Artificial Intelligence (AI) in Healthcare & Medical Field

Shrankhla Saxena1(corresponding author)

IET-CSE , Mangalayatan University, Aligarh, India. [email protected]

Javed Wasim2
IET, Faculty of Engineering & Sciences, Mangalayatan University, Aligarh, India.
[email protected]

Abhinav Juneja3
Computer science & information technolgy, KIET group of institutions, Ghaziabad, India.
[email protected]

ABSTRACT: Different technologies are adopted and experimented with increasing automation
in the medical field. Now a day, AI is being introduced in the medical field to keep a medical
record in digital format and conduct patient checkup using smart technologies. It provides
solutions, especially in targeted treatments, uniquely composed drugs and personalized
therapies. AI is an innovative technology that helps to guide the surgeon during medication,
treatment and operation. The main application of this technology is for better decision making
for complicated cases. It can also help to track, detect, investigate and control the infection in
the hospital. This technology develops and optimized online patient appointment platform. In
future, it will helpful in all medical areas to serve humanity.

Keywords: Test case prioritization, Genetic algorithm, Machine learning.

1. INTRODUCTION
In the current scenario, AI seems the best technology which can be used for providing higher
life expectancy. It provides AI-assisted robotic surgery for a complicated case. This technology
creates information through different virtual support and regularly communicates with the
patient. In rural areas, there is a shortage of healthcare providers and this technology can be
gainfully employed to fulfil this shortage. It improves the quality of medical students to fulfil
any urgent demand for rural areas8,9. This technology not only improves the efficiency of
health professionals but also improves the quality of healthcare service at a lesser cost. It
provides advice to doctors towards accurate diagnosis10,11.
Artificial Intelligence plays a significant role in scanning technologies like X-rays, Computed
Tomography (CT), Magnetic Resonance Imaging (MRI) and 3D scanners. These are helpful to
make a better decision regarding the patient. AI suggests proper diet and eating habits for better
health. It efficiently manages patient booking and reminds about doctor’s appointment12-14.
This technology is also helpful in virtual interaction with doctors and with its deployment, the
medical field becomes efficient to solve various challenges
.
2. BACKGROUND AND RELATED WORK
The goal of prioritizing the test cases is to sort test case as per their priority so that maximum
faults can be detected in less time hence decreasing the cost of regression testing also. Since,
testers need to run the test cases first to know the fault detection efficiency of any TCP
technique, therefore researchers have proposed various coverage- based metrics for test
case prioritization. Researchers have distributed coverage-based TCP methods into two main
classes- metrics for white box and black box. White box coverage metrics mainly include
statement coverage, block coverage, branch coverage, function coverage etc. Researchers have
also proposed various methods for black box coverage metrics such as t-wise technique,
calculated diversity of inputs with help of NCD, Jaccard distance, Levenshtein distance etc.

XXX-X-XXXX-XXXX-X/XX/$XX.00 ©20XX IEEE

After selecting criteria for prioritization for most of the above-mentioned methods, test cases are
sorted by greedy algorithm.

Figure 1. Methodology of TCP Technique

Luo et al. [4] performed study on many real-world programs to analyze TCP techniques and
provides some important observations and results. This study includes two metrics of
effectiveness which are average percentage faults detected (APFD) and cost-cognizant APFDc.
Researchers in this study used dynamic techniques for TCP using statement level coverage and
found that static techniques are better and give better results while using APFDc. Their results
are based on study and comparison of four static TCP methods which are call graph based
method, string distance based method, topic model based method with four dynamic TCP
methods which includes total greedy, additional greedy, adaptive random and search based
methods. Their results show that values of APFDc and APFD are more consistent at the level of
test class than the level of test method.
Khatibsyarbini et al. [6] performed review of 110 papers and found the recent trends in machine
learning(ML) for test case prioritization(TCP). This paper represents the increasing use of ML
for TCP since 2004. They also found that after 2004 some new ML approaches have been
introduced for TCP. ML taxonomy for TCP is given in figure 2.

Figure 2. Taxonomy of Machine Learning

After review of 110 papers, it was found that classification techniques have been mostly used
ML techniques for TCP. In classification methods K nearest neighbor and decision tree are most
popular techniques. In regression techniques, linear regression, regression trees and support
vector machine (SVM)

3. TEST CASE PRIORITIZATION USING GENETIC ALGORITHM(GA)

Genetic algorithm is an evolutionary approach of ML which is inspired by the rule of nature i.e.
survival of the fittest[7]. It is a heuristic search method which is used to solve optimization
problem in ML i.e. problems which need long time to solve. Process of Genetic algorithm is
given in figure 5.

4. HYPERVOLUME INDICATOR
Hypervolume indicator is mostly used in evolutionary multiple objective optimization[11]. It can
measure the amount of the affected portion of the work space. It includes the most dominating
features of pareto front. The role of hypervolume indicator is to direct the search in multiple
objective optimization and evaluate its performance. Problems of multiple objective optimization
are converted to single objective optimization with the help of hypervolume indicator, since it
finds the finite number of solutions and increases the hypervolume parameter[10].
6. RESULTS
Artificial intelligence can help to monitor and proper treatment for patients. It makes an
assessment in image/result without the requirement of doctors, surgeon and clinician. AI-based
technologies provide decision which can help to predict medical emergencies. It is helpful to
provide medical consultation through a digital app. The practical implication of this technology
is to increase accuracy in diagnosis and treatment. It can be used in the prevention of disease
and reduce medical cost.

