Pop Quiz #4: Post Midterm Exam Introduction, Narration, Confirmation, Concession, Refutation(s), and Summation

Argumentative research paper on the role of AI in our lives.

Laying down the foundation: 9 pro scholarly sources and 3 con scholarly sources

Priscilla Laguna

ENG 2105

Dr. Gill

Pro #1 Journal

21 October 2021

“The best writing is rewriting”: 2 Draft, 0 Tutorials, 0 Teacher Conference

Bouletreau, P., et al. “Artificial Intelligence: Applications in Orthognathic Surgery.” Journal of Stomatology, Oral and Maxillofacial Surgery, vol. 120, no. 4, 26 June 2019, pp. 347–354., https://doi.org/10.1016/j.jormas.2019.06.001.

Key quote: “As far as the first component (Basic Knowledge) is concerned, there is little doubt that machines have by far exceed human capacities in storing and organizing data” (Bouletreau et al. 352).

(Provocative Title) AI: An Orthodontist’s Trusty Companion

(Unity/Topic Sentence) Bouletreau et al. consider the application of artificial intelligence (AI) in orthognathic, or jaw surgery, as a worthwhile investment. (Adequate Development/Body) The beginning of the article outlines the evolution of AI. Although development research of AI was first introduced in 1956, it was not widely accepted until the 1990s where an AI beat a world chess champion. Since then, the use of AI has spread from small scale applications, such as smartphone use, to applications such as medicine diagnostics and surgery. Bouletreau et al. consider four advantages of using AI in orthognathic surgery: diagnostic precision, 3D images and models, custom surgical tools and equipment, and improvement of follow-up procedures. After going into depth into each advantage, challenges are also considered. The planification of orthognathic surgery comprises of three phases: basic textbook knowledge for diagnosis of the problem, ability to recognize any aesthetic changes due to surgery, and knowing the patient’s motivation for wanting the surgery. With AI’s enormous potential for storing data, AI wins over humans in first phase of basic knowledge. The other two require human instinct and intuition. Predicting how the surgery will change the physical appearance of the person can be tricky and involves many variables; therefore, coming up with a single algorithm to detect this is not an easy task. As for the third phase, AI still have not been developed with the full capacity of interpreting the drive and motivation of human actions. As a result, patient satisfaction is not in favor for the AI brain. (Coherence/Conclusion) Overall, Bouletraeu and colleagues emphasize the importance of orthognathic surgeons to consider more collaboration with AI to carry out treatment.

Pro #2 Journal

Mirnezami, R., and A. Ahmed. “Surgery 3.0, Artificial Intelligence and the next-Generation Surgeon.” British Journal of Surgery, vol. 105, no. 5, 2018, pp. 463–465., https://doi.org/10.1002/bjs.10860.

Key quote: “Surgical AI should be seen for what it is: an opportunity to enhance surgical care and to complement, rather than replace, the human surgeon” (465).

(Provocative Title) Forming AI and Human Partnerships in Surgical Science

(Unity/Topic Sentence) Mirnezami and Ahmed consider the advances in AI technology as opportunities for doctors and AI to work together. (Adequate Development/Body) The future that they envision is one where doctors learn to consult and rely on the abundance of data provided by their “AI colleague[s]” (464). The authors argue that the main reason why this idea has not been widely accepted yet is, not because of lack of evidence, but the lack of awareness of the benefits and potential AI have in the medical field. For example, before performing a surgery, the surgeon needs to consider numerous factors such as the probabilities of risk, bed rest, complications, how widespread the disease is, and so on. These are complex variables that require an equally large amount of data. With the development of clinical decision support (CDS) algorithms implemented into AI, surgeons can have all this information at the touch of a button. These AI can even be programmed to communicate with other AI from other countries and other continents to access additional databases. Over the course of time, new research results in new surgical methods that prove to be more efficient than previous techniques. The advantage of using AI is that these novel methods can be taught to them in a few hours, which would normally take years to train a human being to do the same thing. (Coherence/Conclusion) Mirnezami and Ahmed understand the opposing view on the ethics of making autonomous AI; therefore, their proposition is for humans and AI to work together instead of letting AI take over completely.

Pro #3 Journal

Briganti, Giovanni, and Olivier Le Moine. “Artificial Intelligence in Medicine: Today and Tomorrow.” Frontiers in Medicine, vol. 7, 5 Feb. 2020, pp. 1–6., https://doi.org/10.3389/fmed.2020.00027.

Key quote: “Empatica received FDA approval in 2018 for their wearable Embrace, which associated with electrodermal captors can detect generalized epilepsy seizures and report to a mobile application that is able to alert close relatives and trusted physician with complementary information about patient localization” (3).

(Provocative Title) Medical Engineers: Products of a Computer Science Based Curriculum

(Unity/Topic Sentence) Briganti and Le Moine outline the current benefits of AI in various medical fields and give suggestions to overcome barriers of opposition for their use. (Adequate Development/Body) Briganti and colleagues define the purpose of AI as a tool to provide both doctors and patients more control over their health “by performing early diagnosis, reducing complications, optimizing treatment and/or providing less invasive options, and reducing the length of hospitalization” (1). Overall, their opinion aligns with that of Mirnezami and Ahmed that AI should not replace but compliment the work performed by doctors. Briganti and Le Moine also give various examples and studies of present-day applications of AI in diagnosing or monitoring diseases using smartphone apps and even Apple watches in the fields of cardiology, pulmonary medicine, endocrinology, nephrology, gastroenterology, neurology, and oncology. They argue that studies of comparison between the capabilities of AI and human doctors should focus on the results of doctors that do not work with AI versus doctors who do use AI to help them diagnose and carry out surgeries. Additional evidence that supports the author’s stance on AI as beneficial to society is that “medical technology is one of the most promising markets of the 21st century, with an estimated market value rapidly approaching a thousand billion dollars in 2019” (3). These statistics point to the conclusion that it is worth investing in educating future doctors to apply AI in their medical practices. Briganti and Le Moine suggest that universities modify the medical curriculum by adding classes in the subjects of physics, engineering, and computer science. (Coherence/Conclusion) As a result, these new doctors will serve as pioneers of innovative medical technology and help erase any doubt from any party who opposes the implementation of AI in medicine.

