Enhancing Machine Learning Model Efficiency through Quantization and Bit Depth Optimization: A Performance Analysis on Healthcare Data