Page 1 of 7
Advances in Social Sciences Research Journal – Vol. 8, No. 12
Publication Date: December 25, 2021
DOI:10.14738/assrj.812.11425. Jarrett, J. E. (2021). Update and Acceleration of Health Care Using Artificial Intelligence in Medical Treatments and Diagnostics.
Advances in Social Sciences Research Journal, 8(12). 394-400.
Services for Science and Education – United Kingdom
Update and Acceleration of Health Care Using Artificial
Intelligence in Medical Treatments and Diagnostics
Jeffrey E. Jarrett
Ph.D., Professor Emeritus
Professor, Management Science, Data Analytics and Finance
2021 edition, University of Rhode Island (COB)
ABSTRACT
Researchers support the growth of artificial intelligence and similar methods in
health and medical care for the purpose of continuously improving processes. By
focusing on the growth on data analytics, statistics, applied mathematics, and
computer methods including machine learning, the future of health-care methods
will change. The development of computerized methods and the growth of data
systems produce ample materials for artificial intelligence to develop and to bring
physician assistance programs to enable continuous improvement resulting in
superior health and medical care. This includes applications in intensive care as
well as diagnostic therapies. The focus is on examples in the use of the promising
developments in data science methods, the accumulation of medical and research
data. With quality and continuous improvement in process control applications
where one determines the usefulness of data analytics, there are great possibilities
of change in the improvement in medical applications as well as the management of
medical and health-care treatment and diagnostic facilities.
Key Terms: Data analytics; Artificial intelligence; (AI), Autoregressive moving average
Modeling; Machine Learning; Multivariate Models; Autocorrelation of Data.
INTRODUCTION
Data analysis and analytics is now everywhere one looks, from the production of most
scientifically manufactured component parts to the checkout lines at most supermarkets,
hardware stores, and automatic consumer buying via the internet. We refer to this as
automation, but it is the advances in computer technologies that drove this mechanization of
seemingly simple but technologically advanced tasks to streamline production methodologies.
The growth of these technologies in the future will be accelerated by breakthroughs in artificial
intelligence (AI), which will continue the mechanization of tasks to improve the quality of
output. By including AI into health-care procedures is not simple, but it includes the
methodology of statistical and/or mathematical science as it applies to data-driven
methodologies. In this study, we focus on one such plan that involves the analytics associated
with a volume of diagnostic tests to produce plans to generate treatments.
AUTOMATING THE QUALITY MOVEMENT IN DIAGNOSTICS
Machine learning approaches to problem-solving are growing rapidly within
healthcare, Improvements in diagnostic care, whether in hospitals, treatment and diagnostic
centers, and other health-care units, are a central function of quality health care. In many places,
Page 2 of 7
395
Jarrett, J. E. (2021). Update and Acceleration of Health Care Using Artificial Intelligence in Medical Treatments and Diagnostics. Advances in Social
Sciences Research Journal, 8(12). 394-400.
URL: http://dx.doi.org/10.14738/assrj.812.11425
they are the principal methods by which patients can secure care. The example of Planned
Parenthood clinics is one where patients can receive care and treatment in an affordable and
often convenient manner for men’s health care. Planned Parenthood provides services that
often are not available to those who do not have sufficient (or even any) places to obtain
affordable care. A client with a severe set of conditions enters the clinic to have scientific tests
performed in order to ascertain or determine a diagnosis and therapeutic plan to produce a
treatment to successfully reduce the problem and achieve positive results. The process includes
a total quality movement (TQM), which is a plan to achieve a successful outcome to the patient’s
health problem. In the future, we expect machine learning, applications of statistical science, AI
and TQM to spread everywhere. Just look at the current research in automobiles and the
relative changes made by the driverless vehicle.