Pro #4 Journal

Liang, Xuebing, et al. “Artificial Intelligence in Plastic Surgery: Applications and Challenges.” Aesthetic Plastic Surgery, vol. 45, no. 2, 2020, pp. 784–790., https://doi.org/10.1007/s00266-019-01592-2.

Key quote: “As the technique of AI-guided material discovery matures, better and safer synthetic implants can be designed for use in plastic surgery” (Xuebing et al. 787).

(Provocative Title) AI as a Tool for Increasing the Success Rate and Patient Satisfaction of Plastic Surgery

(Unity/Topic Sentence) Xuebing et al. follow the same pattern of opinion as the previous authors that AI should be used to improve surgical planning rather than replace the human factor completely. (Adequate Development/Body) The purpose of AI in the medical field is to have the capability to interpret various data through programmed algorithms and choose the best surgical method based on simulations of various scenarios. Studies analyzed by Xuebing and colleagues have shown that AI can detect subtle patterns in medical imaging and statistics that would otherwise be overlooked by the human surgeon. The traditional method that surgeons use to plan out a surgery was the tree method. Although this method did take into consideration the various complex variables and probabilities that affect the success of the surgery, it was limited to only analyzing one outcome. Based on this fact, Xuebing et al. researched the use of the Markov decision-making model by AI. The results indicated that this method proved very useful since the AI had the ability to multiple scenarios at the same time. Basically, the AI analyzed one step of the planning and surgical process at a time. The AI noted the possible pathways that could be taken in each step and considered which path resulted in the more desirable outcome. Then, the AI would move on to the next step and repeat the process. Xuebing et al. concluded that this technique allowed for a personalized treatment plan for the patient undergoing the plastic surgery. Taking this a step further, Xuebing et al. considered the use of AI to conduct research and improve materials and methods used in plastic surgery. For example, a study made by other researchers applied AI developed by Cognovi Labs to analyze the emotional response of people on social media to keywords such as “liposuction”, “plastic surgery”, and “cosmetic surgery” in order to determine what types of plastic surgeries would be most profitable in the future. Xuebing et al. also give their own suggestions of how machine learning by AI can be used to make plastic surgery more cost-effective by searching through a database of chemical compounds to discover new materials that are safer for the patient and cheaper to produce. (Coherence/Conclusion) Xuebing et al. express with the evidence gathered by their own research and the research of others high hopes for the enhancement of AI in the area of plastic surgery.

Pro #5 Journal

Hashimoto, Daniel A., et al. “Artificial Intelligence in Surgery: Promises and Perils.” Annals of Surgery, vol. 268, no. 1, 2018, pp. 70–76., https://doi.org/10.1097/sla.0000000000002693.

Key quote: “By using clinical variables such as patient history, medications, blood pressure, and length of stay, ANNs, in combination with other ML approaches, have yielded predictions of in-hospital mortality after open abdominal aortic aneurysm repair with sensitivity of 87%, specificity of 96.1%, and accuracy of 95.4%.” (73).

(Provocative Title) Four Aspects of AI Learning

(Unity/Topic Sentence) Hashimoto et al. advocate for surgeons to work together with data scientist to develop and improve AI capability to analyze and interpret clinical data to ensure the best medical care for patients. (Adequate Development/Body) Hashimoto and his colleagues consider four main aspects of AI and their applications in surgical settings. The first is machine learning (ML) where, much like humans, AI learn to recognize patterns and even learn from their mistakes. Another characteristic of AI is natural language processing (NLP) where the AI learns to interpret the underlying meaning and emotions associated with words in the spoken language. This is a huge step in the sophistication of AI because it allows medical researchers and surgeons more freedom to write in a more narrative and natural way rather than technical and to the point. To make AI as human as possible, in the sense of cognitive capability, scientists have attempted to replicate the biological nervous system by creating layers of simple computational units that work similarly to neurons in a brain. As the AI makes connections between these “neurons”, studies have shown that AI develop the capability to distinguish more complex and subtle patterns. Lastly, computer vision considers an AI’s interpretation of images and videos. Applications of this include computer-aided diagnosis, image-guided surgery, and virtual colonoscopy. (Coherence/Conclusion) As the cherry on top of the cake, the capabilities described previously can be stored for easy share and access to all the gathered information from pre- to post- surgical steps and results with other doctors across the globe.

Pro #6 Journal

Schwartz, John T., et al. “Applications of Machine Learning Using Electronic Medical Records in Spine Surgery.” Neurospine, vol. 16, no. 4, 2019, pp. 643–653., https://doi.org/10.14245/ns.1938386.193.

Key quote: “It would allow surgeons to better predict and prepare for postoperative complications, more efficiently utilize hospital resources, and prioritize patient surveillance on those who are at the greatest risk. In fact, multiple studies have demonstrated the ability to build predictive models using EMR data for major perioperative complications in spine surgery, particularly surgical site infections.” (Schwartz et al. 648).

(Provocative Title) Creating a Brain for AI to Aid in Spinal Surgery

(Unity/Topic Sentence) Schwartz et al. explains the advantages of machine learning (ML), natural language processing (NLP), and deep neural network (DNN) in spinal surgery. (Adequate Development/Body) Most of the data gathered in preparation for spinal surgery is image data from CT and MRI scans; therefore, AI algorithms need to be able to take measurements and distinguish between different topographical features present in these images. Basically, the AI needs to be able to perform three tasks: image classification, or detect the presence of vertebrae; object detection, or tell the difference between bone and tissue; segmentation or identify the outlines and boundaries between each vertebra and the surrounding tissue. As an example, the University of Cincinnati created a “guess-and-revise” ML algorithm in which the AI makes an approximation of where each vertebra is located and makes revisions based on further data gathered from the location of the rest of the vertebrae. This process usually takes a trained person about 15 minutes per patient, but the algorithms can expedite the process with about the same accuracy. During a spinal operation, surgeons, even with their level of expertise, need to eyeball where to place pedicle screw placements and to measure Cobb angles. Most of the time, surgeons among themselves can disagree with these measurements by a 10% difference. Use of DNN algorithms can help with more accurate and more consistent measurements. In the case of NLP applications, the reports that surgical doctors must make for each patient can be very burdensome and time consuming. NLP algorithms can generate templates for these reports with prefilled data included, alleviating clerical work. Finally, AI can help surgeons predict post-operative risks and implications for each patient. This is not only beneficial to the patient, but also for the hospital. The application of AI will ensure that the hospital’s scarce resources are allocated in the best way possible. (Coherence/Conclusion) The literature research and analysis made by Schwartz et al. provide evidence of the benefits of AI application in spinal surgery.

