fbpx

Procedure Step Detection… On Steroids

Can we skip to the good part?

Have you ever recorded an entire concert with your phone? Or a fireworks show? No one goes back to look at the whole thing, you want to fast-forward to the meaningful moments. Fast-forward to your favorite part or when something interesting happens. 

It’s the same with surgery. Surgeons don’t have the time to go back and review hours of surgery after scrubbing out from a long day in the operating room.

The two longest surgical videos in our database are (hours:minutes:seconds):

    • Robotic anterior resection: 9:52:28
    • Robotic whipple: 8:30:19

Even procedures this long, are automatically structured. With more than 3,000 procedures captured by Theator every week at a median duration of 42 minutes, that is a lot of data. A lot. 

The video is yours, we just make sense of it

Surgical videos need to be structured to be useful. Not just with tools and time, but with all procedure steps, intraoperative events, surgical practices, and more. Structuring with procedure steps is the first step to extracting value from surgical video capture. A deep tech stack enables this to be possible and importantly, enables scalability. 

When a procedure is structured, surgeons can easily navigate to meaningful moments, trainees can quickly review interesting clips, surgical teams can quickly get answers to questions (read how one hospital confirmed no foreign body was left in a patient using quick video review), and so much more. 

But, don’t just take it from us. A few abstracts and publications examples, written by our customers, have shown the ability of our AI tech to automate procedure step annotations (which, is just one element we structure). Check them out for yourselves below:

Urology

Gynecology

    • Hysterectomy – published in the International Journal of Gynecology and Obstetrics 

General Surgery

Want to check these out all in one place? Download the cheat sheet here.

Do Patients Want AI in the Operating Room?

Do Patients Want AI in the Operating Room?

Short answer: Yes! 

 

In fact, not only do patients place a high importance on choosing where to get surgery based on their innovative nature and adoption of technology, but half of patients believe they will receive better care if that is the case.

 

Much of the literature about implementing AI technology in the OR views the issue from the perspective of the physician or healthcare organization. And while patient outcomes are a key force driving this discussion, patient opinions are rarely part of the conversation. So how do patients feel about the use of AI technology during their own surgical procedures? The answer may surprise you. 

Surgery Is Already A Difficult Decision For Patients

Unsurprisingly, patients do not take the decision to undergo surgery lightly. Furthermore, most have understandable concerns about surgical risks and complications. In our 2023 survey of 200 post-operative patients, 82% of patients reported they were nervous to undergo surgery. Specifically, 65% worried about dying on the table and 66% were concerned their surgery may create more health problems. 

 

 

In fact, our data show that patient preoperative anxiety is not misplaced. While 93% of those surveyed were pleased with their surgical outcomes, 26% of this group experienced post-surgical issues that required further intervention to correct.

Choosing Surgeons and Hospitals

What factors go into a patient’s decision about where to have surgery and whom to trust with their care? A vast majority of patients (87%) trusted their surgeon prior to undergoing their procedures. This trust largely carried through to the post-operative phase, even for those who experienced complications they believed could have been prevented. Interestingly, 21% of patients felt their surgeon did not clearly explain what to expect, yet clearly still trusted them.


In terms of surgical location, access to the most advanced technologies was the most important factor patients considered when deciding where to undergo surgery. This outranked consideration of the hospital where a physician practiced and where the surgeon was trained. Patients see the value in innovative care and are willing to forgo institutional loyalty to access advanced technologies.  

Intraoperative Video Recording and AI

According to our results, a majority of patients (55%) wished their surgical procedure had been recorded and placed high value on the potential uses of recorded surgical data. Of the patients surveyed, 60% would want to watch the recording of their surgical procedure if it were available. Additionally, 38% wish they knew more about what happened in the OR during their procedure and 58% of those who experienced complications (15% of total respondents) felt their complications were avoidable, which may explain their interest in viewing their own procedure. 

Notably, patients are not simply interested in the potential for recorded surgical video to demonstrate what took place during their own procedure — they see the value in using recorded surgical data for quality improvement. In this regard, 76% of respondents believed that surgical video should be captured and used to better understand surgery. 

When asked about the use of AI tools in the operating room, 50% of those surveyed felt they would receive better care if advanced technology like AI was used in the operating room. This means that when patients are selecting a hospital for their surgical procedure, many see the use of AI as exactly the type of high-value advanced technology that would warrant choosing one hospital over another. Rather than being wary of these tools, patients see the potential for them to support surgical decision making.

Surgical Expectations and Outcomes

A majority of respondents believed that surgeons should be measured on complication and readmission rates, as well as length of stay. Of note, 72% of patients surveyed experienced a length of stay that was expected or less than expected. 

