Hospitals are drowning in data and still starving for time.

For two decades, healthcare IT has optimized for capture: more fields, more clicks, more dashboards, more “insights”. Meanwhile, the lived experience inside a hospital, especially around surgery, has become overloaded with administrative burden: more documentation pressure, more handoffs, more coordination debt, more systems to reconcile, more “just one more thing” piled onto clinicians who are already at capacity.

Hospitals suffer from coordination overload, not information scarcity. They already know what’s happening. What they lack is a scalable way to make the necessary work happen without more human effort.

If you want to understand why this matters, start in the operating room.

The OR is the most expensive clock in the hospital

Every minute in an operating room has a measurable price tag. A large, multi-hospital analysis using hospital financial data estimated the mean cost of OR time at roughly $50-$100 per minute (FY2014, inflation adjusted), with most settings landing around $71 per minute on average and high-complexity surgeries topping $140 per minute. Add to that costs of overtime and/or holiday rates and you can be pushing $150-$250 a minute. This means that even small delays of 10 minutes can cost a hospital upwards of $1,000.

That number is the core economic engine of a health system. Surgery is often the biggest revenue and cost driver, as OR access determines throughput downstream (PACU, ICU, floor beds). Delays propagate: a late start becomes a missed add-on, a bumped case, an overtime staffing bill, and a frustrated patient.

So why do so many digital transformation efforts in perioperative care still focus on generating more analytics instead of removing work? Because "insight" is easier to build than execution.

The “insights trap”: we keep measuring the mess instead of eliminating it

Hospitals have become very good at instrumenting complexity. We’ve created layers of systems that can tell you:

  • Turnover time is up.
  • First-case on-time starts are down.
  • Block utilization is below target.
  • Cancellations are increasing.
  • Documentation is incomplete.

But dashboards don’t close loops. People do. The insights trap is when we keep measuring the mess instead of eliminating it.

The underlying problem isn’t that hospitals lack awareness. It’s that they’re forced to manage too many things manually, especially at the seams:

  • between scheduling and staffing,
  • between pre-op clearance and day-of-surgery readiness,
  • between intraop events and downstream documentation/coding,
  • between what happened in the room and what gets entered into the record later.

That seam work is where efficiency goes to die. What does not fix inefficiency is more dashboards, more predictive models without execution, more alerts, more documentation, and another portal for staff to manage.

The hidden tax: administrative work consumes the clinical day

The most painful irony in modern healthcare is that digitization has made documentation more legible but not less costly. That goes for time, too.

One large-scale analysis of EHR use (based on ~100 million encounters across ~155,000 physicians) reported that physicians spent about 16 minutes per encounter in the EHR on average, with major time allocation to chart review, documentation, and ordering.

And when you look at physician work broadly, not just in-visit activity, the pattern is the same: substantial time is spent on indirect care tasks like documentation and order entry. The AMA’s 2024 survey summary, for example, reports meaningful weekly hours spent on indirect patient care and admin work.

Nursing faces a similar problem. Time-motion research has found nurses spend a meaningful portion of their time documenting (e.g., ~19% in one multi-hospital analysis). Our research at VitVio aligns with recent industry findings highlighting widespread frustration with duplicative or unproductive admin among nurses and the broader surgical staff.

Now map that onto perioperative care, where the operational tempo is faster, the handoffs are tighter, and the margin for error is smaller.

Periop doesn’t just need better information. It needs less to babysit.

Perioperative work is coordination work

Periop is a chain of dependencies:

  • pre-op evaluation and clearance,
  • patient readiness,
  • equipment, implants, blood availability,
  • staff and anesthesia coverage,
  • room readiness,
  • documentation and charges,
  • PACU capacity and bed placement.

When that chain breaks, humans become a bottleneck. They chase missing items, reconcile inconsistent statuses across systems, call the patient again, re-enter the same fact in three places, and patch gaps with heroics. That’s not inefficiency in the abstract; it’s time taken directly out of patient care.

This is why “more data” often fails. The hospital already knows what’s happening. What it doesn’t have is a scalable way to make the necessary work happen without more human effort.

AI’s real opportunity in periop: turning observation into action

AI in healthcare has been marketed as prediction: detect risk, flag deterioration, forecast schedules, find “insights.” However, periop is where prediction is only valuable if it converts into action fast, safely, and in the hospital’s existing workflow.

The most promising AI stories of the last two years are not about better dashboards. They are about eliminating clerical work. Ambient documentation tools have been associated with reductions in documentation time and burnout in real-world settings. In  inpatient settings, ambient AI scribes substantially reduced documentation time and improved documentation quality scores.

This direction matters because it reframes AI from insight generation to time liberation. Periop is arguably the highest-leverage place in the hospital to do that, because time here is literally monetized at the minute-by-minute level.