Pro #7 Journal

Rimmer, Lara, et al. “The Automaton as a Surgeon: The Future of Artificial Intelligence in Emergency and General Surgery.” European Journal of Trauma and Emergency Surgery, vol. 47, no. 3, 2020, pp. 757–762., https://doi.org/10.1007/s00068-020-01444-8.

Key quote: “AI possesses substantial promise in the field of emergency surgery… to refer appropriate patients through the emergency department efficiently, alert any surgical issues on imaging, and even predict operative risks based on vital sign observations and clinical history to allow a surgeon to provide a personalized risk profile” (Rimmer et al. 760).

(Provocative Title) Preventing Unnecessary Surgery and Expediting Time-Sensitive Surgeries with AI Logarithms

(Unity/Topic Sentence) While most of the articles analyzed so far consider the use of AI for pre-planned surgeries, Rimmer et al. focused on AI’s potential to help surgeons detect the need for emergency surgeries. (Adequate Development/Body) One emergency surgical procedure is the removal of the appendix. It is usually very hard to diagnose causing both life-threatening erroneous negative diagnoses and unnecessary surgery from erroneous positive diagnoses. A sample size of 590 patients with both positive and negative diagnoses of appendicitis in Germany was implemented in an AI algorithm to detect which biomarkers indicate appendicitis. The researchers concluded that two-thirds of cases where surgery was done unnecessarily could have been prevented if AI assistance was used. Similar to the information presented in other articles, Rimmer et al. suggest positive outlooks for AI in creating personalized surgical risk profiles for each patient’s unique situation. For example, two factors to take into consideration before performing an emergency operation are blood loss and the length of postoperative stay in ICU. Research conducted by Lei et al. focused predicting postoperative acute kidney injury by analyzing factors such as the two mentioned previously. Their sample size consisted of analyzing 12,303 cases of ICU patients after performing an emergency surgery. The mathematical model that they created had a success rate between 85% and 87% for predicting long-term stays in ICU. Additionally, an ongoing study involves developing AI that can be put inside the patient to regenerate tissue such as that seen in birth defects. While the operation is taking place, AI machines can be programmed to gives instantaneous feedback to the surgeon on their quality of surgical precision, such as indicating how much force the surgeon is putting on the soft tissue. (Coherence/Conclusion) While Rimmer et al. do acknowledge the present opposition on the legal and ethical debates on the growing use of AI, they do believe that, with time, AI assistance can be a valuable tool for surgeons.

Pro #8 Journal

Solanki, Sohan Lal, et al. “Artificial Intelligence in Perioperative Management of Major Gastrointestinal Surgeries.” World Journal of Gastroenterology, vol. 27, no. 21, 2021, pp. 2758–2770., https://doi.org/10.3748/wjg.v27.i21.2758.

Key quote: “An ML algorithm with a custom Python script has been shown to optimize the efficiency of operating room booking times that resulted in a reduction in nursing overtime of 21%, which was equivalent to saving of half a million dollars” (Solanki et al. 2763).

(Provocative Title) Present and Future Applications of AI in Surgery and For Fighting the COVID-19 Pandemic

(Unity/Topic Sentence) Solanki et al. stress the importance of surgeons from all areas to familiarize themselves and keep up to date with the development of AI in surgical procedures. (Adequate Development/Body) For now, the AI algorithms developed for gastrointestinal surgeries are still in the trial phase and still not ready to be applied in a real operative setting, but the case studies analyzed by Solanki et al. provide evidence of promising experimental results. Solanki and colleagues divided this case studies into three phases of the surgical process: before surgery, during surgery, and after surgery. Preoperative advantages of AI include imaging, risk assessment, and endoscopy. For example, 3D imaging of the liver can help surgeons to plan out complex surgeries that would otherwise be deemed as impossible. With these images, virtual and navigation-assisted surgeries can be performed to predict the best path to take for surgery. As for the intraoperative phase, intubating and operating robots along with enhanced monitoring can help increase the surgeon’s efficiency in performing the surgery. Jacob et al., a group of researchers, invented a robotic scrub nurse called Gestonurse that functions without the need for verbal or manual commands. Instead, it can interpret hand gestures to give the surgeon the necessary surgical instruments. Recognizing the potential dangers of giving AI complete autonomy in the operating room, Solanki et al. point out that AI should be seen as an opportunity to enhance human capability and performance in surgical procedures; this goal is accomplished by AI providing better imaging and statistics rather than having AI replace the human surgeon completely. The goal set forth for AI in postoperative treatment is to be equipped with an algorithm that can predict unwanted surgical outcomes such as anastomotic leaks. Solanki et al. hope that this will make patient recovery less painful by being able to prevent postsurgical complications. Taking a step further, Solanki et al. consider the possible applications of AI to help fight the COVID-19 pandemic. Due to the virus’ airborne capability of spreading from person to person, many surgeries were put on hold. The transition to more robotic surgeries supervised by human surgeons can decrease the risk of infection. Additionally, with hospital resources scarcer than ever, “An ML algorithm with a custom Python script has been shown to optimize the efficiency of operating room booking times that resulted in a reduction in nursing overtime of 21%, which was equivalent to saving of half a million dollars” (Solanki et al. 2763). Therefore, using AI algorithms can help save time and money for hospitals. (Coherence/Conclusion) Although AI does have lots of room for improvement in the area of gastrointestinal surgeries and for fighting the pandemic, trials and research are underway to smooth out any kinks in functionality.