When asked how their surgical experience could have been improved, patients cited the following:

    • Knowing my surgeon is using the most advanced technology (47%)
    • Knowing my surgeon is being assessed based on their postoperative outcomes rather than cost (46%)
    • Having more visibility into what happened while I was asleep in the operating room (41%)
    • Being at a hospital known for having the most advanced technology (32%)

Insights for Future Decision Making

Patients are consumers of healthcare and have high expectations for the level of surgical care they receive. Rather than being viewed as risky, technological innovations in surgical care, such as the use of AI tools, are viewed by the general public as not only desirable, but a feature worth seeking out when deciding where to obtain care.

As physicians and hospital administrators make decisions about how and when to integrate AI tools into their ORs, it is essential to keep this patient perspective in mind. Just as those in the healthcare sector see the potential for AI to improve patient care, so do patients, whose interest in advanced surgical technologies is likely to grow as these tools become more widely available.

Surgical Outcome Reports Aren’t Enough

Improving The Modern Surgical Outcomes Report

In recent decades as evidence-based medicine became the standard of care, physicians and hospital systems turned to outcomes reporting as a means of indirectly measuring quality of care. Such information, especially when it is publicly available, allows for the transparency necessary to assuage patient concerns about the quality and safety of medical care and provides a basis for identifying systemic and individual opportunities for improvement. 

No other area of medicine lends itself to outcomes reporting and tracking quite as well as surgery. The universal nature of surgical procedures allows for standardization of outcome metrics, and the inability of patients to provide an account of surgical occurrences reinforces the need for maximal transparency to monitor quality.

While outcomes reporting plays an undeniable role in improving care, as we enter the era of surgical AI augmentation, the time has come to ask ourselves whether outcomes reporting is enough. With so much data now at our fingertips, how can it be harnessed to increase transparency and further improve outcomes?

The Actual Impact of Reporting Surgical Outcomes

The use case for surgical outcomes reporting came from cardiothoracic surgeons in the UK. This formal reporting process was born out of public pressure related to high mortality rates of pediatric patients in the region undergoing cardiothoracic surgery. With all pediatric cardiac units required to participate in outcomes reporting, mortality decreased 75% between 1985 and 2002. 

Following this, the Society for Cardiothoracic Surgery in Great Britain and Ireland began publishing hospital- and physician-specific data in 2005, resulting in a 50% reduction in risk-adjusted mortality. Since 2005, seven additional surgical specialties in the UK started publishing their outcomes data.

This practice has a number of advantages, most especially for patients for whom the transparency of outcomes reports provides trust and autonomy in their healthcare decision making. Surgeons also benefit from outcomes tracking, which allows identification of hospitals and surgeons with the best outcomes who can, in turn, inform best surgical practices. Additionally, publishing outcomes data provides a transparent piece of the complex puzzle that is healthcare costs.

Surgical Outcome Reports Don't Tell The Whole Story

Without an automated means of capturing outcomes data, continued dependence on human reporting will be inherently suboptimal due to incomplete recall and documentation. 

 

Related to this is the fact that outcomes alone do not take into account patient-specific pre-surgical risk and other factors that affect the complexity of a case. In other words, the intraoperative death of an 85-year-old patient with multiple comorbidities is clinically different from that of a 25-year-old previously healthy patient undergoing the same procedure. While some outcomes reporting systems attempt to account for this risk difference with adjusted scores, there is no perfect way of representing these factors

 

In addition, outcomes reports are inherently inaccurate to a certain extent due to their dependence on retrospective reporting. One study found that routine reporting by surgeons identified only 62% of adverse outcomes and even medical chart review identified only 78%. 

Without an automated means of capturing outcomes data, continued dependence on human reporting will be inherently suboptimal due to incomplete recall and documentation. 

How Do We Go Beyond Surgical Outcomes Reporting?

Outcomes reporting provides an important level of professional and public transparency that can build patient trust and potentially result in practice improvement. However, the latter is not a guaranteed result of outcomes reporting alone. One study that compared 263 hospitals that participated in The American College of Surgeons National Surgical Quality Improvement Program (ACS-NSQIP) to 526 that did not found no statistically significant improvement in outcomes across a 3-year period at participating hospitals, leading the authors to conclude that feedback on outcomes alone is not enough to prompt improvement.

 

For outcomes to serve as meaningful tools for improvement, they must be associated with additional information that tells the story behind how and why an outcome occurred. For surgeons, this data lies mainly in the events that occur in the OR during a surgical procedure. Intraoperative video recording offers a goldmine of data that, when paired with information about surgical outcomes, can provide the key to translating these outcomes into improved surgical practices. 

