- Healthcare organizations are faced with repositories of unstructured data, waiting to be used for valuable clinical insights. However, many organizations lack the health IT infrastructure tools needed to take advantage of that data.
Tools such as IBM Watson Health are currently being deployed, using artificial intelligence to gain much needed insight into patient health. Watson Health discovers trends and applies them to patients for better and more accurate diagnoses, as well as a better understanding of long term conditions and symptoms.
Watson Health is an advanced tool, and it’s important for healthcare organizations to understand how it is deployed and what IT infrastructure requirements it has.
“Any organization that wants to speed up the ability to turn their health data into critical insights would benefit from Watson,” Watson Health Chief Security Officer Carl Kraenzel told HITInfrastructure.com. “Today, healthcare organizations have access to unprecedented volumes and variety of high quality data, but as much as 80 percent of all data is only minimally usable by the computers and systems that store it. That data is called dark data — also known as unstructured data that is ‘hidden’ from systems and provides little insight.”
Deploying a cognitive system like Watson allows organizations to take advantage of all the data they collect and discover new ways to collect data.
For example, Watson for Clinical Trial Matching finds patients who are eligible for certain clinical trials to participate in. Healthcare organizations have trouble finding patients to participate in trials, with less than 5 percent of cancer patients participating in trials, according to Kraenzel. Matching patients with clinical trials will significantly widen the pool of results.
Organizations looking to deploy Watson need to consider how it is deployed and what the requirements are. Watson is deployed as a secure medical-grade cloud service, which simplifies the deployment process than having to deploy it on-premise. Organizations need to update their existing proxy servers and LDAP servers to apply SSL/TLS and SAML to link Watson into their end-user application.
Entities are also faced with training Watson to answer questions on the subject in which they are applying Watson.
“The system uses natural language processing and machine learning to read through massive volumes of documents to extract and organize information about a particular topic and then refines its understanding of that subject based on human feedback,” Kraenzel explained. “The amount of time it takes to train Watson varies depends on how complicated the domain is and how much data the system has access to.”
Watson is also compatible with other health IT systems organizations already have in place. Watson has APIs, connectors, and gateways that are used to connect it with EHRs, data repositories, and workflow systems.
Organizations that may not be able to deploy Watson right away, but are interested in adopting it in the future can take steps to prepare their health IT infrastructure.
“The first step for every organization is to refresh their interpretation of data security policies and associated procurement and compliance practices,” Kraenzel advised. “When getting started, typically organizations discover they have older ‘20th century’ style interpretations that make it hard to adopt cloud. That would then also inhibit any adoption of Watson, along with so many other critical industry innovations happening today.”
“Empower and challenge your compliance leadership to adopt a freshened 21st century interpretation to regulations and policies, with risk assessments and control equivalence and contractual structures that make it easy for your IT teams to send PHI into a fully encrypted medical grade cloud,” Kraenzel continued. “This makes it easy for your procurement teams to rent without trying to rewrite a simple SaaS contract into a complex expensive outsourcing agreement. There are many organizations that have already made this shift, and as such they are getting a crucial head start over their less agile competitors on critical innovations like Watson.”
Organizations also need to consider where their unstructured data is and how it is currently being stored.
“Organizations should consider where there are large repositories of unstructured data (text or images) in their business, and estimate what benefit would come if their experts could somehow magically read all that information to stay current on it,” said Kraenzel. “Those are candidate settings to plug in Watson capability, as an assistant to the human experts, an advisor that stays current on the content for them.
“Before starting with Watson, it's important to have a view of the potential value that could be unleashed from that data, plus some thought about the processes and costs to keep that data clean and updated ongoing. If the expense to curate that data exceeds your possible value, you won't gain any benefit by adding Watson there.”
As healthcare organizations continue to collect unstructured data, tools like Watson are necessary to gain any insight from that data.
“We believe the power of cognitive computing can change the way we view healthcare and how we process and relay information to both patients and providers,” Kraenzel concluded. “To succeed in an increasingly complex environment like the healthcare sector, new approaches are required for value creation and outcomes-driven performance.”