Artificial Intelligence: Nurses Integral Role in Deployment

March 19, 2020

How Nurses Use Data Generate by Smart Algorithms

Yale New Haven Hospital nursing staff was one of the earliest users of the Rothman Index. The index is a tool used to reflect acuity and risk levels for patients. Clinical Informatics Manager Leslie Hutchins stated the Yale New Haven Hospital implementation of technology was aiming to provide the right advisory, at the right time, in order to pull the data that is meaningful to the cause of achieving desired patient outcomes.

The Rothman index uses electronic medical record data for calculations. 26 variables (including eleven nursing assessment metrics are put into graphs. The Rothman Index was met by a great deal of skepticism regarding validity and reliability in addition to the accuracy of actionable results. Initially, the index didn’t have enough peer-reviewed literature for nurses and clinician to back the tool. Today research has suggested that the Rothman Index performance has been impacted positively by nursing assessment data. This makes the potential for nurses to impact the care their patients receive phenomenally.

Nurse SWAT teams at YNHH use the Rothman Index for the identification of at-risk patients. SWAT teams are made up of nurses who are experienced in critical care, trauma, and cardiovascular life support. Immediate warning notifications are sent to cell phones when a patient’s index shows signs of deterioration.  SWAT will review the patient’s EMR and assess the patient. The clinical nurses, doctors, and SWAT then collaborate to determine the focal points of the patient’s care. The SWAT nurses have taken on the label of the “second set of eyes”.

The data that is used to calculate the index come from the documentation all nurses do routinely. The nurse’s input of their assessment data in a timely manner is at the critical core of calculating index scores because they derive their updates in real time from the patient’s EMR.

New algorithms will be integrated into patient care and nurses will have to garner experience in the interpretation of multiple data results to integrate new information into their daily practice. Based on their own Rothman index experience, the YNHH nurses offered their best advice on integrating new data into patient care seamlessly.

  • The mindset of the team of nurses must be open to growth. Be prepared to learn new ways to harvest and utilize patient data.
  • When integrating a tool into existing practice it is essential to offer word-of-mouth frontline stories that complement the training sessions.
  • Be sure the tools you choose to integrate are user-friendly and that the output interpretation is intuitive.
  • The main goal is that patient care benefits from the integration of any artificial intelligence tool. Nurses should be able to spend more time bedside and understand their patient’s medical condition and needs better as a result of implementation.

 

Are Robots Changing the Profession of Nursing?

Robotic engineers are working every day on the advancement of what robots are able to do and how they can now respond emotionally to certain circumstances. Robots that are able to respond with emotion are commonly referred to as social robots or robot companions. We have not reached the epic Hollywood cinematic levels of robots taking over, but they are making their way into healthcare facilities, our residences, and our work sites.

Social robots are made to interact with humans in ways that make them human-like by responding to the interactions. Sophia is a widely known social robot that was conceived to be a companion for senior citizens. Sophia is a representation of technology’s potential in improving how robots can function on a human-like level. Globally researchers are creating robots to teach driver’s training, interact live on telehealth applications, intercede when suicidal tendencies are present, and much more.

Robots are being made to perform nursing functions like taking vitals signs, ambulation support, administering medications, and learning infectious disease protocol. As this becomes a reality and robots are integrated into healthcare settings, the traditional role of the nurse will change. Research says that eight to sixteen percent of the time nurses spend on the job is allocated to activities that are not solely nursing duties and could be delegated to another member of staff. When nurses have robots to support them, they will have the control to take this time back and dedicate it directly to the patient’s care.

Since 2014, the National Science Foundation has funded nursing centered robotics grants to promote and support the utilization of robots in the healthcare setting. To date over $3 million has been invested into learning how robots can be used to perform nursing activities. This by far does not mean that nurses are going to become obsolete. Actually, the polar opposite is happening. Nurses are engaging in the creation of these robots that will one day revolutionize healthcare. Most nurses view these robots as assistants that will help them at bedside and in the community as elder support.

A collaborative team at Duke University Pratt School of Engineering and School of Nursing are working together to develop TRINA (the Tele-Robotic Intelligent Nursing Assistant). TRINA is controlled remotely to address healthcare staff that is at risk for infections because of routine nursing interactions with patients, handling material that is contaminated, and difficulties with removing any protective gear safely. TRINA is being tested in a simulation lab and performs over half of the duties that nurses perform, but is twenty times slower. There are no plans for release any time soon, but research and development have proven to be learning experiences that will assist in future endeavors.

 

Nurse Collaboration on Development of AI Applications

Microsoft CNO and National Director Molly McCarthy says AI is a tool that will significantly support nursing as a whole by making it possible for nurses to practice at the top of their license. Nurses today spend too much time on tasks that should be performed by another staff member with different training.

Research predicts that in the future AI will relieve nurses of these activities so that they can focus on activities that use all of the knowledge, training, education, and experience they have gained.

Microsoft has been applying AI technology to three of the main health care elements: operations, finances, and clinical. Working alongside nurses they are identifying the areas where AI could be mutually beneficial to both healthcare staff and the patients. Nurses are the key component to designing and developing healthcare AI technology. These are the experts that will give Microsoft the insight that is needed to deliver the power that nurses globally deserve during the industry’s transformation in this digital trek. The implementation of new ways of administering healthcare is challenging and integrating AI technology will not be the exception to the rule. Microsoft suggest that AI development teams will be more successful if they:

  • breed a mindset that is open to growth and will embrace new practices within the team. Failures must be expected as part of the process. Strong leadership of a cultural foundation is essential.
  • Identify outcomes that can meaningfully measure progress and legitimately be considered a contribution to the goal.
  • Respect clinician’s time limits. Be consistently prepared with pertinent data and answers to questions that are common.
  • Look beyond the healthcare industry to see what AI has done to advance other industries and be willing to learn from their experiences.

