Sharecare is a digital company that provides a mobile app with artificial intelligence to consumers. It has strong views on AI and how it should be used.

Sharecare believes companies that use augmented analysis and AI to analyze data using business intelligence software are missing out on the benefits and advantages of data fluency, federated AI and data fluency. Sharecare uses data fluency as well as federated AI to discover hidden similarities in data that other business intelligence tools are unable to detect in health settings.

To get a deeper understanding about data fluency, federated AI and healthcare IT news, Healthcare IT News interviewed Akshay Sharma (executive vice president for artificial intelligence at Sharecare) for an in-depth interview.

Q. How does federated AI differ from any other type of AI?

A: Federated AI (or federated Learning) guarantees that data is not lost from the device. Applications that run at the edges of the network can learn to process data and build more efficient models. They also share a mathematical representation of key clinical characteristics.

Traditional machine learning relies on centralizing data in order to build and train a model. Edge AI and Federational Learning combined with other privacy preserving techniques and zero trust Infrastructure make it possible for models to be built in a distributed data structure while lowering risk of attacks from any one point.

In cloud settings, federated Learning can also be applied. Data doesn’t have leave the systems where it exists. However, it can still allow for learning. This is federated cloud-learning, which can be used by organizations to collaborate while keeping the data private.

A: What is data Fluency? Why is it important for AI.

A: Data fluency can be described as a set and framework of tools to quickly unlock clinical data’s potential value. All stakeholders must participate simultaneously in a collaborative environment. A machine learning environment with a data fluency framework engages clinicians, actuaries, data engineers, data scientists, managers, infrastructure engineers and all other business stakeholders to explore the data, ask questions, quickly build analytics and even model the data.

This unique approach to enterprise information analytics was created for healthcare. It is intended to improve collaboration, workflows, and rapid prototyping of new ideas before spending money building models.

Q. How can data fluency platforms help analysts, engineers, data scientists, and clinicians collaborate more efficiently and easily?

A Many traditional healthcare systems are extremely siloed and organizations struggle to find the value in their data, unlock actionable trends, and gain clinical insights. Data transformation systems are often isolated from data creation teams. Furthermore, data scientists and engineers use coding languages to create data while clinicians and finance team use Word and Excel.

A disconnect can lead to data knowledge that is not being translated within the programming environment. Transforms between system boundaries can be lossy without feedback loops that could improve the algorithm or the code. To build effective and transparent health algorithms, all stakeholders must have early access to the data.

The modern healthcare platform facilitates collaboration between cross-functional teams using a single data-driven point-of-view in Python Notebooks with non-engineering partner UIs. It can be costly and time-consuming to build AI models. Therefore, it is important to get early prototype input from experts in other domains.

Data fluency provides an environment for critical stakeholders to discover the value on top of the data or insights and in a real-time, agile and iterative way. Instant feedback can help to improve the code in the notebook or the model.

Each domain expert can have multiple data views that facilitate deep collaboration and data insight discovery, enabling the continuous learning environment from care to research and from research to care. Data fluency is possible with cloud-native technologies. Many of these techniques can also be extended to computing on the edge, where patient data and other information reside.

Q. What do you think the future of analytics for healthcare is?

A: Traditional analytics in healthcare focuses on understanding a particular set of data through the use of business intelligence-focused tools. These tools are often used by business users, analysts, statisticians, or engineers.

Traditional enterprise data analytics has a problem: you can’t learn anything from it. Instead, you just understand what it contains. To learn from data, you have to bring machine learning into the equation and effective feedback loops from all relevant stakeholders.

Machine learning helps surface hidden patterns in the data, especially if there are non-linear relationships that aren’t easily identifiable to humans. Active collaboration at the data layer allows transparency into the building of models and analytics metrics. It also makes it easier for biases to be identified and corrected in real-time.

Data fluency (federated AI) and data fluency are also designed to address the barriers that prevent data acquisition. These obstacles, while not always technological, can include privacy and trust, regulatory compliance, intellectual property, and privacy. This is particularly true in healthcare where patients and customers expect privacy and organizations want to protect the data’s value. They are required to comply with regulatory laws like HIPAA in the United States or the GDPR in the Eurozone.

Access to healthcare information is very difficult and protected behind strict security walls. Access is granted to de-identified information with multiple security precautions. Federated AI and data fluency principles can share a model, but not the data used to train it. It will be crucial in understanding the insights from distributed data silos and navigating around compliance barriers.

Privacy-preserving solutions to unlocking the health data’s value are essential for the future of healthcare. The point is to improve healthcare machine learning adoption and understandability to drive actionable insights and better health outcomes. Federated AI goes well beyond traditional enterprise analysis to create a machinelearning environment for data explanation and data fluency that enables parallel training from automated multiomics pipelines.

Twitter: @SiwickiHealthIT
Email the writer: bsiwicki@himss.org
Healthcare IT News can be found in the HIMSS Media publications.

Leave a Reply

Your email address will not be published. Required fields are marked *