Dev Kapadia, Applied Sciences, Fall 2020

Figure: The image depicts a Hill-Rom hospital bed patient control panel. Using this panel, patients can alarm nurses for any needs or concerns. Nurses can respond to hundreds of these alarms each day with almost none of them relating to urgent medical concerns. Using artificial intelligence and machine learning, these concerns can be filtered through to determine which alarms are urgent and cut down on the “alarm fatigue” for healthcare professionals.

Source: Wikimedia Commons

As each year passes, digitization becomes increasingly important. Digitization of resources has allowed individuals to perform tasks much more easily, and the new field of automation has eliminated some traditional, repetitive jobs. McKinsey Global Institute, one of the leading companies producing research into trends on the global economy, anticipates that artificial intelligence (AI) or machine learning (ML) will automate anywhere from four hundred million to eight hundred million currently held jobs by 2030. The trend is expected to particularly hit richer nations, like the United States, where it is estimated that anywhere from thirty-nine to seventy-three million jobs will be eliminated by 2030 because of automation (Manyika et al., 2017).

However, this is not by any means a cause for concern. Many workers of these jobs will be transferrable to other industries, and automation could actually serve to improve worker efficiencies and job numbers. While the McKinsey report estimates that physical workers and office support will surely be hit the hardest due to automation, other occupations such as managers, teachers, scientists, and more, are expected to see huge increases in employment numbers.

One field that will be enhanced with the presence of AI and ML is healthcare.  Healthcare professionals help save lives every day, and the wonders of automation are expected to not only reduce the time spent on menial tasks but also analyze data in ways that current technologies cannot. The popular medical blog “The Medical Futurist” outlines seven different applications of AI in healthcare (The Medical Futurist, 2020).

The first major setting where automation is expected to touch healthcare is in healthcare logistics. In this setting, AI assistants are expected to rectify the numerous patient absences and frustrating wait times experienced every day. With automation, the industry can expect to connect patients with transportation to their nearest hospital as well as better regulate what time patients should arrive (The Medical Futurist, 2020).

The next use-case of AI in healthcare is simply in automating repetitive tasks. Whether it be copying down notes from a provider’s clipboard to the Electronic Health Record (EHR) or sending reminders to patients for their appointments, automation can free professionals’ time to not only serve more patients but also serve them in a more meaningful way. This could mean spending time with them to create a customized plan for care or even discussing novel treatment methods. Either way, it is apparent that automations in this regard will greatly improve the quality and efficiency of the healthcare industry (The Medical Futurist, 2020).

Another area where AI and ML can be used is within the drug design processes. The drug design process suffers from many pressures. The two primary stresses that are placed on companies developing drugs is financial pressure from investors in the company and practicality pressures from the complexity of the science involved in producing effective drugs. These companies are pushed to produce effective drugs on a short timeline that overcome the scientific hurdles of molecular interactions and antimicrobial resistance. Companies can use AI and ML trained on datasets of past successful therapies to produce new treatments even faster (The Medical Futurist, 2020). In fact, this technology is already being used by companies like Atomwise, which identified two drugs that were likely to reduce Ebola infectivity during the 2015 outbreak (“New Ebola Treatment Using Artificial Intelligence,” 2015).

The fourth area where automation can be useful to healthcare workers is in working conditions. “Alarm fatigue” is the phenomena in which healthcare workers become desensitized to alarm signals because they hear hundreds of hospital alarms each day with only a few signaling true urgent care needs. With automation, machines can sort through the requests to severely cut down on the amount of alarms heard each day (The Medical Futurist, 2019). An algorithm developed by Fernandes et al. was actually so effective that it cut down on notifications to caregivers by 99.3% (Fernandes et al., 2019). The algorithm tracked anomalies in patient monitoring systems as well as frequency of alarming to send slightly delayed urgent messages to a caregiver’s phone app. Therefore, though the notification came in just seconds after the alarm was sent, they were always addressed immediately with the proper attention. Although this accuracy is very effective in limiting healthcare professional burnout, there is clearly much more work to do as even leaving 0.7% of important alarms unanswered could lead to life or death scenarios. This is just one example of the work that still needs to be done surrounding this industry.