7. CONCLUSION AND FUTURE SCOPE

This technology is used to answer patient questions and reduce unnecessary hospital visits. AI
provides an excellent source and identifies problems during the shortage of doctor. This
technology first understands the biological origin of disease for better treatment of the patient. It
easily imagines medical images and accurately performs the task. AI helps identify cancer and
its treatment with the analysis of patient data. Its applications are for the diagnosis of heart
disease. It speeds up the clinical trials to produce a conclusive result. AI constructs analytical
algorithms of different features from the patient data, which is helpful to provide information
about the patient, survival times and diseases level. In upcoming years, its applications will be
for digital supervision in hospitals to improve patient care.
Table 1. comparison of HGA with other ML techniques

S Technologies Description References

No
1 Recording and  It collects, stores and analyses the medical data to [61-65]
storage of provide faster access and decision making
medical data  Every patient data is electronically stored which
facilitates diagnosis and treatment
 Provide day to day improvement history of the
patient
 Data stored digitally can help to identify the cause of
diseases and help research and development activity
 It records the all medical record of an individual
patient and further compares the database of illness

2 Analyze  Artificial intelligence accurately analyse different [66,67]

different test tests like X-ray, ultrasound, MRI and CT scan
 Capability to check the improvement and significant
causes of the disease
 Quickly shares patient information in the emergency
case which makes doctors and surgeon job easy
 Efficiently perform, evaluate, validate, predict and
analyse the data using different scanning
technologies

3 Patient  It helps to monitor patient condition and follow up all [68,69]

monitoring treatments
 Provide proper monitoring to get more exercise,
needs and habits of the patient
 Helps in monitoring and access information of the

REFERENCES
[1] D Nucci, D., Panichella, A., Zaidmain, A. and De Lucia, A. 2020. “A Test Case Prioritization Genetic
Algorithm Guided by the Hypervolume Indicator,” in IEEE Transactions on Software Engineering, vol. 46,
no. 6, pp. 674-696, doi:10.1109/TSE.2018.2868082.
[2] Bagherzadeh, M., Kahani, N. and Briand, L. 2022. “Reinforcement Learning for Test Case
Prioritization,” in IEEE Transactions on Software Engineering, vol. 48, no. 8, pp. 2836-2856, doi:
10.1109/TSE.2021.3070549.
[3] Lachmann, R., Schulze, S., Nieke, M., Seidl, C. and Schaefer, I. 2016. “System-Level Test Case
Prioritization Using Machine Learning,” 15th IEEE International Conference on Machine Learning and
Applications (ICMLA), Anaheim, CA, USA, 2016, pp. 361-368, doi: 10.1109/ICMLA.2016.0065.
[4] Luo, Q., Moran, K., Zhang, L. and Poshyvanyk, D. 2018. “How Do Static and Dynamic Test Case
Prioritization Techniques Perform on Modern Software Systems? An Extensive Study on GitHub
Projects,” in IEEE Transactions on Software Engineering, vol. 45, no. 11, pp. 1054-1080, 1 Nov. 2019,
doi: 10.1109/TSE.2018.2822270.
[5] Satpathy, S., Khalaf, O. I., Shukla, D. K., Algburi, S., & Hamam, H. (2024). Consumer electronics
based smart technologies for enhanced terahertz healthcare having an integration of split learning with
medical imaging. Scientific Reports, 14(1), 10412.
[6] Khatibsyarbini, M., Isa, M. A., Jawawi, D. N. A., Shafie, M. L. M., Wan-Kadir, W. M. N., Hamed, H.
N. A., Suffain, M. D. M. 2021 “Trend Application of Machine Learning in Test Case Prioritization: A
Review on Techniques, “in IEEE Access, vol. 9, pp. 166262-166282,
doi:10.1109/ACCESS.2021.3135508.
[7] Sabharwal, S., Sibal, R. & Sharma, C. 2014. Applying Genetic Algorithm for Prioritization of Test
Case Scenarios Derived from UML Diagrams., Int J Comput Sc Issues. 8.
[8] Bajaj, A. and Sangwan, O. P. 2019. “A systematic literature review of test case prioritization using
genetic algorithms,” in IEEE Access, vol. 7, pp. 126355-126375, doi: 10.1109/ACCESS.2019.2938260.
[9] Deb, K., Pratap, A., Agarwal, S. and Meyarivan, T. 2002. “A fast and elitist multiobjective genetic
algorithm: NSGA-II,” in IEEE Transactions on Evolutionary Computation, vol. 6, no. 2, pp. 182-197, doi:
10.1109/4235.996017.
[10] Anne, A., Johannes, B., Dimo, B. and Eckart, Z. 2009. Theory of the Hypervolume Indicator: Optimal
µ-Distributions and the Choice of the Reference Point. Proceedings of the 10 th ACM SIGEVO Workshop
on Foundations of Genetic Algorithms, FOGA’09. 10.1145/1527125.1527138.