Pro #9 Journal

Zhou, Xiao-Yun, et al. “Application of Artificial Intelligence in Surgery.” Frontiers of Medicine, vol. 14, no. 4, 2020, pp. 417–430., https://doi.org/10.1007/s11684-020-0770-0.

Key quote: “Learning from demonstration (LfD), … is a popular paradigm for enabling robots to perform autonomously new tasks with learned policies. The common framework of LfD is to first segment a complicated surgical task into several motion primitives or subtasks, followed by recognition, modeling, and execution of these subtasks sequentially” (Zhou et al. 423).

(Provocative Title) Pushing the Boundaries of AI Capability

(Unity/Topic Sentence) Zhou et al. break down current research and development of AI algorithms for preoperative planning and robotic surgical procedures and guidance. (Adequate Development/Body) A training process for AI called deep learning consist of multiple layers of “neural” networks in an algorithm that can help the AI detect subtle patterns that would go unnoticed by humans. This advantage is used for predicting mortality rate, kidney failure, and internal bleeding which can help prevent discomfort and pain for patients post surgically. Also, Gibson et al. and other fellow researchers are coming up with strategies to help train AIs to differentiate between a surgical instrument and an organ tissue during operation. This can improve accuracy and speed of using these instruments for the surgeon in areas of the body where the naked eye is not of much use. A more complex application consists of 3D imaging of a specific instance during the surgical procedure. Usually, constructing an image of great precision and detail is time consuming, but researchers are working hard to decrease the time and number of images needed. As of now, surgical robots can perform simple tasks, such as suturing, by using a method called deep multistage detection that helps identify the shape of the thread in between all the tissue. Another approach is learning from demonstration where the AI watches a human surgeon complete a new procedure that it has never done before and then imitates the techniques faster and with more precision. Application of augmented and virtual reality from the gaming world to surgical procedures can enable remote cooperation between surgeons who are far away but can give real-time advice on how to carry out the operation. (Coherence/Conclusion) Going more into the future, Zhou et al. propose the development of nanobots for noninvasive surgeries to create more flexible, smaller, and cheaper surgical methods.

Con #1 Journal

AlHasan, AJMS. “Bias in Medical Artificial Intelligence.” The Bulletin of the Royal College of Surgeons of England, vol. 103, no. 6, 2021, pp. 302–305., https://doi.org/10.1308/rcsbull.2021.111.

Key quote: “Machines are only as impartial as the data that have been fed into them” (AlHasan 303).

(Provocative Title) Raising AI to be Prejudiced

(Unity/Topic Sentence) AlHasan describes four areas in which AIs have shown bias in treating patients: gender, race, language, and socioeconomic standing. (Adequate Development/Body) The basis by which AI learn their respective functions is by analyzing data given to them by their human inventors. Unfortunately, nondiverse and small sample sizes mess up AI’s efficacy in diagnoses. For example, a recent study made by Sjoding et al. in 2020 found signs of racial bias in pulse oximetry sensors. On average, Black patients suffered lower blood oxygen levels than White patients at the same oxygen saturation readings read by the sensor. This caused much alarm because monitoring low blood oxygen levels in patients with the COVID-19 virus is of utmost importance in keeping them alive. Another example, an algorithm was designed to detect which patients needed extra care now in the hospital to reduce future healthcare costs. Although, the algorithm was designed with good intentions, it discriminated against those who needed this help the most. It was found that the sicker and poorer patients were neglected and not given the extra care that they needed. Instead, more money and care were spent on people who were better off and experienced less in-hospital stay. Natural language processing (NLP) is a tool that teaches AI to interpret spoken language. The problem is that this type of programming is very sensitive to certain language dialects. For instance, the University of Toronto created an algorithm to detect speech impairment as a sign of Alzheimer’s. The group who designed the algorithm noticed that the AI was only good at recognizing Canadian English and had a hard time identifying speech patterns of French speakers and those of other indigenous languages in Canada. (Coherence/Conclusion) In order to correct the AI bias, AlHasan suggests implementing a four-stage strategy —data, development, delivery, and dashboard— that stresses testing for bias in each step before implementing the AI in the medical field.

Con #2 Journal

Doshi, Hiten, et al. “Machine Learning in Atrial Fibrillation—Racial Bias and a Call for Caution.” Journal of Medical Artificial Intelligence, vol. 4, 30 Sept. 2021, p. 6., https://doi.org/10.21037/jmai-21-12.

Key quote: “Many of the most commonly used ECG datasets only report limited demographic data, including the patient’s age, gender, and/or baseline clinical characteristics, without reporting racial or ethnic background.” (Doshi et al. 6).

(Provocative Title) Possible Solutions for Fixing AI Bias in Medical Diagnosis

(Unity/Topic Sentence) Doshi et al. offer an explanation as to why AI systems show racial discrimination towards diagnosis of atrial fibrillation. (Adequate Development/Body) Diagnosing atrial fibrillation (AF) is a challenge. Most of the time, it is missed despite analysis of electrocardiograms (ECGs). Hope for better diagnosis has been put on AI through machine learning of data input. Unfortunately, most of the data provided come from White patients; therefore, Black patients who are at risk of AF are usually not diagnosed. Doshi and colleagues point out that the most common databases of which AI developers feed it with are very limited including only the patient’s age, gender, and other general clinical demographics but no mention of race or ethnicity. This poses a problem, especially for AF, because it is widely known among medical practitioners that AF symptoms vary greatly depending on race. Without specification of race in the data, there is no way for the AI to detect AF in non-White patients accurately. This causes a phenomenon called “racial paradox” where Black patients have a higher risk of AF, but the data shows low incidence rates of AF in Blacks. Doshi et al. conclude that this miscorrelation comes from underdiagnosis of AF in Blacks. In retrospect, Doshi et al. propose three solutions for correcting the bias algorithms implemented in AI systems. First, AI algorithms created for detecting AF should be tested for bias on a variety of ethnic groups before releasing them for public use. Second, more emphasis should be put on reporting ethnic and racial information on the datasets used for creating these algorithms. Third, hospitals should use their own database to construct these algorithms instead of using the common and flawed datasets that are in use now. (Coherence/Conclusion) With these guidelines in mind, Doshi et al. hope that future researchers and developers can decrease the gap in AI bias.

