|
| Ref. No | Technologies used | Key contributions | Limitations |
|
| [61] | Federated semantic segmentation models | In this study, federated semantic segmentation models performed on multimodal brain scans are similar to models trained for data sharing | Data acquisition is a major challenge |
|
| [63] | DL techniques | The objective of this review is to present an overview of current research on applying DL to clinical tasks derived from EHR data, in which we examine the variety of DL techniques and frameworks applied to various types of clinical tasks | (i) Model interpretability |
| (ii) Data heterogeneity, and |
| (iii) Lack of universal benchmarks |
|
| [64] | A descriptive and inferential statistical analysis | The purpose of this survey was to assess electronic communication and awareness of HIPAA privacy and security rules, especially in the context of text messaging | (i) First, there was a low response rate, raising concern for nonresponse bias |
| (ii) Second, survey results may be skewed by cognitive biases |
|
| [67] | Federated- autonomous deep learning (FADL) method | This study finds that FADL exceeds traditional federal methods of learning and that balancing global to local formation is an important feature of distributed techniques, especially in the field of healthcare | Accessing data is complex and slow due to:
(i) Security |
| (ii) Privacy |
| (iii) Regulatory and |
| (iv) Operational issues |
|
| [69] | FL framework | This study reveals that while differential privacy in a federal system is commonly adopted, it can lead to considerable losses in model performance in healthcare applications | (i) Distributed data silos |
| (ii) Privacy issues |
|
| [70] | An FL framework can develop global ADR prediction models, based on local health data held at different locations | In this study, we focused on algorithms conducive to distributed solutions, including gradient descent, as a method supported by FL | Frameworks for predicting adverse drug reactions (ADR) using centralized learning |
|
| [78] | Blockchain and AI | In this study, we have provided a comprehensive coronavirus (COVID-19) investigation utilizing blockchain and AI | The challenges are analyzed in this article from four different perspectives:
(i) Regulatory considerations |
| (ii) Maintaining people’s privacy |
| (iii) The security of blockchain and AI ecosystems, and |
| (iv) A lack of unified databases |
|
