English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Tools annual

Global: Zendy Free Plan

Global: Zendy Tools monthly

Global: Zendy Plus monthly

Accurate cost prediction of healthcare resources is challenging as diverse factors affect the overall prediction. The cost of healthcare providers is increasing exponentially as different healthcare providers charge differently for the same service due to various factors, majorly the sky rocketing inflation and increased population. It increases the importance of predicting healthcare costs to avoid unpleasant surprises. This study aims to provide the expected cost of healthcare providers that helps the patients in resource allocation and strengthens decision-making according to their resources. This paper proposes three hybrid Deep Learning (DL) models, Visual Geometry Group and Stacked Autoencoder (VGG-SAE), Visual Geometry Group and Deep Neural Network (VGG-DNN), and Stacked Autoencoder and Deep Neural Network (SAE-DNN), which optimize learning the hidden patterns from the given data more efficiently than individual models. The three hybrid DL models estimate the cost of healthcare providers effectively. The preprocessing is performed using the mode imputation for handling the missing values, Z-score for removing the outliers and standard scaler for standardizing the data. To train the hybrid models on optimum parameters, the Random Search technique is used that provides the best hyper-parameters of each hybrid model. The interpretation of the hybrid models&#x2019; output is achieved using the SHapley Additive ExPlanations (SHAP) technique. The performances of VGG-SAE, VGG-DNN, and SAE-DNN are compared with the baseline DL models such as SAE, DNN, and VGG. To assess the robustness of the proposed approach, the hybrid models are trained on two different datasets of healthcare such as Healthcare Providers and Hospital Inpatient Cost Transparency. With the hyper-parameter tuning of the Healthcare Providers Dataset, VGG-SAE achieved MSE of 0.01, RMSE of 0.13, MAE of 0.02, and R-squared of 0.98. VGG-DNN achieved MSE of 0.01, RMSE of 0.12, MAE of 0.02, and R-squared of 0.99. SAE-DNN achieved MSE of 0.01, RMSE of 0.11, MAE of 0.02, and R-squared of 0.99. With the hyper-parameter tuning of the Hospital Inpatient Cost Transparency Dataset, VGG-SAE achieved MSE of 0.007, RMSE of 0.08, MAE of 0.03, R-squared of 0.99, and execution time of 1680 seconds. VGG-DNN achieved MSE of 0.0006, RMSE of 0.08, MAE of 0.03, R-squared of 0.99, and execution time of 645 seconds. SAE-DNN achieved MSE of 0.003, RMSE of 0.06, MAE of 0.02, R-squared of 0.99, and execution time of 850 seconds. Our proposed hybrid combinations outperformed other deep models and Machine Learning (ML) techniques such as SAE, DNN, VGG, SVR and GBR, which ensures high efficiency of the proposed models in terms of healthcare providers cost.

New hybrid deep learning models to predict cost from healthcare providers in smart hospitals