 

The concept of recording all surgical procedures as standard practice has understandably been met with trepidation by some surgeons. After all, as Alexander Langerman, MD states, “no one wants to fall below the bell curve.” This, coupled with the litigious nature of today’s healthcare system, can lead surgeons to disproportionately focus on the risks of transparency while disregarding the benefits. 

 

However, the benefits of intraoperative video analysis are well documented. For example, one study that used AI to analyze hundreds of intraoperative hysterectomy videos in search of practice differences to explain variability in outcomes found that not all surgeons were performing the recommended step of identifying the ureters during the procedure. This critical part of a hysterectomy procedure is key to preventing ureter injuries, and yet it was not being consistently performed. 

 

This is only one example of how AI analysis of intraoperative video can serve as an essential component of the quality improvement process. AI offers the ability to review and analyze surgical video in a manner that is impossible for surgeons to do manually due to time constraints. Without this tool and the transparency it offers, the reasons behind variability in surgical outcomes cannot be easily identified, and the practice modifications needed to improve these outcomes will remain a mystery.

Taking Surgical Transparency to the Next Level

Digital tools, including AI technologies that analyze intraoperative surgical video, are no longer the wave of the future. Instead, they are already present in ORs and being used by surgeons who see the value they offer in increasing transparency, identifying practice variability, and improving outcomes. The takeaway in the dawning of this new era in which medical care and technology are more intertwined than ever is that increasing transparency by embracing these tools will be a defining characteristic of surgeons who strive to continuously improve, and those who do not embrace these technologies will be left behind. 

Will AI Replace Surgeons?

AI tools have become a hot topic of conversation in healthcare over the past year, and with good reason. Although patients are unlikely to have a visit with Dr. Chatbot anytime soon, AI technologies are now used on a daily basis by a variety of clinicians, including surgeons. This means it’s likely the role of AI in surgical care will grow, leading many surgeons to ask, “Is AI going to replace me?”

 

The short answer to this question is, “No, but…” Let’s dive in to discuss who/what is more likely and less likely to be replaced by AI technologies.

What Can AI Do That Surgeons Can’t?

Although the adoption of digital technologies, including AI, in the healthcare industry lags behind most other professional sectors, currently available AI tools are able to perform certain tasks that are not feasible for physicians, usually because of time constraints.

The crux of all AI technologies, which makes them so appealing, is their ability to process voluminous amounts of data in a short time period and generate useful output about this data. This is possible because these tools are trained on large datasets, which allows them to create algorithms. 

 

There is no shortage of available data in healthcare. In fact, the situation is quite the opposite- we have more data than time to process and make use of it.  This is what AI can do that doctors can’t. It has the ability to mine large quantities of data and generate patient-specific recommendations for management, such as individualized preoperative risk stratification.

What Can AI Do That Surgeons Don’t Want To?

While the idea of AI taking over tasks that physicians currently perform may sound ominous, there are a number of instances in which this is likely to be a welcome change. 

One such example is the use of AI technologies for clinical documentation. Surgeons spend a substantial amount of time on clinical documentation, often outside of normal working hours. Not only do these hours spent on clinical documentation take surgeons away from direct patient care, they also contribute to physician burnout

 

Imagine an AI tool that can pair with intraoperative video recording already in use and take what it sees to generate an accurate operative report. Such technology already exists, and its widespread use could greatly reduce the administrative burden for surgeons and, therefore, decrease burnout. 

How Can AI Augment Surgical Care?

Rather than viewing AI as a tool that will replace the work of surgeons, it makes more sense to think of it as a means of augmenting the care that surgeons currently provide, like an additional surgical instrument. 

 

AI technologies being developed for use in the OR have the ability to identify critical anatomic landmarks and steps in a surgical procedure. This information can then be used for clinical decision support and improved workflow efficiency. 

For example, currently available AI tools are able to identify the Critical View of Safety during laparoscopic cholecystectomy with 84% accuracy. Such technologies have the potential to reduce errors and improve quality by encouraging best surgical practices in every case. 

 

Furthermore, these same tools that pair with intraoperative video already being recorded are able to detect surgical case steps, meaning they have the potential to offer feedback to the user about the best next steps while simultaneously providing real-time information to OR staff about the expected time remaining in a case, allowing for streamlined OR workflow.

 

So, will AI replace surgeons? It’s unlikely. However, given that AI tools have the ability to improve patient safety, clinical outcomes, quality of care, and clinical documentation, according to AMA President Jesse Ehrenfeld, MD, “Physicians who use AI will replace those who don’t.”

 “Physicians who use AI will replace those who don’t.”

