Healthcare IT in 2025: Triage for the Age of AI

The modern concept of “triage” dates to the Napoleonic Wars, when French battlefield surgeons developed a system for assessing the wounded and determining treatment priority to govern appropriate medical action in the chaos and confusion of war. It efficiently leveraged limited resources under challenging circumstances, increased the number of survivors, strengthened the army’s continued fighting capability, and transformed care delivery strategy in ways that still influence the field of medicine today.

Amid the chaos and confusion of healthcare IT in our age of AI, that triage concept may be the best guide for organizations fighting to stray strong and capable in 2025.

A challenging and dynamic landscape

Technologically, the healthcare industry now functions along a vast spectrum of capabilities and limitations. On one hand, recent research shows that more than half of healthcare organizations use AI tools for some purposes, indicating that incredibly sophisticated data-driven technologies are already being broadly deployed. Conversely, hospitals lose around $8 billion annually due to inefficient IT systems, and the whole sector is still regularly beset by the business world’s most-expensive cybersecurity crises, reflecting constrained technological resources perpetually under siege. 

The puzzle that is healthcare IT is a challenging one to solve. It almost always involves tending to data infrastructure that must be compliant and secure and operational 24/7 (under tight budgets and at significant cost), while trying to adapt it to new threats and new “must-have” capabilities surging over the horizon and beating at the doors daily. 

For example, most hospitals and healthcare organizations really only started making their push into the cloud over the past five years. They may still be trying to fulfill initial roadmaps and refine a cloud strategy, while the landscape is already shifting to multiple clouds. Say you’re a health system that transitioned to the cloud on AWS, and now Microsoft releases new Azure and Fabric tools that can integrate with your Epic EHR, and oh yeah, your research arm wants to use Google Cloud’s Healthcare Data Engine. Maybe you have different lines of business with different domains with their own tools for health plan management versus the care delivery side of the organization versus procurement and supply chain, etc. And maybe you are using Snowflake or Databricks as your go-forward data platform and they both announce open source Apache Iceberg-related tools and new suites of features supporting data lakehouses and a variety of analytics engines that could let you facilitate autonomy and simplify your data architecture. How do you manage all that sensitive and regulated data, and how do you support and architect an optimal system to use all three public clouds, as well as all those wonderful tools and features, and whatever else is going to arrive tomorrow?

And all of this barely touches on accommodating the elephant in the room — AI. 

AI’s impact

Matt Turck publishes popular “state of the union” maps of logos representing important companies in data, analytics, machine learning, and AI ecosystem every year. The very first version in 2012 had just 139 logos. Last year, there were 1,416. And this year, there are 2,011! Over just the past 12 months, there’s been a veritable arms race in AI tools and additions and enhancements for every platform imaginable — it’s near impossible to keep up with every single feature. One of them may be game changing, but it may not be the one screaming the loudest. And maybe something as utilitarian as optimizing your cloud storage and setting up a more resilient architecture to be able to handle all these other tools is really where your organization needs to focus its budget and resources. It’s not sexy, but it is really critical. 

I know I’m preaching to the choir on the weight of complexity and the continuing onslaught of innovations, but facing the new year with a triage mindset really may help clarify how your organization can best meet the moment: What will survive with or without intervention? What will make a positive difference in outcome if tended? What will perish regardless of your efforts? Here are a few examples of what triage might look like in IT practice.

Prioritize assessment: Measuring the ROI of a data initiative in healthcare IT has traditionally been an exercise that data teams either do not do or don’t do very well. This will become even more pronounced with AI because organizations are really building multiple “new muscles” and they need to understand their impact: using the new tech, understanding how to forecast its usage, and having a strong framework for measuring and validating whether it is benefiting the organization as intended or it’s a boondoggle that drains resources from the organization’s other IT priorities. Healthcare organizations cannot afford to spend another year experimenting without producing results. 

Focus on data strategy & architecture: With an increasing number of technology vendors and new capabilities for existing vendors, there needs to be a strong approach to support selecting the right tool for the right job, which likely will not come from one vendor. Open-source data lakehouses such as Iceberg offer a lynchpin in a “bring-your-own-tool” architecture, so you aren’t pigeon-holed as the landscape evolves and shifts, and the ability to have a completely agnostic data layer for multiple analytics engines is within reach. Much of the tooling to facilitate that capability has only been available to the masses for a few months, and there’s an awful lot of new activity in this area, but it’s a topic worth monitoring and analyzing against your strategy and roadmap repeatedly in the new year.

Build design thinking & automation expertise: This is an extremely desired IT skillset for healthcare — and every other industry — in 2025 and beyond. AI represents immense opportunity, but there is nothing worse than automating a process that shouldn’t exist in the first place. Teams that can rapidly grasp the current state, understand the art of the possible with new tooling, weigh build versus buy opportunities, and consistently connect that back to business outcomes in a pragmatic fashion are priceless.

Triage, at its heart, is about pragmatism in the face of havoc and uncertainty — and that is what healthcare IT really needs in the age of AI. Build a roadmap and make it pragmatic by systematically tying it to your particular organization’s circumstances, resources, and desired outcomes to chart your course. You don’t have to make all the decisions on everything all at once. You just have to know the right “next one” that you need to make and why it matters.

Photo: nevarpp, Getty Images

This post appears through the MedCity Influencers program. Anyone can publish their perspective on business and innovation in healthcare on MedCity News through MedCity Influencers. Click here to find out how.