Computer vision and ambient sensing is the missing ingredient for periop automation

Ambient sensing in the operating room refers to the use of computer vision, audio, and contextual signals to automatically detect and timestamp perioperative events (e.g., induction start, procedure start, closure, room turnover) without requiring manual documentation by clinical staff.

This matters because the hard truth is that a large portion of what matters operationally in the OR is not reliably captured in structured data, let alone in real time.

Room ready? Wheels in? Induction started? Procedure start? Closure?

These events may exist in documentation, but are often late, incomplete, or inconsistently recorded, because the team’s priority is the patient, not the timestamp.

Computer vision-based OR platforms are starting to change that by detecting and timestamping perioperative events directly from the environment. For example, a 2025 BMJ Health & Care Informatics study used a camera-based AI system that segmented cases into workflow phases (induction, prep, procedure, cleanup, setup, etc.) to provide insights and analysis of variability across thousands of cases.

However, here lies the true pivot point on making valuable impact that this paper did not address:

  • Insights-only approach: “We can show you where time is being lost.”
  • Automation-first approach: “We can remove the work that causes time to be lost.”

The latter is where “fewer things to manage” becomes real.

The causal chain of ambient sensing

Ambient sensing → reliable event detection → automated documentation + workflow triggers → fewer manual handoffs → more predictable OR days with less admin → time returned to clinical teams.

What “fewer things to manage” looks like in periop

If you take this seriously, you stop designing AI as a reporting layer and start designing it as an execution layer. Software that absorbs operational chores the way autopilot absorbs routine flying. In periop, that means systems that can:

1) Convert the timeline into documentation automatically

Documentation shouldn’t be a second job after the case. It should be a byproduct of the case. When the environment itself can produce a reliable timeline (phases, key events, handoffs), you eliminate the endless “who updates what” and the “what happened” problems outright. Research into ambient documentation is already showing time savings and burnout improvements in multiple settings. The periop analog is obvious: a large portion of operative documentation is structured, repetitive, and tightly linked to what happened when.

2) Drive coding and billing accuracy as a downstream effect

When the system knows the phase transitions, devices used, and key events, it can pre-populate the “billing story” without forcing clinicians to re-enter it. This matters because the OR is not only costly; it’s also where a substantial amount of revenue is lost today.

3) Close coordination loops automatically

A delay should automatically inform the next dependent step: staffing, PACU readiness, bed planning, patient communication, downstream clinic adjustments. Humans should handle exceptions, not the baseline.

4) Reduce the number of systems staff must actively manage

Hospitals don’t need another portal. They need their existing systems to stop demanding constant attention. They need to lower their cognitive load.

AI should sit on top of the hospital’s systems of record and understand context, take routine actions, escalate only when uncertainty or risk warrants it. This is the core difference between analytics and operations.

The uncomfortable metric: “time returned” should outrank “insights delivered”

Most healthcare AI products still market themselves on detection rates, model performance, or “actionable insights.” Those metrics matter but they are not the hospital’s bottleneck.

The bottleneck is:

  • minutes of clinician attention,
  • minutes of useful OR time freed up,
  • minutes of schedule stability,
  • minutes of delay avoided,
  • minutes of after-hours documentation eliminated.

If an AI system produces ten new risk flags but creates five new tasks, it is not helping. It is creating more strain on the system and adding pressure to the team

A better north star is quite blunt:

"Does this reduce the number of things the hospital has to manage day to day?"

This philosophy is explicitly reflected in how we build at VitVio. We prioritize proactive, context-aware automation over over-informing, and use sensing in the OR to let AI agents handle routine tasks such as phase logging, calling for the next patient, and drafting operative documentation. Periop AI that does work instead of merely describing work.

The impact: safer care, faster throughput, less burnout—without heroics

When you reduce the “things to manage,” you get compounding benefits:

  • More predictable days for periop teams.
  • Fewer late starts and fewer cascading delays (because dependencies are handled earlier and more reliably).
  • Cleaner documentation and better revenue capture (because the record becomes a product of reality, not memory).
  • Less cognitive load (because staff stop acting as the glue between systems).

And yes, you get better analytics too, but as a byproduct, not the core product.

The next decade of periop systems will be won by execution, not insight

Hospitals don’t need more data that they don't have time to look at. They need fewer responsibilities, because time is expensive and coordination is both fragile and challenging.

The winning AI systems in the OR won’t be the ones that just generate insights and treat that as the product: Those will be the baseline. The winners will be the systems that quietly, relentlessly, give time back to nurses, anesthesiologists, surgeons, schedulers, the wider perioperative team, and to the patients waiting in gowns while someone hunts down a missing clearance.

The next decade of perioperative systems will be won by execution, not insight. That is what “fewer things to manage” really means, and this is the difference between a hospital that is informed and a hospital that is actually relieved.

—

Have thoughts, questions, or a perspective worth sharing? I’d love to hear from you, reach out to Thomas Knox – Founder & CEO, VitVio