This sentiment is echoed by Keith Horvath, MD, AAMC’s senior director of clinical transformation, who notes that “AI is not going to replace physicians, but physicians who use AI are going to replace physicians who don’t, and that may be the cautionary tale.” 

 

AI’s inability to quickly evolve or work in a non-linear fashion means that these tools are unable to match the critical thinking skills of surgeons, who perform their roles in this manner on a daily basis. In addition, AI will never be able to offer the key ingredient of medical care: empathy. There will always be a need for human-to-human interaction that cannot be replaced by AI.

 

However, if surgeons experience better outcomes, improved efficiency, and decreased burnout by augmenting their work with AI tools, it makes sense that experts expect those who utilize them will be replaced by those who do not. The key for surgeons in this new era of medical care is to learn and adapt to change, as the benefits AI promises for patients and surgeons alike mean that it is here to stay.

The Role of Artificial Intelligence in Surgical Care

AI in Surgical Care

In recent years, novel AI technologies have become increasingly available in the healthcare industry. From clinical decision support to documentation to diagnostic radiology, one thing is certain: AI in medicine is here to stay

While this may sound like a daunting assertion, the benefits that emerging AI technologies can offer to patient care and clinician workflow have the potential to improve outcome measures in ways that are not currently possible. As you can imagine, one area that is primed for process and safety optimization using AI tools is surgery. 

You may be wondering what technologies are waiting in the wings to improve surgical care. And what are the implications of these tools? Read on to find out!

Emerging Surgical AI Technologies

Although there are numerous specific new and emerging surgical AI tools, many can be grouped into broad categories based on their intended function. 

AI Tools For Workflow Optimization

Developing and maintaining streamlined workflows is essential to OR efficiency, and ultimately to providing timely and appropriate surgical care. Tools that predict the duration of a surgical case based on real time data from video streams or surgical instruments can assist with reducing OR turnaround time. 

Another area of workflow optimization using AI includes OR scheduling. One hospital that used an AI tool to identify and automate release of incentivized OR block time used 336 hours more of their block time in one month than they would have without the use of AI.

AI Risk Assessment Tools

Surgeons have long embraced the use of pooled data and risk assessment tools to predict individual surgical risks for patients, an important method for ensuring high-quality care. Incorporating AI technology into risk assessment adds an additional layer of analysis that personalizes risk information beyond that which is currently available.

 

Because AI technology goes beyond data compilation and has the ability to use algorithms to make predictions, tools that are pre-programmed with large data sets can apply what they have learned from this data to individual risk prediction

 

Similar tools are already being utilized in the hospital setting. The Epic Sepsis Model, a feature of the Epic electronic medical record, is a predictive model for sepsis that was trained on data from 500,000 patient encounters to identify early signs of sepsis. A single-center study found its use as a screening tool to be associated with a 44% reduction in the odds of sepsis-related mortality.

AI For Intraoperative Guidance

Surgeries are among the highest stakes medical care provided by hospitals. As such, the ability of AI to support surgeons intraoperatively in making procedural decisions offers the opportunity to provide real time feedback to improve patient outcomes. 

For example, use of an AI tool in the form of an overlay on the surgeon’s video screen during laparoscopic cholecystectomies can guide physicians as to which areas are safe or less safe for operating. 

In orthopedic surgery, automated bone registration and tracking tools use depth cameras to identify the surface geometry of a target bone. These can then be compared to preoperative images to assess bone alignment. Such technologies offer the potential to augment physician knowledge and skill to optimize patient outcomes.

AI For Diagnostic Enhancement

Intraoperative surgical pathology is a common source of time consumption and missed diagnoses in surgical cases. What if AI could be used at this critical point of care to save time and increase diagnostic accuracy? 

Such technology already exists and has been used to accurately diagnose brain tumors in under 3 minutes. Furthermore, the same tool was able to accurately distinguish between tumor tissue and healthy tissue, thus aiding with identification of appropriate surgical margins. 

For pediatric patients, AI tools have been developed to identify acute and chronic otitis media using training data from patients presenting for myringotomy tube placement. Early data suggests this technology performs with higher diagnostic accuracy than expert otoscopic examination, a promising advancement for improved diagnosis of a common but diagnostically challenging condition. 

AI Tools For Intraoperative Video Analysis

While the capability to record intraoperative video is not new, the application of AI tools to this technology offers the potential to gain insights from intraoperative surgical video that were not previously available. 

By performing advanced assessment of intraoperative video, AI can extract information that can be used in multiple areas to improve surgical care. Utilizing AI to analyze voluminous surgical video in a manner that time and cost restraints make impossible for surgeons to do personally exponentially advances OR quality improvement practices.

