Diabetes prediction using Shapley additive explanations and DSaaS over machine learning classifiers: a novel healthcare paradigm

Technologies like cloud computing, Artificial Intelligence (AI), and Machine intelligence technologies must combine to accomplish computational intelligence. To deliberate the tasks promptly and effectively, the software systems must possess data science competencies. The data science capabilities include intelligent predictive analytics, an optimal solution with high precision, efficient resource utilization, and extracting meaningful information from vast quantities of data. In this paper, we deeply analyzed the confluence of cloud-based technologies with AI, IoT, and data science capabilities, where data science is introduced as a Service (DSaaS) platform for cloud-based services to predict diabetes. To this end, a paradigm for smart healthcare systems using data Science and cloud-enabled platforms is proposed. The feature ranking uses MRMR, ReliefF, and ANOVA followed by Shapley additive explanations (Shap) for attribution selection. The predictions are performed using the Neural Network model for female patients suffering from diabetic diseases. The accuracy achieved by the Neural Network (NN) classifier is 77.9% on a sample dataset of 768 instances and 9 attributes. The Positive Predictive Value (PPV) achieved by the classifier is 79.3%.