Con #3 Journal

Key quote: “Data included 1,535 (14.4%) adverse events with significant negative patient impacts, including injuries (1,391 cases) and deaths (144 cases), and over 8,061 (75.9%) device malfunctions.” (Almezadeh et al. 7).

(Provocative Title) The Robot Crime

(Unity/Topic Sentence) Alemzadeh et al. gathered data available from the FDA MAUDE database to analyze cases in which surgical injury or death was caused by robotic operators. (Adequate Development/Body) An example of an incident report is provided in the article where the commonly used Da Vinci surgical robot experienced communication complications in the electronic system between the robot’s arm and the software that controls it. Most of the reports comprised of failures from the Da Vinci surgical robots with a wide variety of reasons from broken instrument tips, electronic failure, and user error. Most of the evidence found points to complications in surgeries that are more complex such as cardiovascular surgery. An increase in the number of incidents may be due to the increase of robots over the years. (Coherence/Conclusion) In conclusion, evaluation of the data presented by Alemzadeh et al. points to flaws in the AI operating system that need to be addressed.

Brainstorm: Prewriting Steps

Priscilla Laguna

ENG 2105

Dr. Gill

Prewriting on Images, Media, and Privacy

15 November 2021

“The best writing is rewriting”: 2 Drafts, 1 WC Tutorial (Amanda Kirschner), 0 Teacher Conference

Step 1: Who is my audience?

My audience consists of people who have little to no knowledge about AI use in surgical procedures; therefore, they doubt the benefits and fear their use in the medical field.

Step 2: What is my purpose?

My purpose is to convince the audience that AI use in medicine, specifically surgery of all kinds, are beneficial. Rather than seeing them as replacements for human surgeons, the audience’s perspective should focus on seeing AI as complements for surgeons in carrying out these procedures.

Step 3: What is my premise?

My premise is, despite opposition based on the complications of legalities, ethics, and cost of AI, the benefits of using AI for carrying out surgical procedures outweigh the disadvantages because AI in the medical field are created with the intention to complement, not replace, human surgeons, AI in the medical field have the capability to prevent deadly accidents during surgery by formulating personalized risk factor algorithms for each patient before undergoing surgery, and AI in the medical field have been found to allocate scarce hospital resources in more cost and time efficient ways.

Step 4: What is my chosen quotation?

“Learning from demonstration (LfD), … is a popular paradigm for enabling robots to perform autonomously new tasks with learned policies. The common framework of LfD is to first segment a complicated surgical task into several motion primitives or subtasks, followed by recognition, modeling, and execution of these subtasks sequentially” (Zhou et al. 423).

Rough Draft:

Priscilla Laguna

ENG 2105

Dr. Gill

Images, Media, and Privacy Argumentative Research Paper

15 November 2021

“The best writing is rewriting”: 1 Draft, 0 Tutorials, 0 Teacher Conference

(Provocative Title) AI: The Complementary “Eve” for Surgeons

(Hook) Evaluating artificial intelligence capabilities, Rimmer et al., research scientists in medical robotics, argue, “AI possesses substantial promise in the field of emergency surgery… to refer appropriate patients through the emergency department efficiently, alert any surgical issues on imaging, and even predict operative risks based on vital sign observations and clinical history to allow a surgeon to provide a personalized risk profile” (Rimmer et al. 760). (Bridge) Rimmer and colleagues envisioned Artificial Intelligence (AI) as an autonomous aid in emergency medicine, ranging from referrals to warnings based on observations or stored data. (Divided thesis/stasis sentence: Opponent’s claim informed by 3 scholarly sources) Although rising controversy revolves around the complications of legalities, ethics, and cost of AI, (Rhetor’s main claim informed by 9 scholarly sources) the benefits of using AI for carrying out surgical procedures outweigh the disadvantages because (Reason/Support 1) advanced machine learning methods such as artificial neural networks give AI higher learning capability and accuracy, (Reason/Support 2) AI algorithms help surgeons create a personalized surgical procedure for each patient by predicting possible risks and period of postsurgical recovery, (Reason/Support 3) and AI 3D imaging during surgical procedures seek to complement, not replace, human surgeons in carrying out these operations.

(Narration) An unprecedented event in December of 2017 marked a breakthrough in computer science. An AI research company under Google called DeepMind created AlphaZero, a computer algorithm with the ability to teach itself. In just under four hours, AlphaZero mastered and developed its own strategies for the complex game of chess. When put to the test against the world champion chess program, AlphaZero’s performance resulted in an astonishing victory (Mirnezami 463). This incident sparked a domino effect inciting more research and development in AI such as the Chinese government investing “$2.1 billion on an AI industrial park in 2018” (Xuebing 784). This influence has trickled down to expanding the use of AI in surgical operations from 3D modeling to creating an artificial “neural system” like the human brain. Supporters for the use of AI in surgical procedures believe that machine learning algorithms can close the gap in misdiagnoses by detecting subtle patterns in medical imaging that would otherwise be overlooked by a human surgeon. They also believe that a partnership between surgeons and computer scientists can smooth out any kinks in the AI operating system to ensure the best medical care for patients. On the other hand, critics believe that the racial bias present in the non-diverse data input causes more detrimental effects than positive effects in patient healthcare. The missing link in both lines of reasoning is that the ability for AI to reteach itself comes from imitation of the human model.

(Confirmation) The beauty behind AI systems lies in their adaptability to changing circumstances without the frustration of surgeons relying on a programmed computer that only functions properly in the preset parameters. Xuebing et al. put this into context after analysis of various literature on AI applications in plastic surgery, “As the technique of AI-guided material discovery matures, better and safer synthetic implants can be designed for use in plastic surgery” (Xuebing et al. 787). This statement suggests that the growth potential for AI to improve on mistakes made through trial and error has no boundary. In correlation, the 1983 movie “War Games” relates the learning process of a computer who learns that not all games can be won or lost; rather, sometimes the outcome is a tie.