For surgeons, intraoperative video capture and AI analysis offers a novel tool for linking what happens in the OR to patient outcomes. Coupling these tools together means that AI can do the work of annotating surgical footage, giving surgeons detailed insight into how their intraoperative techniques and decisions relate to later outcomes.

Surgical AI: What are the Implications?

Most experts agree that it’s unlikely AI will be replacing surgeons. Instead, AI tools being developed for use by surgeons are intended to augment and automate current processes and procedures. This has the potential to affect multiple areas of patient care and clinician workflow.

    • Quality Improvement: Intraoperative tools that assist with diagnosis and analyze surgical video aim to increase diagnostic accuracy, reduce operative time, and improve patient outcomes.
    • Cost Savings: By streamlining processes, such as OR scheduling, AI tools have the power to improve the efficiency of OR operations and optimize staffing costs. 
    • Medical Education: AI tools that offer intraoperative clinical decision support or detailed analysis of intraoperative video can provide valuable feedback and guidance for training the next generation of surgeons.

As with any new medical technology or device, adoption of AI technologies for surgical care should be subjected to the highest standards of evaluation prior to widespread adoption. Nonetheless, there are new and emerging AI tools in the surgical field poised to enhance care delivery and improve outcomes, and surgeons and hospital administrators should keep a close eye on these rapidly evolving technologies. 

How AI Can Improve Surgical Care

Although the ultimate role of artificial intelligence (AI) in surgical care is still evolving, one thing is clear: AI is here to stay. The potential advantages AI tools offer to process large amounts of data and generate meaningful information that can improve patient outcomes means that its integration into surgical care has myriad potential benefits. Read on to learn what they are!

How Can AI Tools Be Integrated Into the OR?

To understand how AI technology can be used to improve patient care, it’s important to first understand how AI tools can be incorporated into current surgical workflows. Because the power of AI rests in its ability to rapidly process large data sets and produce reasoned output, data that is currently routinely being captured in ORs in the form of intraoperative video recordings can serve as the input for AI tools. These specially trained programs can then process this input and generate feedback for surgeons and OR staff to improve surgical workflow, technique, and quality

The first step in generating useful information from surgical video recordings is teaching these AI tools to accurately identify the steps of a surgical procedure. Currently available technologies have already demonstrated the ability to perform this task in different surgeries. One study that analyzed the recognition of surgical steps in recordings of 619 totally extraperitoneal (TEP) inguinal hernia repairs found per-step identification accuracy to be as high as 94%. Another study demonstrated successful real-time annotation of two separate urologic procedures.

What’s Next for Surgical AI Technology?

Now that we know AI tools can identify surgical steps, how does this information benefit patient care? Here are just a few of the multiple ways this data can be harnessed.

Workflow Optimization

By accurately identifying surgical steps and segments of surgical procedures, AI tools can provide real time predictions of factors that affect OR workflow efficiency, such as the amount of time remaining in a case. 

Accurate prediction of such metrics allows staff and anesthesiologists to better plan patient flow and make the best use of daily OR time, an especially valuable proposition in the dynamic environment of busy ORs. In difficult or complicated cases, AI technologies may someday even have the ability to trigger an alert for extra support or surgical backup based on variations noted in surgical segment identification.

Landmark Identification

Correct identification of anatomical landmarks during surgery is key to preventing medical errors. Just as AI tools are able to accurately recognize surgical segments, they can also assist with identification of critical anatomical structures. By doing so, this augmented technology can aid with reducing misidentification and enhance patient safety.

One example of this technology in action is the use of an AI tool to identify the Critical View of Safety during laparoscopic cholecystectomy, a widely agreed upon surgical best practice. After training on 2,000 surgical videos, this technology was able to achieve 84% accuracy in detecting the Critical View of Safety.

Clinical Decision Support​

A natural extension of the ability of AI tools to mark surgical segments and identify landmarks is the processing of this data to offer real time intraoperative feedback and guidance to surgeons. Based on anatomical data, AI technology may be able to provide updated predictions of surgical risk that augment those done preoperatively based on patient characteristics and risk factors alone.

Furthermore, the ability of AI tools to predict next surgical steps means they can offer suggestions on how to proceed based on their previously programmed data. This feature may be especially valuable for trainees who require additional guidance until they reach a sufficient experience level.

Operative Report Creation ​

Operative reports are a necessary part of surgical care. However, they are also a task that takes physicians away from patient care, and they do not always accurately reflect each step that took place during a procedure. This often makes them less than useful as both a medical-legal document and a learning tool.

Given the ability of AI tools to track surgical segments, it follows that these documented steps can then be compiled into an accurate and timely operative report. One study of this technology found that an AI tool reproduced major components of operative reports 91% of the time across 117 cases. This holds great promise for both reducing physician administrative burden and improving surgical transparency.