To consider the depth of AI and modern data analytics topics in health care, refer to the
publications by Jarrett (2008) and (Jarrett and Pan 1981, 2015, 2016). In addition, see Patel et
al. (2009), Machado and Costa (2010), Khoo and Quah (2003) and more recently, Acampora et
al. (2013) added specific {in addition, Radiol (2019) specified other improvements in 2019 of
new computer-based methods. Technology firms such as Google, Amazon, Microsoft, and Apple
in recent years made huge investments in AI to deliver tailored search results and build items
called personal virtual assistants. The methodology is seeping down to hospital care and other
forms of diagnostic and treatment methodology in health care in general. With reforms in
health care and health-care reform law, AI will assist physicians and other health-care
personnel in choosing medicines and treatments for patients in an efficient and timely manner.
For example, a physician who has a patient with a particular diagnosis will be able to choose
the best medicine to counter the effect of a severe ailment quickly. With the huge number of
medicines available for a physician to prescribe, much decision making will be automated
thanks in part to the push for computer systems to prescribe the best treatment available from
medical science. No longer will a physician need to peruse volumes of databases to find the
optimal treatment. The computer will find and inform health-care personnel to act quickly and
optimally.
Today, data collection by health statisticians includes volumes of patient demographic and
clinical and billing data, which are in an electronic format for analysis by intelligent software.
For these difficult tasks, AI software can analyze quickly to perform the tasks of recommending
medicines, treatment protocols, and general advice to assist physicians in attacking the
problems associated with difficult diagnoses. For example, applications of AI have been utilized
in intensive care for nearly a generation (Hanson and Marshall 2001; Liu and Salinas 2017). In
other examples, new digital devices and home tests are allowing a more thorough patient
examination remotely, which addresses some of the previous setbacks of telemedicine. Remote
diagnostic tools such as perception of telemedicine. Heartbeat and respiration rate can now be
checked remotely. The same is true for blood pressure, blood glucose, body temperature, and
oxygen levels. A device may contain a high-definition camera that can look down the throat and
ear canals. Cameras can also provide high-resolution images of the skin to examine lesions,
suspicious skin changes, and other dermatological problems. Urine-testing kits may also be
employed in the home or specific diagnostic centers to provide information to medical
personnel to suggest a treatment without the patient being in the same physical location as the
medical personnel.
Page 3 of 7
396
Advances in Social Sciences Research Journal (ASSRJ) Vol. 8, Issue 12, December-2021
Services for Science and Education – United Kingdom
At this point, we should consider automated statistical quality control (ASQC) or automated
statistical process control (ASPC) as it applies in TQM. These terms are no longer new in
diagnostic and treatment terminology; however, they are based on previous applications in
industry, banking, and everywhere one seeks assistance in the analysis of data where the timing
of decisions is very important. TQM is the field that ensures that management maintains
standards set and continually improves the processing of successful goals and achievements.
Instead of final, end-of-service inspection (whether the patient is found healthy or not after the
treatment ends). TQM according to Lee and Wang (2003) and Weihs and Jessenberger (1999)
provides. Instead of end-of-service inspection and decision making, TQM emphasizes
prevention, integrated source inspection, process control, and continuous improvement. The
mitigating of risks of type I and type II errors are the prime purpose of these methods. In
addition, AI will provide software, services, and analytics solutions to the ambulatory care
market. In a health-care information technology and services company that delivers the
foundational capabilities to organizations that want to promote healthy communities. The
technology provides a customizable platform that empowers physician success, enriches the
patient care experience, and lowers the cost of health care and, in turn, health insurance. Stated
simply, AISQC monitors the incidence characterized by the results of multiple tests on a similar
fluid per period of a short interval over a lengthy period (e.g., 10–20 weeks). The monitoring
requires an intelligent system analyzing items (e.g., control charts) and seeking whether there
are common causes of variation or special causes of variation. In industrial applications, these
were called Shewhart charts. Later, others suggested additional methods including the use of
exponentially weighted moving average (EWMA) control charts (See Griggs and Spiegelhalter
2007).
