Summary of Amisha et al.’s “Overview of artificial intelligence in medicine”

TO: Professor Jason W. Ellis

FROM: Tasnuba Anika

DATE: Sept 18, 2020

SUBJECT: 500-Word Summary

            Artificial intelligence (AI) is one of the fields of the computer science that highlights the fabrication of intelligent machine that performs different tasks and acts like humans. There are many artificial intelligence methods like fuzzy expert systems, Bayesian networks, artificial neural networks, and hybrid intelligent systems are playing significant role in health care.

            AI in medicine are divided into two categories such as virtual and physical. The virtual section deals with securing patient’s medical history into an electronic database or providing neural network-based recommendation to the physicians. The neural networks on computer process information and can suggest solution to the critical problems like human brain. Using robots for surgeries, intelligent prostheses for disable people are the physical usage of AI.

           Computers can identify patient’s health problems in two different ways. One is flowchart style and the other one is database. In the flowchart style doctors must provide the common syndromes and diseases that patients usually experience into machine-based cloud networks. Afterwards when the physician put the syndromes in the machine that a patient is facing, it shows what kind of disease that patient has. However, this approach of diagnosing patient’s sickness is not very useful. Sometimes the machine cannot spot the illness. On the other hand, database technique has proven to be very functional. As it deploys the object identification systems. In this system computers are fed thousands of clinical or radiological images, different types of diseases and symptoms. In this way computers can recognize patient’s illness with 75% accuracy.

          The usage of AI is very broad in medical field. AI has been utilizing in online scheduling, digitalization of medical records, prescribing medicine dosage, immunization dates for children and pregnant women and so on. Computer-assisted diagnosis (CAD) is getting popular in assessing mammography. In CAD process it indicates whether the screening has been done appropriately or not. It also points out negative tests in X-rays or MRIs specially in hospitals which have fewer staffs. Moreover, there is a system called Germwatcher which spots and examine germs and infections. With help of an online course called AI-therapy now patients can treat social anxiety. Besides that, doctors are now using robotic arms that can imitate human hand movement and provides 3D view with zooming options. Robotic arms help surgeons to perform precise surgical cuts. In addition, health trackers like Fitbit and apple track human activities like heart rate, sleeping and so on. In this way doctors could get better insight of patients’ health.

         The use AI is rapidly increasing for which physicians and medical trainees should get themselves familiarized with the modern technologies and get trained properly. As the author in this article mentioned “The goal should be to strike a delicate mutually beneficial balance between effective use of automation and AI and the human strengths and judgment of trained primary care physicians.” This is basic since AI totally supplanting people in the field of medication is a worry which may somehow, or another hamper the advantages which can be gotten from it. Machines can never perform critical thinking, creativity, or emotional intelligence like human. That is why patients would still need doctors to treat their disease.

                                                           References

Amisha, Malik, P., Pathania, M., & Rathaur, V. (2019). Overview of artificial intelligence in medicine. Journal of Family Medicine & Primary Care, 8(7), 2328–2331. https://doi.org/10.4103/jfmpc.jfmpc_440_19

Summary of Huda et al.’s “Mobile-Based Driver Sleepiness Detection Using Facial Landmarks and Analysis of EAR Values”

TO: Professor Jason W. Ellis

FROM: Mahim M. Pritom

DATE: Sept 17, 2020

SUBJECT: 500-Word Summary

This memo is a 500-word summary of the article, “Mobile-Based Driver Sleepiness Detection Using Facial Landmarks and Analysis of EAR Values,” by Choirul Huda, Herman Tolle & Fitri Utaminingrum, researchers at Brawijaya University, Malang, Indonesia.  

The article discusses the facial recognition system capable of identifying closed eyes and generating results of the Eye Aspect Ratio(EAR) by analyzing Facial Landmark points with an accuracy rate of about 92.85%.

According to the studies, sleepiness while driving causes injuries from minor to life-threatening that results in death in both developed and developing countries. Even though automotive industries started to develop special devices designed to keep drivers awake while driving, only certain brands of vehicles such as Mercedes-Benz, BMW, Volvo have the technology leaving the people driving a vehicle equipped with simple technology still at risk of sleepiness driving-related injuries due to the expensive production cost.

Rahman et al. conducted a study to identify sleepiness based on eye blink analysis by using a webcam connected to a computer. He used the Viola-Jones method to detect the eye area of the face using the Haar-like feature algorithm that helps to expose the sleepy eyes and evaluating the frame of the closed-eyes caught on camera. Jacobe et al. conducted another study utilizing two models of Artificial Neural Network(ANN) to predict sleepiness in a driving simulation. Even though both studies gave satisfying results, there were problems such as converting the color of each frame, inadequate detection process, and obstacles to install the camera inside the car. Mohammad et al. utilized the Haar Cascade Classifier by using mobile to detect and observe the eye sclera both closed and opened situation to determine the color reflecting back. Although this method can generate high accuracy result, there are some limitations such as the driver’s face must be aimed directly towards the camera, needs further development to recognize eye areas precisely. Another method was proposed by Jabbar et al. to detect drowsiness by combining Deep Neural Networks and extraction of facial points. Although this system could identify the drowsy drivers by 81%, the application process of this method requires a computer with high specifications because of the complex learning process which is very expensive and time-consuming.

Soukupova and Cech used the EAR method to measure the facial landmark in the eye area using Euclidean Distance that helps to recognize the blinks using 6 different landmarks. Since the number of extraction points applied in this study was limited, the most influential points in closed eye detection are selected to speed up the computing process. The authors, on these limited extraction points, emphasizes, “The Facial Landmark method qualified to identify driver sleepiness accurately based on analysis of EAR (Eye Aspect Ratio) values utilizes the extraction of 4 points around the eye area” (Huda et al., 2020, p. 27). The tree Regression method is applied to complete the precision procedure of incompletely processed points that are produced during the extraction process. The library is used to determine the number of dots generated on the face. Face detection is utilized as the library to detect the extracted facial landmark dots using a face detector developed by Google. Soukupova and Cech assigned a threshold value of 0.20 compared with the EAR resulting in opened eyes if EAR>0.20 or closed eyes if EAR<0.20.

Reference

Huda, C., Tolle, H., & Utaminingrum, F. (2020). Mobile-Based Driver Sleepiness Detection Using Facial Landmarks and Analysis of EAR Values. International Journal of Interactive Mobile Technologies, 14(14), 16–30. https://doi.org/10.3991/ijim.v14i14.14105