The fifth area in healthcare calling for AI integration is in finding new signs of diseases. There are often signs, whether they be internal or external, that our body is not functioning properly due to foreign substances or unnatural processes. However, many of these signs are often disregarded, too subtle to be noticed, or tied to several other features that make analysis difficult. Using AI, researchers can find new ways of assessing measurable qualities to determine the risk or presence of a disease. There is a widespread interest in this subject with software companies like Google entering the healthcare field through this data-driven method, a field in which they have formidable experience in (The Medical Futurist, 2020).

There is also application of searching for disease beyond the patient themselves, particularly in pandemic identification, prevention, and response. This area of AI expansion is painfully suitable for the current environment. Although the Novel Coronavirus seemed to have caught entire countries by surprise, it is likely that with the proper technology and attention put towards pandemic detection, this outbreak could have been identified and potentially contained much sooner. In fact, in January of 2020, months before the virus was a concern in the United States, Lai et al. identified the virus in Wuhan and modeled how it could spread beyond Wuhan city using relative inflow and outflows from hotspots in and around the city. Though the model predicted large spreads around the world, it certainly did not predict the magnitude that we see today (Lai, Bogoch, Watts, et al., 2020). In fact, the article was later revised and published again in March to show a more accurate projection, unfortunately once the virus had already spread around the globe (Lai, Bogoch, Ruktanonchai, et al., 2020). Had the models been more robust and broadcasted at a larger scale, it is possible that the outbreak could have been contained to its source before becoming a worldwide pandemic (The Medical Futurist, 2020).

The next application is potentially the most well-known use for automation and AI in healthcare: analyzing large data sets for patient care. This goes beyond simply analyzing current patient characteristics to analyzing EHR data, other anonymized patient data, trends in environmental factors affecting health, and more to serve patients better. Many healthcare companies like Cerner or UnitedHealth Group are producing technologies to serve this need, and its existence can significantly improve patient outcomes in the future (The Medical Futurist, 2020).

Whether the world is ready or not, the age of automation is coming. However, rather than being characterized solely by its threat to worldwide job markets, it will also stimulate growth that will lead to more workers needed in other industries, as well as developing industries to build and maintain these automatized capabilities. Particularly as it relates to the healthcare industry, automation has the potential to save lives, and we should not just accept it but welcome and build upon it to improve our lives for the better.

 

References

Fernandes, C. O., Miles, S., Lucena, C. J. P. D., & Cowan, D. (2019). Artificial Intelligence Technologies for Coping with Alarm Fatigue in Hospital Environments Because of Sensory Overload: Algorithm Development and Validation. Journal of Medical Internet Research, 21(11), e15406. https://doi.org/10.2196/15406

Lai, S., Bogoch, I. I., Watts, A., & Khan, K. (2020). Preliminary risk analysis of 2019 novel coronavirus spread within and beyond China. 19.

Lai, S., Bogoch, I., Ruktanonchai, N., Watts, A., Lu, X., Yang, W., Yu, H., Khan, K., & Tatem, A. J. (2020). Assessing spread risk of Wuhan novel coronavirus within and beyond China, January-April 2020: A travel network-based modelling study. MedRxiv, 2020.02.04.20020479. https://doi.org/10.1101/2020.02.04.20020479

Manyika, J., Lund, S., Chui, M., Bughin, J., Woetzel, J., Batra, P., Ko, R., & Sanghvi, S. (2017, November 28). Jobs lost, jobs gained: What the future of work will mean for jobs, skills, and wages | McKinsey. https://www.mckinsey.com/featured-insights/future-of-work/jobs-lost-jobs-gained-what-the-future-of-work-will-mean-for-jobs-skills-and-wages

New Ebola Treatment Using Artificial Intelligence. (2015, March 24). Atomwise. https://www.atomwise.com/2015/03/24/new-ebola-treatment-using-artificial-intelligence/

The Medical Futurist. (2019, April 30). The Real Era of The Art of Medicine Begins With Artificial Intelligence. The Medical Futurist. https://medicalfuturist.com/artificial-intelligence-and-the-art-of-medicine

The Medical Futurist. (2020, October 21). 7 Things You Can Expect From A.I. In Healthcare. The Medical Futurist. https://medicalfuturist.com/7-things-you-can-expect-from-a-i-in-healthcare