The great rise of health information systems enables AI in the very early stages of its
development to match one’s own intelligence. Computers certainly cannot diagnose like
physicians; however, AI software and computer technology are capable of processing vast
amounts of data and identifying patterns that humans cannot. AI solves the complex algorithms
that analyze these data and is a useful tool to take full advantage of electronic medical records,
transforming them from mere e-filing cabinets into full-fledged physician analysts that can
deliver clinically relevant, high-quality data in real time.
AISPC AND AISQC IN HEALTH-CARE ENVIRONMENTS
SPC/SQC environments usually assume a steady process behavior where the influence of
dynamic behavior does not exist or is ignored. The focus of control is where there is only one
variable (e.g., medical test) over a lengthy interval of time. SPC controls for the changes in either
the measure of location or dispersion, or both. These procedures as practiced in each phase
may disturb the flow of the service production process and operations. We note that in recent
years the use of SPC to address processes characterized by more than one test or treatment
emerged. First, we review the basic univariate procedures to improve the process of SPC and
allow AI to enter the process.
Shewhart control charts were the central foundation of univariate (one variable) SPC, which
has a major flaw. The process considers only one piece of data, the last data point, and does not
carry the memory of the previous data collected. Often, a small change in the mean of a random
variable is not likely to be detected quickly (Griggs and Spiegelhalter 2007). EWMA control
charts improved upon the detection process of small process shifts. Rapid detection of
Page 4 of 7
397
Jarrett, J. E. (2021). Update and Acceleration of Health Care Using Artificial Intelligence in Medical Treatments and Diagnostics. Advances in Social
Sciences Research Journal, 8(12). 394-400.
URL: http://dx.doi.org/10.14738/assrj.812.11425
relatively small changes in the characteristic of interest and ease of computations through
recursive equations are some of the important properties of the EWMA control chart that
makes the process attractive and easy to use intelligent software to detect changes.
The EWMA chart is used extensively in time-series modeling where the data contain a gradual
drift (Box and Draper 1998). EWMA provides for identifying gradual shifts in medical tests by
predicting where the observation will be in the next period of time. Hence, the EWMA process
improves decision support in the future and is dynamic (Hunter 1986). The EWMA statistic for
monitoring the results of lengthy period of tests having short interval when the actual tests are
performed. Furthermore, the method gives less and less weight to data as they become more
remote in time. Montgomery’s (2013) work contains the development of models for finding
control limits in this univariate process but appears to be another example of where intelligent
software applies.
ADDITIONAL APPLICATIONS USING UNIVARIATE MODELS
In many applications of univariate analysis, the sample size used in the test process is one.
Stated differently, the sample consists of an individual unit the control chart for a sample of one
(the individual chart) employs a moving average of two successive observations to estimate the
process variability. Obviously, small samples lead to incorrect decisions (stated as an increase
in the probability of a type II error) point out problems and issues associated with statistically
based evaluations which must be included in intelligent software. A solution may be provided
by examining the average run length (ARL) of a proposed solution over a variety of alternative
process shifts. ARL performance for an in-control state and for a single shift in a process for
which the proposed detection program optimizes must be evaluated. If the system is not
optimized, misplaced control limits may result. The system for detection of shifts is sub
optimized, and better techniques should be sought. Yeh and Hwang (2004) suggest processes
whereby the units are dynamic. In provider-of-care treatments, the distinction between phases
I and II of SQC solutions is often not clear. Hence, ARL is often the choice used to assist the
providers of care with the assistance they need to recommend courses of treatment.
Alwan (1992) finds that the great majority of SPC applications studied result in control charts
with misplaced control limits and essentially false signals to the providers of care. The
misplacement results from autocorrelated process observation. The autocorrelated time-series
observations violate an assumption associated with Shewhart control charts (Woodall 2005).
