8 issues every CIO needs to solve about artificial intelligence
Technology investments must be planned to take advantage of AI as healthcare organizations demand results.
Steps CIOs must take to take advantage of AI technology
While many in the healthcare industry acknowledge that artificial intelligence has potential to deliver results, success with AI is data driven. For most healthcare organizations legacy limitations on ingesting, integrating and understand data that is structured and unstructured will severely limit AI ambitions, according to MarkLogic, which markets a database for integrating data from silos. Bill Fox, global chief technology officer of healthcare and life sciences at MarkLogic, offers eight examples of why data limitations are holding back true artificial intelligence.
1. Poor data quality
If an organization doesn’t know when the data was entered, where it came from, if it was modified, when and by whom, then how can it be trusted to fuel AI initiatives?
2. Conflicts in priority
AI adoption is emerging at a time of tremendous chaos. Whether it is regulation, like the Affordable Care Act or real world evidence requirements of the Food and Drug Administration, business leaders want to implement AI to assist their efforts, while data scientists are handling fire drills and wrangling data, leading to haphazard AI attempts that fall short of producing actual business value.
Sharing sensitive data openly and securely is a complex hot-button topic. As AI innovation moves into heavily regulated industries with massive amounts of personally identifiable information and protected health information, barriers to entry could arise as gatekeepers will need serious assurance that the risk is worth the reward for them and their organizations.
4. Regulatory change
Any new security regulations that come out of the Equifax breach, or new protected health information regulations coming out of the Anthem breach, will change over time and require new questions to be asked of data.
5. Outdated infrastructure
The data integration step can’t be ignored, and if an organization is working on a decades-old architecture, or a cobbled together open source stack, it will be very difficult if not impossible to securely integrate and curate data, such as patient records, insurance claims and doctor’ notes to create a high value AI architecture. In today’s fast paced, data-driven marketplace, healthcare providers simply don’t have the luxury of wasting time and money.
6. Overstated capability
Artificial intelligence solutions are in their early days.An organization must build judiciously to adapt to AI as it rapidly evolves, rather than making investments that set an organization up for a cycle of costly IT projects.
7. Data lake reality
Healthcare organizations that have poured millions of dollars and countless hours into building a data lake are now coming to the hard truth that a data lake is not a database. A file system, no matter what is piled on top, is not going to be as agile, secure and easy to govern as a multi-model database.
8. Competing investment priorities
With data now such an integral part of healthcare business today, IT budgets generally have not kept pace with demand. chief information officers and chief data officers are forced to make decisions on which projects to fund, leaving some leading edge projects on the cutting room floor.