Medical image-based research requires heavy computational workload associated with image analysis and collaborative device independent platforms to incorporate expert opinions from multiple institutions. Cloud-based resources such as Microsoft Azure Machine Learning Studio (MAMLS) provide such a platform that is conducive to the medical-image-based data analysis. This paper fosters the advantages of the cloud-based computing frameworks (such as MAMLS) and presents a practical work-flow well-suited for the standard machine learning tasks seen in medical image research viz., binary classification, multi-class learning, regression and so on. The proposed automated generalized workflow allows medical researchers/practitioners to focus on data inferencing rather than dealing with the intricate details of predictive modeling, such as feature and model selection. The scalable architecture of the proposed flow utilizes the MAMLS framework to processes data sets that require partial core storage space in the virtual machine to one complete core storage space in a common flow. Also, the proposed flow invokes multiple feature ranking and predictive models in parallel for automated selection and parameterization of the optimal data model. The performance of the proposed flow is bench-marked on 14 public data sets and four local medical image data sets (~0.12 MB–1.22 GB) using a single common flow, while ensuring better (~ 8% improvement) or atleast similar generalization capability with respect to existing works.