Artificial intelligence (AI) is the simulation of human intelligence processes by computer systems. Such processes include learning and reasoning. Alan Turing, John McCarthy, and Arthur Samuel were just a few of the field’s pioneers in the mid-20th century. In the last few decades, interest in AI has grown tremendously with the rapid developments in computational resources. Machine learning and deep learning are primary AI applications, which have been applied in various domains, including linguistics, healthcare, energy, and education.
The healthcare sector has always motivated innovation. Today, AI is bringing a paradigm shift to clinical decision-making, powered by the increasing availability of healthcare data and the rapid progress of analytics techniques. On the other hand, the WHO reported that the prevalence of chronic diseases is expected to rise by 57% by the end of 2020. Fortunately, AI-driven advancements are anticipated to minimize the financial burden of such diseases.
In 2018, the World Health Organization (WHO) and the International Telecommunication Union (ITU), established a Focus Group on AI for Health (FG-AI4H), which seeks to develop standardised benchmarking processes for AI models in healthcare.
It is clear that the growth of AI in healthcare can not only reduce cost, but it could free up the physicians’ time by taking on repetitive tasks in order to increase productivity, precision, and efficacy. It is noticed that physicians who used AI-supported documentation such as “dictation assistance”, engaged in better with patients than those who did not use such services. Some advancements will improve “Digital Consultation” and “Telemedicine” as patients can use their devices to communicate with their consultants for disease identification and to determine what care is necessary.
Moreover, AI-enabled applications will help physicians provide personalized protocols for patients, contributing to the development of precision medicine. “Personalized treatment” plan, that predicts the patient’s possible response to a particular treatment, and “Medication non-adherence issues” are just examples of the problems that might be affected by the efficiency of AI technologies.
The AI applications in healthcare extend from robotics to predictive analytics. Examples include AI-assisted robotic surgery and smart prostheses for the disabled. On the other hand, some applications have also been developed for computational purposes, such as using data extracted from electronic health record systems. Nowadays, AI is applied in the administrative services in healthcare, ranging from automated scheduling of appointments to predicting the influx of patients at emergency departments. Recently, many innovative technologies have already been investigated within genomics, proteomics, cell biology, stem cell therapy.
Machine learning, particularly Deep Learning algorithms, have recently made huge advances in automatically diagnosing diseases, and reducing errors. Some ambitious systems learn from a combination of data sources (CT scans, MRI scans, genomics and proteomics, patient data, and even handwritten files) in assessing disease and its prognosis. AI has also been adopted in applications within the drug industry, as many of the analytical processes involved in drug development could benefit from improved efficiency.
The feasibility of AI in healthcare is unknown to date since most findings are currently based on retrospective analysis. Yet, recent results illustrate the potential of AI in improving current clinical practice. However, AI may also lead to a decline in employment, loss of data privacy, and a decrease in human interactions. Such risks might cause reluctance in adopting AI technologies.
Genomics is the study of the function and information structure encoded in the DNA sequences of living cells. Since 1953, it has been understood that DNA molecules are the physical storage medium for genetic information. In the following decades, this branch of science has gained popularity and has grown exponentially. The human genome is comprised of approximately 4 billion base pairs, where the order of the pairs determines the information encoded within the genome. DNA sequencing entails learning the exact order of the bases in a strand of DNA. As such, sequencing has become indispensable for basic biological research and diagnosis and opens enormous opportunities for precision medicine. However, the interpretation of the structure, function, and the meaning of the genetic information remains to be a challenge.
Today, the use of AI is being investigated within genomic studies, from structural genomics to functional and regulatory genomics, with practical applications including the prediction of gene expression from genotype, the prediction of genomic association, and the classification of mutations and functional activities
The challenges faced by AI in genomics are mainly related to data. For example, AI applications can only learn efficiently from large datasets. However, such data is difficult to access due to privacy concerns. Scientists and regulatory bodies are working on addressing these challenges, as the tremendous benefits of AI in genomics that will affect the practice of the medical and health sciences such as precision medicine, diagnostics, and cancer therapy cannot be overlooked.
Medical diagnostics are a category of medical specialties designed to help detect medical conditions and diseases. In 2015, the Institute of Medicine reported that “diagnostic errors contribute to approximately 10% of patient deaths, and 6-17% of hospital complications.
In recent years, clinicians, scientists, and medical equipment companies have been collaborating to develop AI-driven diagnostic tools. Machine learning has shown impressive diagnostic capabilities through imaging analysis within radiology, pathology, and dermatology. Other sources of data, such as electronic health records, have also been used for diagnostic purposes. The predictions of machine learning models can be used to support the decision-making process of clinicians. Additionally, the machine learning systems can improve efficiency, overcome the lack of experience in a specific disease area, assist professionals in identifying abnormalities, and reach timely decisions especially within high-burden settings like the emergency department.
Although the diagnostic performance of AI models might appear promising equivalent to that of health-care professionals, some studies highlighted the models’ lack of transparency. Additionally, the performance of the systems relies on the quality of the data, which could in return limit the accuracy of the interpretation of the reported diagnostic findings. Moreover, curating large-scale datasets sourced from various healthcare facilities, insurance companies, and government agencies requires huge investment.
The role of AI in therapeutics could be noticed in discovery, development, and commercialization. Generally speaking, the discovery period of a new molecule is associated with large spending and long timelines. The AI will aim to accelerate drug development by skipping the long R&D cycles, getting life-saving drugs to patients faster and increase clinical success by matching the drugs to the right patient.
Most AI efforts in pharmaceuticals are focused on drug discovery of new effective molecules. The early process of AI assisted drug discovery could largely benefit from machine learning, starting from the initial screening of drug compounds to predicting the success rate of a drug. AI could also assist in identifying the patients for clinical trials in the development stage. This relies on using genomics, molecular, and cellular structure databases, sequencing, and genome editing, to continually teach the algorithm how to match drugs to patients. Ideally, this would translate to lower drug costs for patients and more diverse treatment choices.
Over the past decades, evidence-based research has been exploring the ability of robots to participate effectively in rehabilitation therapy leading to a potential impact on the neuro- recovery following injuries such as stroke. Current trials are testing robotic devices that can act as exoskeletons to assist the elderly, and people with special needs, enabling them to use their limbs.
Robotic surgery or robot-assisted surgery is another proof of how therapeutic Robotics can impact healthcare delivery. It allows surgeons to perform complex procedures with more precision, flexibility, and control. The utilization of this technology enables surgeons to perform a growing number of complex urological, gynecological, cardiothoracic and general surgical procedures.