Autocorrelation of process observations is common in many applications—for example, cast
steel (Alwan 2000), wastewater treatment plants (Berthouex et al. 1978), chemical processes
many other processes in the health-care industry, especially diagnostic care and similar
applications. In addition, suggest using autoregressive integrated moving average (ARIMA)
charts for decision analysis. Continuous intelligent software can be of particular aid to
identification of the appropriate methods for decision analysis if one follows the works of
Atienza et al. (1998), Box et al. (2008), and West et al. (2002) who employed ARIMA modeling
with intervention. In addition, Jarrett (2016a, 2016b) summarize many of these method in SPC.
All these models are in the process of being computerized to develop intelligent systems that
will enable computers intelligently point to optimal patient treatments and diagnoses. The
notion of physicians having patient-centered diagnostic programs using AI will be of immense
help.
Page 5 of 7
398
Advances in Social Sciences Research Journal (ASSRJ) Vol. 8, Issue 12, December-2021
Services for Science and Education – United Kingdom
MULTIVARIATE QUALITY CONTROLS (MQC)
Multivariate methods use additional analyses due to having two or more variables that are the
results of several diagnostic procedures to determine a specific plan of care (treatment). The
use of univariate analysis may lead to incorrect interpretation of data due to the cointegration
of the tests performed. The most popular multivariate methods (MQC) are those based on the
Hotelling T2 distribution (West, et al. 2002; Woodall 2005; Yang and Rahim 2005) and
multivariate exponential moving average method (MEWMA; Elsayed and Sastri 2007). Other
approaches, such as control ellipse, apply for the case of two correlated variables. There are
other MQC methods including those developed by Kalagonda and Kulkarni (2003, 2004), Jarrett
and Pan (2006, 2007a, 2007b, 2013), Vanhatalo and Kulachi (2015. All the aforementioned
MQC modelers produced results that achieve superiority to SQC analysis because of one or
more of the following factors:
1. The control region of variables is represented by an ellipse rather than parallel lines.
2. The intelligent software is programmed to maintain a specific probability of a type I
error in the analysis.
3. The determination of whether the process is out of control is a single control limit
(ARL).
4. By correcting T2-based MQC analysis, autocorrelation is present. In the data
5. Use of MEWMA, when time-series methods have unique schemes.
As a result, the above methods indicate that intelligent software cannot ignore the various
possibilities to lead to non-optimal decisions. However, proper AI methods will adjust to new
research, and patient-assisted analytical software will be of great use to find diagnoses that
enable one to use AI to solve difficulties with patient care.
LIMITATIONS OF AI, MACHINE LEARNING AND SIMILAR TOPICS
AI, which uses all the methods discussed, is dependent on data science, scientific sample, and
statistical analysis. One of the great problems is that AI has yet to come to grips with the huge
problems associated with the presently insurmountable problem of language. AI professionals
may develop a huge amount of algorithms by combining a small number of mathematical
symbols and, in turn, following a small set of rules. Similarly, one can develop an enormous
amount of sentences by utilizing a relatively modest number of words and rules. A realistic and
useful AI system still needs to cope with the challenges associated with all the possible
sentences that may be created in the conversations developed in the AI interrogatory. Genuine
AI systems in health care need to have simple and realistic combinations of questions and
interpretations that are easily understood and do not require the finite possibility of many
interpretations of results. AI will have problems when correct solutions relate to what is “most
likely true.” Hence, interrogatories must be tight and simple such that AI cannot rely on
insufficient interpretations of questions and answers (see Cambria and White, 2014).
As of this time, the dominant approach to AI is not working out. There is no reason to believe
that researchers in AI should return to the projects of making machines actually share some of
human’s cognitive abilities. Human cognition could be built into machines applications as there
is flexibility in human thought that is goal. Approaching AI has been built into Google Translate
and Google Duplex. The limitations of these applications as a form of human intelligence should
alert developers. If machine learning and what is entitled “big data” cannot deliver any further
Page 6 of 7
399
Jarrett, J. E. (2021). Update and Acceleration of Health Care Using Artificial Intelligence in Medical Treatments and Diagnostics. Advances in Social
Sciences Research Journal, 8(12). 394-400.
