The Chair of Digital Strategy for Critical Care at the prestigious Mayo Clinic in the US has cautioned that a cognitive layer must be built into Ireland’s national EHR in order to avoid swamping clinicians with irrelevant data and adversely affecting patient care.

Prof Brian Pickering, a graduate of Trinity College Dublin, has led the design and deployment of intelligent human-centred technology across acute care environments throughout the Mayo Clinic Health System.

Speaking at last week’s Irish Hospital Consultants Association’s (IHCA’s) FutureCare 2026 conference, Prof Pickering said Ireland’s EHR must be developed around clinical decision-making and not data collection.

“Over the past 20 years, we have generated more digital data than in the existence of Mayo Clinic. That digital data proliferates through vital signs, labs, imaging, et cetera, et cetera. Everything I do is captured. It’s all stored in the digital data. Yet there are many clinicians who work with EHRs, who we will tell you a very similar story – we have a lot of data, but we often don’t have a lot of insight. We’re often lost in the data and don’t get the chance to synthesise that data,” he explained.

Lived experience

“For many clinicians, the lived experience diverges from the promise. Instead of clarifying the picture, the EHR often fragments it. In EHRs, information is organised by tabs. Clinicians don’t think of tabs. The story is never really told by the EHR. You have to stitch it together yourself. You start creating more alerts, documentation expands, and your time at the bedside starts contracting.”

What emerges, he said, is a system that’s “extraordinarily good at capturing data for compliance and billing” but not so good at aiding in the delivery of care.

Ireland has the opportunity to build a system where the data is presented according to relevance rather than completeness.

Prof Pickering and colleagues previously examined the alignment between the data required by clinicians and the data collected by the EHR. The findings were startling. The study found that of the 200-plus data elements collected by Mayo’s EHR, when it came to decisions about admitting complex patients to ICU, the average clinician used just 11 data elements. Some clinicians used zero EHR data elements, and made the decision to admit based on clinical examination alone.

“High-quality clinical care depends on synthesis of information. We go to the bedside. We talk to patients. We take a history. We put everything together. We don’t reason by scanning every available data element. We build mental models with limited data sets. We continuously update those models as new information emerges and change our plan. In practice, doing this requires very little information. Our brains are optimised for relevance rather than completeness,” Prof Pickering said. 

Cognitive layer

Presenting clinicians with too much data will not deliver clinical benefit, and may end up worsening patient care.

“One of the things we found in our study was, the more [information] you threw at people, at some point it broke them. They started making worse decisions, less clear decisions. So there’s a sweet spot between nothing and everything,” Prof Pickering said.

This is why a cognitive layer should be built into Ireland’s EHR at the foundational level. 

“The first thing it has to do is organise information around clinical questions and clinical observances, not by data type. If I go into my EHR, I can see a tab that has all of the lab values in it, and then I go to a different one to look at the vital signs, and a different location for a chest X-ray. I might need pieces of information from all of those to create a medical model of the patient and make decisions. You need to organise your data around those concepts. 

“Clinicians don’t think in tables,” he said. “When everything is visible in your EHR, nothing is important. It is difficult to pull information out. That time you spend on that, that cognition that you spend, that is what gets in the way of the care of patients.”

He added that agentic and generative AI, have been transformative in this regard. “In the nearly 20 years I’ve been working in this field, these are the most powerful tools I’ve encountered. They have really transformed the way I do my job.” 

“What they add is the ability to synthesise data as a narrative. So they can tell you, Mr X has had a dropping blood pressure over the past six hours. Remember, he’s got this problem in his heart. You should maybe go see him.”

‘Ramp’ to care

The result is a hospital-wide surveillance system that continually in the background, synthesising data, so that clinicians are alerted when a patient may be beginning to deteriorate. 

“If you arrive sooner at the bedside of a patient who needs you, that’s really important,” Prof Pickering explained.

Implementing an EHR is “just the on ramp to modern health care”. The critical factor to consider is how the data is used to improve patient care, and AI is revolutionary in this regard, he said. Ireland has an opportunity to develop an EHR that can easily incorporate AI functions,

“Building an intelligence layer right up front is less expensive and less difficult to do than what we’re trying to do, which is spending billions of dollars on an EHR, and then spend billions of dollars taking all the data and putting it in a cloud, and having AI and other things run on top of it.”

“Once you see the mismatch between how clinicians think about patients and how information is presented to them, the natural question that follows is, what would a system look like if it was designed to support clinical reasoning? And that’s the opportunity that you have in Ireland right now,” he said.