Implementation of AI used Algorithms Design Models for to Create Healthcare Smart System

In the field of healthcare spending prediction, various methods, including probabilistic and neural draws near, have been suggested, yet an insufficient stress has been placed on assessing their forecasting powers, alongside ensemble techniques. This work tries to assess different models of statistics, such as persistence (a starting point) and autoregressive mixed moving average (ARIMA), next to computational models of neurons such as multiple-layer perceptron (MLP) as well as lengthy short-term memory (LSTM). The main focus swirls around forecasting weekly average expenses on specific pain medicines. Further, an ensemble model, merging results from ARIMA, MLP, while LSTM, was adjusted for improved accuracy. The study scrutinizes the effect of shuffling conditioning data and adding breakdown in MLPs, when likewise, mixing training data and leaving out dropout in LSTMs while instruction. The results show the better performance of the combined approach over the individual’s dedication, ARIMA, MLP, and LSTM techniques across both pain medicines. Notably, for MLP models, keeping an order of training inputs while instituting dropout gave ideal results, while to stay LSTM models, changing training data without applying dropping proved beneficial. This study underlines the implications of applying statistical, neural, while combined techniques in time-series predictions for healthcare results.