URL: http://dx.doi.org/10.14738/assrj.812.11425
than a ticket to a Broadway Show in the hands of the most capable AI firms and developers, it
is time to reconsider the strategy associated with AI development.
SUMMARY AND CONCLUSION
The purpose of this study is to encourage growth in a very important industry called artificial
intelligence. AI-based platforms for digital transformation will play an increasing role in patient
diagnosis health programs. The growth will occur in treatment and emergency care centers as
well as intensive care units. Intelligent software is being developed, which will suggest to
physicians and other health-care professionals the meaning of studying databases of
information data analytics. In turn, intelligent software will prescribe and set protocols for
treatments of difficult prognoses and intensive care. Intelligent programs are AI-based
platforms for digital transformation. They are modular and an interconnected mixture of
flexible digital technologies that span from robotic automation to machine learning. The
programs learn over time and produce new ways to arrive at results. The study indicates new
ways to get results and in a timely fashion. The blending of intelligent software and
comprehensive data analytics will eventually move health-care analysts from the task of
interpreting results to have protocols produced for them. Intelligent software will blend
seamlessly with a decision maker’s operational insights and produce unique domain expertise
to create better analytical conclusions in the real world. By examining quality operations, we
observe how AI shares the burdens of care and assists health-care personnel in achieving their
goals. As stated before, AI in health care incorporates AI into many health-care procedures that
are not simple, but includes the methodology of statistical/ mathematical science as it applies
the data-driven methodologies.
References
Acampora, G., Cook, D.J., Rashidi, P., and Vasilakos, A.V. (2013). “A Survey on Ambient Intelligence in Healthcare.”
Proceedings of the IEEE, 101, 2470–2494. Doi: 10.1109/JPROC.2013.2262913.
Alwan, L.C. (1992). “Time-Series Investigation of Subsample Mean Charts.” Communications in Statistics—Theory
and Methods, 21 (4), 1025–1049.
Alwan, L.C. (2000). Statistical Process Control. New York: Irwin-McGraw-Hill.
Atienza, O.O., Tang, L.C., and Ang, B.W. (1998). “A SPC Procedure for Detecting Level Shifts of Autocorrelated
Processes.” Journal of Quality Technology, 30, 340–351.
Berthouex, P.M., Hunter, E., and Pallesen, I. (1978). “Monitoring Sewage Treatment Plants, Some Quality Control
Aspects.” Journal of Quality Technology, 10 (4), 149–159
Box, G.E.P., and Draper, N.R. (1998). Empirical Model Building and Response Surfaces. New York: Wiley.
Box, G.E.P., Jenkins, G.M., and Reinsel, G.C. (2008). Time Series Analysis, Forecasting and Control, 4th Ed. New
York: Wiley.
Cambria, E., and White, B. (2014). “Jumping NLP Curves: A Review of Natural Language Processing Research
[Review Article].” IEEE Computational Intelligence Magazine, 9 (2), 48–57. doi: 10.1109/MCI.2014.2307227.
Elsayed, E.A., and Sastri, T. (2007). “Design of Optimum Simple Step-Stress Accelerated Life Testing Plans.”
Recent Advancement of Stochastic Operations Research, edited by S. Dohl, T.S. Osaki, and K. Sawaki. Singapore:
World Scientific.
Griggs, O.A., and Spiegalhalter, D.J. (2007). “A Simple Risk-Adjusted Exponentially Weighted Moving Average.”
Journal of the American Statistical Association, 102 (477), 140–152.
Hanson, C., and Marshall, B. (2001). “Artificial Intelligence Applications in the Intensive Care Unit.” Critical Care
Medicine, 29 (2), 427–435.