 

How Will AI Impact the Role of the Nurse?

 With the rapid pace that AI apps are being developed for the healthcare setting, it’s likely that you (as a nurse) may be involved in creating or testing an introductory AI system. Nurses that get recruited to be part of this process will be asked to provide your feedback to the data scientist. The goal is to garner valid data to design and train an AI system. Nurses are basically the quality control check for the data input and outputs. Nurses will evaluate if the end result makes sense. If data variables are missing, the team will work together to identify them. Nurses will answer questions such as: Are you surprised by any of the results? Are you confident in your knowledge of what the data represents and do you know how to use the results?  And the big question: Would you trust these results enough to base your treatment of your patients on them?

One of the main concerns for AI practitioners is managing the levels of bias. If an algorithm is trained with biased data, they are definitely going to return biased output. According to the Massachusetts Institute of Technology there are three areas of vulnerability in any AI Project where bias can finagle it’s way in:

  • when framing the task at hand
  • choosing data that is flawed (flaws can be that the data reflects reality that is not current or there are pre-existing biases littering the data).
  • Choosing data that have attributes that will warp the algorithm outcomes.

Provision 4 of the American Nurses Association’s Code of Ethics for Nurses states very clearly and concisely that nurses are responsible for and are accountable for all nursing practices and any impact they may have on patient care. Accountability for the judgment, choices, and actions of nurses is specifically addressed in the code. Additionally, the code specifically outlines systems and technology used as aids rather a substitution for the skill and judgment of a nurse. When AI is introduced into clinical setting, they will be (according to the code) adjunct to rather than substitutes for a nurse’s skills and abilities according to their knowledge. This makes it imperative for nurses testing or helping to build new systems responsibly ask questions about the data that is being used to train the AI and find out how the system results are getting checked for any bias.

Another aspect of bias that needs to be taken into consideration is the issue of algorithm transparency. Can the AI system explain the results comprehensively to humans? The US Department of Defense has been leading the effort in explainable AI. They realize that if a combat soldier is supposed to comprehend and trust in the recommendations of AI they have to be able to understand how the decision was reached in the first place. When new technology is deployed it must be clear ahead of time how the results will be monitored in order to judge the accuracy of the AI recommendations. In the healthcare industry there are nurse ethicists. It may be a reality in the near future that the role of AI ethicist is established for the deployment of new projects.

The topic of humans being replaced by AI has been the subject of debate and concern since technology began. But in 2017 the McKinsey Global Institute (MGI) published a report that stirred up a bigger debate than ever.  MGI predictions state that by the year 2030 anywhere from 75 to 375 million global workers will be replaced by AI technology and find themselves needing to create a niche in a new career path. The report does say that there will be new roles and new jobs created by the deployment of AI systems, though. The debate lies in what roles can actually be fully automated with no one monitoring for erroneous output.

 

AI at Work in Existing Healthcare Facilities: The Command Center

 For decades nurses have known that they’ve needed something to maximize the use of hospital beds. The problem has been that changing managerial practices, organizational strategies, and communication methods have struck out time and time again. Hospitals have remained somewhat disorganized and poorly run because there are simply too many departments, too many employees, and too many patients to take care of.

The good news is that with nurses’ input, advice, and help, hospitals are beginning to solve the problem. New centralized command centers are being used in hospitals in the United States, the first one being built at John Hopkins Hospital in 2016. The command centers’ basic job is getting patients in and out of beds in a safe, practical, and cost-effective manner.

The command center uses predictive analytic data and artificial intelligence to present decision-makers with the information they can use to put new policies that smooth patient flow and maximize the use of resources in place. More care is given without any additional costs, including hiring additional employees. Decision-makers can also use the data to make predictions to help them know when there will be a high occupancy of beds in the hospital. This allows the hospital to adopt an anticipatory approach to healthcare rather than just reacting to problems as they arise one by one.

Just some of the important data presented by the command center include:

  • Where have patients waited the longest
  • When and why surgeries are delayed
  • Where have discharges missed their target date
  • Where, when, and why are patients in distress
  • What departments have had employees who have called in sick and why

The AI command center is monitored around the clock by human staff that orchestrate patient care and work to avoid roadblocks. They provide the patient’s care as well.

Using the command center John Hopkins was able to increase the number of times the hospital can accept a very sick patient by 60%, decrease by 70% the time spent waiting for a room after surgery, and decrease waiting time in the ER by 25%. In other words, innovation and technology have helped solve a problem that couldn’t be solved by human-led changes alone. With future input and advice from nurses, the possibilities of the command center’s impact on healthcare are endless.

 

What the Future Holds

 Nursing will most assuredly be impacted when new AI technology assumes tasks that are traditionally performed by nursing staff. AI is going to revolutionize how nurse’s time is spent delivering patient care but the need for nurses will not be eliminated. A nurse’s experience and knowledge teamed with their skill will only transition into new ways of processing data and the nurse will evolve into the data integrator. The nurse will be the health coach, the main element of the delivery of human compassion. The nurse will be supported by artificial intelligence innovations, not replaced by them.

ABOUT THE AUTHOR: Heidi West is a medical writer for Vohra Wound Physicians, a national wound care physician group.  She writes about healthcare and technology in the medical industry.