Of the pool of respondents we surveyed:

The Artificial Intelligence (AI) ecosystem

• 46% of respondents were in C-suite roles, while another 34% were in director or VP positions;
• 35.5% of respondents were focused on the development of analytics, while the rest were users of analytics;
• Inpatient providers made up half of respondents, with ambulatory and post-acute providers contributing another 22%, and payers representing 12%; and
• Respondents represented organizations at every scale, with mid-size organizations having the largest share at roughly 46%.

After analyzing the full results from our survey, here are three key trends and takeaways we found:

1. Leadership of analytics programs is shifting away from technology leaders, with a sharp rise in shared leadership models

Historically, the CIO or some other technology-related leader was in charge of purchasing, implementing, and managing an organization's analytics efforts. However, with the impact of digital transformation and the need for greater innovation, that picture is shifting rapidly. In our 2015 analytics survey, 43.8% of analytics programs reported to the CIO. In 2018, that figure was cut in half to 21.5%. On the other hand, shared leadership models surged, jumping from 2.1% in 2015 to 25.3% today.

What these data mean is that analytics is increasingly decentralized, and the CIO is less frequently the primary decision maker for analytics programs. The more mature an analytics program becomes, the less important it is that it reports to someone with deep technology expertise. The CIO's role is still valuable—there are plenty of business-focused, innovation-oriented CIOs out there—however there's a gradual rise in Chief Analytics Officers, Chief Digital Officers, and leaders with another intriguing new title, the Chief Performance Officer, focused on using data to improve processes.

2. New data sources continue to rise in importance, with strong interest in social determinants of health and mobile data

In our previous surveys, we've asked what data organizations currently incorporate into their analytics programs. Since 2015, the availability of clinical partner data sourced through a health information exchange or other integration channel is up dramatically, from 31% to 71%. Patient-reported data and unstructured clinical and non-clinical data increased dramatically as well, about 30% each. These investments in better data availability are vital to predictive analytics.

Get started with patient-generated health data for patient monitoring

We also asked respondents about what types of data they do not currently use, but plan to add in the next year. With this question, there were two clear winners: 48.8% of systems say they plan to add social determines of health data, and 31.7% say they're working to add mobile device or internet of things (IoT) data to their analytics programs. Although these data points are projections, we have heard through our discussions with members the need to expand their sources of data beyond the typical structured financial and clinical data that currently dominate analytic models.

3. AI's perceived value is strong, with over one-third of respondents expecting transformative value to their systems

Views of AI's potential value have improved substantially this year, with almost 37% of leaders expecting transformative value to their systems, 27% expecting some incremental value, and a small fraction expecting negligible value. There is also just under a third of respondents who are unsure. These observations are in line with our anecdotal experience that shows while there is high interest and the majority of leaders now expect real value, most do not anticipate an immediate return on investment. When evaluating responses according to organizational size, leaders at larger organizations are more likely to believe AI will be transformative than their counterparts at smaller systems. That's significant, because AI and analytics generally benefit from economies of scale, and AI is especially hungry for large data sets, so those large systems are both believers in the technology and have the resources to invest.

Of course, there are still plenty of challenges. When asked to pick their top barriers to adopting AI, respondents noted that the cost of vendor solutions (46.5%), funding (37.7%), and immaturity of technology (35.1%) were their top concerns.

Want to see the full results of our 2018 survey? Check out our webconference recording "2018 Analytics and AI Survey."