Editing Log

1. Principle: coh: Coherence. Ideas and sentences need to flow through the use of transition words or parallel ideas and
structure.
  Error: The 3 supportive arguments given in the thesis statement do not correlate with the claims given in the opponent's
argument.
  Original: Although rising controversy revolves around the complications of legalities, ethics, and cost of AI, (Rhetor’s main
claim informed by 9 scholarly sources) the benefits of using AI for carrying out surgical procedures outweigh the
disadvantages because (Reason/Support 1) advanced machine learning methods such as artificial neural networks give AI
higher learning capability and accuracy, (Reason/Support 2) AI algorithms help surgeons create a personalized surgical
procedure for each patient by predicting possible risks and period of postsurgical recovery, (Reason/Support 3) and AI 3D
imaging during surgical procedures seek to complement, not replace, human surgeons in carrying out these operations.
  Revision: Although rising controversy revolving around AI bias and reports of robotic surgical injuries and deaths create
fear and doubt on the possibility of AI autonomy, (Rhetor’s main claim informed by 9 scholarly sources) the benefits of
using AI for carrying out surgical procedures outweigh the disadvantages, because (Reason/Support 1) AI in the medical
field are created with the intention to complement, not replace, human surgeons, (Reason/Support 2) AI in the medical field
has the capability to prevent deadly accidents during the surgery by formulating personalized risk factor algorithms for
each patient before undergoing surgery, (Reason/Support 3) and AI in the medical field has been found to allocate scarce
hospital resources in more cost- and time-efficient ways.

2. Principle: Word choice: Either the word used does not belong in the context, does not have the correct connotation, or does
not have the right tense.
  Error: The wrong verb tense was used. Since the event described happened in the past, the word “unprecedent” needed to
be changed to the past tense “unprecedented”.
  Original: An unprecedent event in December of 2017 marked a breakthrough in computer science.
  Revision: An unprecedented event in December of 2017 marked a breakthrough in computer science.

3. Principle: MLA in-text citation: Whenever a quote from a source is used in an essay, credit needs to be given to the authors
of the source.
  Error: I did not include the profession of the authors nor the title of the journal article written by them before including the
quote.
  Original: “Machines have by far exceed human capacities in storing and organizing data” (Bouletreau et al. 352).
  Revision: A group of orthognathic surgeons of whom have worked with AI in their line of work acknowledge in their
article “Artificial Intelligence: Applications in Orthognathic Surgery”, “machines have by far exceeded human capacities in
storing and organizing data” (Bouletreau et al. 352).

Final Draft
17 November 2021
“The best writing is rewriting”: 2 Drafts, 2 WC Tutorials (Amanda Kirschner and Jason Leclair), 0 Teacher Conference