The Future of Surgical AI Tools

By processing and analyzing intraoperative video recordings, AI tools can assist in detection of  anatomical landmarks, achievement of surgical best practices, and real time clinical decision making, operative report generation, and maintenance of an optimal workflow. However, this is just the tip of the iceberg in terms of what AI is likely to bring to future surgical care. By augmenting the knowledge and experience of trained physicians with the power of large data sets, technologies such as these are poised to improve quality of care, patient safety, and clinical outcomes. 

The Role of AI in Quality & Safety Improvement​

The future of AI-assisted quality improvement

In the matter of a few years, artificial intelligence (AI) has gone from being a topic discussed only by select innovators in the healthcare industry to one of widespread interest and endless discussion. In fact, one study found that the number of publications about AI doubled in the medical literature between 2014 and 2018.

However, it’s always important to remember that quantity does not equal quality. In other words, AI tools are only worthwhile in healthcare if they improve patient outcomes. So what currently available AI technologies improve quality and safety of care? And what does the future of AI-assisted quality improvement look like? Let’s dive in!

Predictive Analytics

While patient safety techniques have traditionally focused on identifying events and near misses after the fact and working to prevent future repeat occurrences, the introduction of AI-powered predict ive analytics flips this concept on its side by allowing identification of potential patient safety issues before they happen.

One example of this is an AI tool used at NYU’s Langone Medical Center to predict hospital readmission rates, a vital patient safety and care quality measure. This technology is able to predict 80% of readmissions and performs 5% better than standard computer tools at calculating readmission risk. 

In the surgical realm, predictive analytic tools can improve patient outcomes by triaging surgical patients to the most appropriate postoperative location more reliably than traditional methods*. This can lead to significant improvements in patient safety, as patient undertriaging to a surgical floor instead of an ICU is associated with a longer length of stay and higher mortality rate, among other poor outcomes.

Clinical Decision Support

AI technologies have the power to offer clinical decision support in a manner not previously possible: at the time of decision making, using personalized and up-to-date patient data. Furthermore, AI tools that support clinical decision making have expanded in availability from a few select specialties to nearly all fields of medicine.

Radiology was an easy early target for AI technologies given the technological nature of the field. And while it’s unlikely that AI will ever replace radiologists, use of AI-powered diagnostic tools to augment radiologist readings has been associated with reduced diagnostic errors, with one study showing a 19% reduction of this metric. This not only improves quality by increasing diagnostic accuracy, it also saves money by reducing the cost of unnecessary tests. 

In surgical care, AI tools that offer decision support using real time intraoperative data can improve patient safety by assisting with the identification of critical landmarks for injury prevention. One study evaluating the use of AI in detecting the Critical View of Safety in laparoscopic cholecystectomy found this method to have an 84% accuracy rate. Similar intraoperative AI-guided landmark identification was evaluated during endoscopic hysterectomy and also associated with improved safety awareness and reduced intraoperative complications.

Medical Education

Creating a safe environment for trainees to gain necessary skills and knowledge for practicing medicine in the real world is an area where many have tried to innovate over the years. From simulation labs with mannequins to online fictional patient case scenarios, computer-assisted technology has undoubtedly enhanced the ability of students and residents to learn without putting patients in danger. 

AI tools take this concept to the next level by providing guidance and feedback using real patient data, rather than simulations, for enhanced learning. For example, AI virtual patient tools analyze large data sets from actual patients to create true-to-real-life cases for medical students to work through. This not only improves the quality of their training, it also saves valuable time previously spent manually generating such mock cases.

For surgical trainees, harnessing intraoperative recordings and pairing them with AI technology offers a new method of learning and feedback on surgical performance. Identifying critical procedural elements and decision points using AI allows surgical residents to learn from the cases they perform after the fact and use this information to improve the quality of their technique and decision making for future cases.

Improving Quality and Safety Into the Future

The promise of AI for improving healthcare quality and safety lies in its ability to operate in real time, using current patient data to provide assistance and support. This greatly contrasts with traditional models of quality and safety improvement that rely on manual reporting of data with analysis after events have already occurred. Therefore, AI has the power to improve patient care and outcomes in real time, rather than simply using past data to prevent future mistakes.

 

* Loftus TJ, Ruppert MM, Ozrazgat-Baslanti T, et al. Association of Postoperative Undertriage to Hospital Wards With Mortality and Morbidity. JAMA Netw Open. 2021;4(11):e2131669. doi:10.1001/jamanetworkopen.2021.31669

 

I. Levin, Y. Gil, A. Cohen, 7722 Improved Safety Awareness and Intraoperative Complication Reduction after Implementation of Artificial Intelligence in Hysterectomies, Journal of Minimally Invasive Gynecology, Volume 29, Issue 11, Supplement, 2022, Page S101,ISSN 1553-4650, https://doi.org/10.1016/j.jmig.2022.09.325.

