point decline in GDP growth. In conclusion, the modified LSTM model effectively captures the complex patterns in GDP time series data, providing accurate and reliable forecasts. The findings have implications for policymakers, economists, and financial analysts in making informed decisions and developing strategies to mitigate the adverse effects of economic shocks.