(Provocative Title) AI: The Complementary “Eve” for Surgeons

(Hook) Evaluating artificial intelligence capabilities, Zhou et al., research scientists in medical robotics, argue, “Learning from demonstration (LfD), … is a popular paradigm for enabling robots to perform autonomously new tasks with learned policies. The common framework of LfD is to first segment a complicated surgical task into several motion primitives or subtasks, followed by recognition, modeling, and execution of these subtasks sequentially” (Zhou et al. 423). (Bridge) Zhou and colleagues compare the learning capability of artificial intelligence (AI) to that of a child observing and imitating a parent’s speech and actions. (Divided thesis/stasis sentence: Opponent’s claim informed by 3 scholarly sources) Although rising controversy revolving around AI bias and reports of robotic surgical injuries and deaths create fear and doubt on the possibility of AI autonomy, (Rhetor’s main claim informed by 9 scholarly sources) the benefits of using AI for carrying out surgical procedures outweigh the disadvantages, because (Reason/Support 1) AI in the medical field are created with the intention to complement, not replace, human surgeons, (Reason/Support 2) AI in the medical field has the capability to prevent deadly accidents during the surgery by formulating personalized risk factor algorithms for each patient before undergoing surgery, (Reason/Support 3) and AI in the medical field has been found to allocate scarce hospital resources in more cost- and time-efficient ways.
(Narration) An unprecedented event in December of 2017 marked a breakthrough in computer science. An AI research company under Google called DeepMind created AlphaZero, a computer algorithm with the ability to teach itself. In just under four hours, AlphaZero mastered and developed its own strategies for the complex game of chess. When put to the test against the world champion chess program, AlphaZero’s performance resulted in an astonishing victory (Mirnezami 463). This incident sparked a domino effect inciting more research and development in AI such as the Chinese government investing “$2.1 billion on an AI industrial park in 2018” (Xuebing 784). This influence has trickled down to expanding the use of AI in surgical operations from 3D modeling to creating an artificial “neural system” like the human brain. Supporters for the use of AI in surgical procedures believe that machine learning algorithms can close the gap in misdiagnoses by detecting subtle patterns in medical imaging that would otherwise be overlooked by a human surgeon. They also believe that a partnership between surgeons and computer scientists can smooth out any kinks in the AI operating system to ensure the best medical care for patients. On the other hand, critics believe that the racial bias present in the non-diverse data input causes more detrimental effects than positive effects in patient healthcare. The missing link in both lines of reasoning is that the ability for AI to reteach itself comes from imitation of the human model.
(Confirmation) The beauty behind AI systems lies in their adaptability to changing circumstances without the frustration of surgeons relying on a programmed computer that only functions properly in the preset parameters. Liang et al. put this into context after analysis of various literature on AI applications in plastic surgery, “As the technique of AI-guided material discovery matures, better and safer synthetic implants can be designed for use in plastic surgery” (Liang et al. 787). This statement suggests that the growth potential for AI to improve on mistakes made through trial and error has no boundary. In correlation, the 1983 movie “War Games” relates the learning process of a computer who learns that not all games can be won or lost; rather, sometimes the outcome is a tie.
(Concession/Refutation) It is, indeed, true that reports indicating evidence of AI bias and of robotic surgical injuries and deaths create a certain amount of concern. (Scholarly Source 1) For example, expert surgeon AlHasan argues, “machines are only as impartial as the data that have been fed into them” (AlHasan 303). Just like a child’s culture and surroundings greatly influence the child’s point of view of themselves and others, an AI’s behavior and performance depends on the information inputted into it. (Scholarly Source 2) According to the opposition, the extent of an AI’s bias extends to a patient’s race, gender, and socioeconomic status. Based on their years of experience, cardiology physicians Doshi and colleagues observe that the method used by AI to diagnose atrial fibrillation discriminates against Black patients. They found that, unfortunately, “many of the most commonly used ECG (electrocardiogram) datasets only report limited demographic data, including the patient’s age, gender, and/or baseline clinical characteristics, without reporting racial or ethnic background.” (Doshi et al. 6). (Scholarly Source 3) Furthermore, Almezadeh and other computer engineer researchers analyzed fourteen years of surgical incident robots. Their findings “included 1,535 (14.4%) adverse events with significant negative patient impacts, including injuries (1,391 cases) and deaths (144 cases), and over 8,061 (75.9%) device malfunctions.” (Almezadeh et al. 7). At first glance, these statistics give cause for alarm. (Refutation: Rhetor’s Main Claim + Support 1) But, the apprehensions outlined above do not have a basis for concern because the creation of AI focuses on making AI the right hand man of the human surgeon. (Toulmin Warrant) Neither computer scientists nor surgeons intend to repeat Adam and Eve’s tragic sin by giving AI full autonomy over surgical procedures and condemning humankind to ruin. (Scholarly Source 1) To illustrate, a group of orthognathic surgeons who have worked with AI in their line of work acknowledge in their article “Artificial Intelligence: Applications in Orthognathic Surgery”, “machines have by far exceeded human capacities in storing and organizing data” (Bouletreau et al. 352). The time that it takes for a single person to become an expert in their field takes decades of acquiring knowledge and experience. In contrast, the time that it takes for an AI to reach the same level of acquired knowledge ranges from a few hours to a few days. It all boils down to surgeons having access to a plethora of information at the touch of a button. (Scholarly Source 2) Another article entitled “Surgery 3.0, Artificial Intelligence and the next-Generation Surgeon” written by Mirnezami and Ahmed, researchers at St. Mark’s Hospital in London, humanizes AI calling it a surgeon’s “colleague” and “a valued member of the surgical multidisciplinary team” (Mirnezami and Ahmed 464). As a vision of the future, imagine walking into the office and seeing, Billy, the AI, sitting at the conference table with other surgeons sharing the knowledge stored in the computer brain. As surprising as it may seem, Mirnezami and Ahmed indicate that things are gearing in this direction. (Scholarly Source 3) Additionally, medical robot researchers Zhou and colleagues explained in “Application of Artificial Intelligence in Surgery” that “with the help of AI, surgical task-oriented HRI (human-robot interaction) allows surgeons to control cooperatively surgical robotic systems with touchless manipulation” (Zhou et al. 424). Usually applications of AI, such as Siri and Alexa, in daily life involve voice commands, but surgical robots have no need for a voice command. These medical AI can recognize and interpret simple hand gestures, gazes, or other types of body language to carry out a needed task. (Conclusion) Evaluation of the facts listed above show that an AI has the capability to reteach itself and correct any error, such as bias.
(Refutation: Support 2) In response to the opponent’s second claim about the danger AI pose while performing surgery, AI’s have been designed to avoid this by predicting possible complications during and after the proposed surgical procedure before the surgeon performs it. (Toulmin Warrant) Surgeons take an oath swearing to do everything in their power to protect human life; therefore, it would not make sense for a surgeon to utilize a tool, such as AI, that could potentially harm a patient on purpose. (Scholarly Source 1) A team comprised of two surgeons and two computer science professionals, Hashimoto et al., reviewed various journal articles to assess the proficiency of AI to predict risks of surgical procedures in “Artificial Intelligence in Surgery: Promises and Perils”. One of the sources they read involved a study where “by using clinical variables such as patient history, medications, blood pressure, and length of stay, [AIs] have yielded predictions of in-hospital mortality after open abdominal aortic aneurysm repair with sensitivity of 87%, specificity of 96.1%, and accuracy of 95.4%” (Hashimoto et al. 73). Considering these odds, AIs have a larger track record in saving lives than in endangering them. (Scholarly Source 2) In agreement with Hashimoto et al., Schwartz and other affiliates at the department of spine surgery in a research school in New York reiterate in their article, “Applications of Machine Learning Using Electronic Medical Records in Spine Surgery”, “multiple studies have demonstrated the ability to build predictive models using EMR data for major perioperative complications in spine surgery, particularly surgical site infections.” (Schwartz et al. 648). Clearly, there is no lack of evidence to support the fact that AIs help save lives; therefore, one can be assured to trust one’s life with one. (Scholarly Source 3) Expanding on AI potential, experts in medicine, Rimmer and colleagues in “The Automaton as a Surgeon: The Future of Artificial Intelligence in Emergency and General Surgery” mention, “AI possesses substantial promise in the field of emergency surgery… to refer appropriate patients through the emergency department efficiently, alert any surgical issues on imaging, and even predict operative risks based on vital sign observations and clinical history to allow a surgeon to provide a personalized risk profile” (Rimmer et al. 760). If surgeons have complete confidence in using AI, even in the extreme case of an emergency surgery, what reason exists for the rest of the human population to distrust AIs? (Conclusion) Therefore, careful planning beforehand can prevent many of the reported accidents of robotic surgeries mentioned by Almezadeh et al.
(Refutation: Support 3) The efficiency of AI in saving hospitals time and money as the third point deserves consideration. (Toulmin Warrant) Time is money, and, extending it to the surgical setting, time dictates a patient’s chance of survival. (Scholarly Source 1) One example that demonstrates AI efficiency involves a paper written by Giovanni and Le Moine, doctors in medical research, entitled “Artificial Intelligence in Medicine: Today and Tomorrow”. They mention called Empatica that ‘received FDA approval in 2018 for their wearable Embrace, which associated with electrodermal captors can detect generalized epilepsy seizures and report to a mobile application that is able to alert close relatives and trusted physician with complementary information about patient localization” (Briganti and Le Moine 3). Instead of having to keep a patient at the hospital for observation, this wristwatch provides all the necessary data, in real time, for a doctor to examine. Otherwise, the time that the patient would have spent in the hospital waiting for something to happen could have been used for another patient in need of immediate attention. (Scholarly Source 2) At the beginning of the COVID-19 pandemic, social distancing put on hold many surgical operations. As a result, an interminable waiting list put the lives of patients in critical condition. Solanki and colleagues, professors of anesthesiology and critical care, observed in the article “Artificial Intelligence in Perioperative Management of Major Gastrointestinal Surgeries” that a solution to this problem proposed the use of an AI algorithm that “optimize[d] the efficiency of operating room booking times that resulted in a reduction in nursing overtime of 21%, which was equivalent to saving of half a million dollars” (Solanki et al. 2763). With medical staff already in scarce quantities, the algorithm helped to alleviate additional strain of the nurses’ workload. While the sanity of healthcare workers was kept at a reasonable level, the hospital also saved in monetary resources. (Scholarly Source 3) In agreement with Zhou et al., plastic surgeons with personal experience using AI in surgery, Liang and colleagues, in “Artificial Intelligence in Plastic Surgery: Applications and Challenges”, have observed that a mutual understanding, or bond, formed between machine and human through natural language processing “allows computers to understand and evaluate statements that are naturally utilized by humans and can help residents sort through vast amounts of information.” (Liang et al. 786). Before when doctors had to write reports about the diagnosis and treatment of their patients, they had to use specific keywords and phrases to organize and save the data for future reference. Now that AI has the capability to understand human emotion through use of language, doctors feel more at ease knowing they can write these reports in a more narrative manner. (Conclusion) Put under the ultimate test of the pandemic, evidence strongly suggests that AI has the capability to adapt and benefit patient care in times of need.
(Summation: Argue that your stance on the issue is best for society) The developing journey of AI over the years has seen improvements in its performance for the medical industry, specifically surgical procedures. Doctors can count on a valuable tool, AI, to provide a patient’s medical information to handing out the necessary surgical instruments with just a simple sweep of the hand. The evidence presented previously give ample examples of successful AI applications from predicting possible surgical risks to efficient time management for critical operations. By meditating on what would have happened if the algorithm mentioned by Solanki et al. designed during the pandemic never came into existence, surely the thought of the drastic consequences is enough to send shivers down the spine. Surgeons fighting for a time slot in the operating room and a higher mortality rate due to patients not given the care needed due to lack of organization does not look like a sight to enjoy. Therefore, the best thing for society is to continue supporting AI research so that surgeons and AI can work together to save lives.