The Role of Artificial Intelligence in Healthcare

AI's Role in Healthcare

The release of ChatGPT’s artificial intelligence took the world by storm in 2022, leaving most casual users amazed at the type of content it could produce. Want to build a customized workout plan? Check. Want to rewrite your resume? Check. A list of topics for a dinner party or work lunch? Check and check. While it’s clearly impressive that ChatGPT has knowledge on an unfathomable number of topics, its primary awe-inspiring feature is its ability to rapidly create novel content that generally reads as if it was written by a human.

 

This revelation set off ongoing discussions in nearly every field about the opportunities, threats, risks, and benefits of AI technology. Healthcare is no exception to this discussion, especially given the high-stakes nature of patient care. So how exactly can we expect AI to be used (and not used) in the healthcare industry?

What is Artificial Intelligence?

Before jumping into the role of AI in healthcare, it’s important to understand what defines artificial intelligence. The original concept of AI dates back to 1956, when John McCarthy described it as the science and engineering of making intelligent machines. On a big picture level, AI refers to technology that is able to perform tasks that typically require a human level of intelligence and insight.  

All AI technologies have the same foundational mechanisms. They are programmed with sets of data to develop algorithms that allow them to quickly generate output based on pattern recognition. AI tools like ChatGPT are programmed with enormous data sets. This is why they are capable of both generating your grocery list and recommending the next book you should read. Other tools, such as those used in the healthcare industry, are programmed on more limited data sets related only to their intended use. So while the “A” in AI stands for artificial, in reality it functions more as augmented intelligence that helps humans perform all kinds of tasks.

How is AI Used in Healthcare Today?

Although the term “artificial intelligence” still has a futuristic ring to it, the truth is that AI has been used in the healthcare setting for decades. Current uses of AI in healthcare include data analysis, clinical decision support, and disease diagnosis and treatment, among others.

 

Radiologists were early adopters of AI tools, which makes sense given the technology-forward nature of their work. As of 2020, the American College of Radiology reported that 30% of radiologists had adopted AI technologies. AI tools are currently being used by radiologists to detect intracranial aneurysms and pulmonary embolisms. Furthermore, they supplement routine radiologist workflows by tracing tumors and measuring the amount of fat and muscle on a CT.  

Another example of the current role of AI tools in healthcare is the use of natural language processing in clinical documentation. Natural language processing describes the way in which technologies like ChatGPT can interpret typical human language input to generate meaningful output. Tools like Nuance’s Dragon Ambient eXperience are able to transcribe a patient/clinician interaction and use this information to generate appropriate electronic clinical documentation.

What’s the Future of AI in Healthcare?

If you’ve ever used Siri on your iPhone, had Netflix suggest movies you may like, or used Google Maps to get to your destination, then you already know that AI technology is here to stay, and the healthcare sector is no exception to this. So what does this mean for the future of medical care and those who provide it?

Most experts agree that AI will not replace doctors or other healthcare professionals, and it’s unlikely that patients will be scheduling visits with a ChatGPT-like bot anytime soon. Instead, AI technology will be used to enhance processes and workflows, improve quality, and assist with making sense of the massive sets of patient data that exist in healthcare organizations. 

While any new technology used to provide patient care requires meticulous vetting and consideration of its ethical implications prior to widespread use, the benefits that high-quality AI tools can offer in the healthcare industry have the potential to substantially improve care delivery and reduce costs and administrative burden.

Cost Reduction:

It’s no secret that the US spends more money on healthcare than other economically similar countries. AI technologies that automate, streamline, or improve processes can reduce healthcare costs. For example, one healthcare system noted a savings of $3 to $4 per visit when they changed to an automated scheduling system.

Improved Patient Care:

AI-powered patient monitoring tools offer the ability to not only monitor metrics such as vitals signs, but also to take that data to the next level by looking for patterns that may indicate an impending medical emergency. Such tools are being developed for use both in the hospital setting and for home monitoring of patients. 

Reduced Physician Burnout:

Burnout notably affects a significant number of doctors, nurses, and other healthcare professionals, which has the downstream effect of growing numbers of healthcare workers leaving their jobs. Therefore, AI tools that can alleviate pain points that contribute to burnout, such as time spent on clinical documentation, can serve to reduce this threat to the healthcare workforce. For example, an AI tool that collects patient health information in advance of a doctor’s visit and automatically generates clinical documentation was shown to reduce intake and documentation time by 90%.