Works Cited

Alemzadeh, Homa, et al. “Adverse Events in Robotic Surgery: A Retrospective Study of
14 Years of FDA Data.” PLOS ONE, vol. 11, no. 4, 20 Apr. 2016, pp. 1-20.,
https://doi.org/10.1371/journal.pone.0151470.
AlHasan, AJMS. “Bias in Medical Artificial Intelligence.” The Bulletin of the Royal
College of Surgeons of England, vol. 103, no. 6, 2021, pp. 302–305.,
https://doi.org/10.1308/rcsbull.2021.111.
Bouletreau, P., et al. “Artificial Intelligence: Applications in Orthognathic Surgery.”
Journal of Stomatology, Oral and Maxillofacial Surgery, vol. 120, no. 4, 26 June 2019, pp. 347–354.,
https://doi.org/10.1016/j.jormas.2019.06.001.
Briganti, Giovanni, and Olivier Le Moine. “Artificial Intelligence in Medicine: Today
and Tomorrow.” Frontiers in Medicine, vol. 7, 5 Feb. 2020, pp. 1–6.,
https://doi.org/10.3389/fmed.2020.00027.
Doshi, Hiten, et al. “Machine Learning in Atrial Fibrillation—Racial Bias and a Call
for Caution.” Journal of Medical Artificial Intelligence, vol. 4, 30 Sept. 2021, p. 6.,
https://doi.org/10.21037/jmai-21-12.
Hashimoto, Daniel A., et al. “Artificial Intelligence in Surgery: Promises and Perils.”
Annals of Surgery, vol. 268, no. 1, 2018, pp. 70–76.,
https://doi.org/10.1097/sla.0000000000002693.
Liang, Xuebing, et al. “Artificial Intelligence in Plastic Surgery: Applications and
Challenges.” Aesthetic Plastic Surgery, vol. 45, no. 2, 2020, pp. 784–790.,
https://doi.org/10.1007/s00266-019-01592-2.
Mirnezami, R., and A. Ahmed. “Surgery 3.0, Artificial Intelligence and the
next-Generation Surgeon.” British Journal of Surgery, vol. 105, no. 5, 2018, pp. 463–465.,
https://doi.org/10.1002/bjs.10860.
Rimmer, Lara, et al. “The Automaton as a Surgeon: The Future of Artificial Intelligence in Emergency and
General Surgery.” European Journal of Trauma and Emergency Surgery, vol. 47, no. 3, 2020, pp. 757–762.,
https://doi.org/10.1007/s00068-020-01444-8.
Schwartz, John T., et al. “Applications of Machine Learning Using Electronic Medical Records
in Spine Surgery.” Neurospine, vol. 16, no. 4, 2019, pp. 643–653.,
https://doi.org/10.14245/ns.1938386.193.
Solanki, Sohan Lal, et al. “Artificial Intelligence in Perioperative Management of Major
Gastrointestinal Surgeries.” World Journal of Gastroenterology, vol. 27, no. 21, 2021, pp. 2758–2770.,
https://doi.org/10.3748/wjg.v27.i21.2758.
Zhou, Xiao-Yun, et al. “Application of Artificial Intelligence in Surgery.” Frontiers of
Medicine, vol. 14, no. 4, 2020, pp. 417–430.,
https://doi.org/10.1007/s11684-020-0770-0.