Enhanced Quality & Safety:

AI’s ability to quickly analyze large sets of data leads to important implications for patient safety and quality of care. Examples of this include AI tools that accurately predict which patients are developing hospital acquired infections and others that monitor hand hygiene practices and provide reminders to clinicians to improve compliance.

AI in Healthcare: The Future is Now

Current trends in the development and implementation of AI in the healthcare setting all point in the same direction: AI is here to stay. AI tools offer the potential to address some of the most pressing concerns in today’s healthcare industry, including rising costs, physician burnout, quality, and patient safety. In particular, one area of healthcare that is ripe for disruption with AI is surgical care

While the adoption of AI technologies in healthcare should be held to the same standards the industry uses for other elements of patient care, such as drugs and medical devices, it’s clear that organizations that are slow to adopt AI as it becomes mainstream may be left significantly behind their competitors. Therefore, the time is now for healthcare leaders to explore evolving technologies and the potential solutions they may offer. 

From Concept to Creation: The Path to Surgical Intelligence

Challenging the status quo of surgical care

Bringing a new idea to life is all about the four ‘P’s: purpose, persistence, (the right) people, and a solid plan. At Theator, our purpose was clear — challenge the status quo of surgical care. And by applying these principles in pursuit of that goal, we developed a first-of-its-kind product, transforming how surgery is performed today and improving quality of care for generations to come. Here’s what it looked like – from concept to creation. 

We started with the end user (surgeon) in mind

Any great product solves an unmet market need, even if the end user is unaware they have one. Theator was founded based on the realization that there is significant variability in surgical outcomes today — not only in different parts of the world but even within the same hospital. We ultimately knew what needed to be solved, now it was a matter of how to best solve it

Theator Technology Advisor and former Netflix Chief Product Officer, Neil Hunt, emphasized the importance of understanding our end users’ real habits, needs, and motivators. We needed to ask the right questions, in a non-leading way, to gather unbiased input that would illuminate our path forward. For example, “what frustrates you about being a surgeon? or “what do you hope to accomplish within the next 2-3 years?”. Based on their responses we discovered there’s a world of untapped surgical data, that if unlocked and leveraged properly, could aid in providing the highest level of care without the extra burden. 

We focused on the features that add value

Surgical video contains a ton of invaluable data, yet surprisingly it isn’t being routinely captured today. 

So we started with the low-hanging fruit: enable touchless, automatic video recording for every procedure where a camera already exists. This includes all laparoscopic and robotic procedures. Next, you have to structure the data in a way that simply highlights only the moments that matter. Otherwise you’re just left with a mountain of raw footage and manual, time-intensive work. And the last thing we wanted was to add to a surgeon’s overflowing workload. Thanks to our advanced AI and computer-vision technology, this information is automatically presented to you in an easily digestible format. But it doesn’t end there. You still need to be able to link specific actions and techniques to outcomes. We are able to connect the dots along the entire patient’s journey (pre-operative, intra-operative, and post-operative) in order to, for the first time ever, draw actionable insights and disseminate best practices. This is Surgical Intelligence in a nutshell.

We knew our tech needed to standout from the rest

In order to deliver on all the above, we needed superior AI-technology. Conventional AI-capabilities were not suitable for delivering both real-time and accurate results that could scale across multiple surgical specialties, so we set out to do something different. 

Our first aim was to develop an algorithm that could accurately detect, structure and annotate procedures in real-time. This was achieved through our proprietary VTN approach, which allows for immediate, end-to-end action recognition within the surgical domain. This is critical given that conventional technologies are only capable of capturing individual frames or brief clips, lacking long-term context of what happened before or after. 

Next, we wanted to quickly scale this across the majority of procedures and specialties. So we became the first to apply a proven machine learning concept, called Transfer Learning, to surgery. This enables us to train new procedures, using less videos. The more information the algorithm is fed, the more knowledgeable it becomes, and the more transferable that knowledge is. Then, applying multi-task functionality enables simultaneous predictions across these various procedures and specialties. So not only can it detect what’s happening, it can also predict what’s going to happen.

An algorithm that is fast, accurate and scalable provides real-time functionality which you can trust. Theator is proven to be the only company that can do this across hundreds of procedures, spanning five specialties and rapidly growing!

Never stop improving.

Our product team’s work is never done. We continuously challenge ourselves to test, optimize, refine, and re-evaluate. Is our solution still aligned with our users’ needs? What is the usability of our application? This ensures we continue to deliver a high-quality, high-value product versus just flashy features. This is only the beginning for Surgical Intelligence and we’re excited for what’s next – stay tuned!