“THE PROSPECT OF ARTIFICIAL INTELLIGENCE IN HEALTHCARE”

– Ananya Singh (Vivekananda Institute of Professional Studies) 

ABSTRACT:-

As a result of the rising complexity and variety of data in healthcare, artificial intelligence (AI) will be used more frequently. Several types of AI are already in use by payers and providers of care, as well as sciences firms. The key categories of application involves diagnose and treatment recommendations, patient involvement and adherence, and administrative operations are the most common types of applications. Although there are numerous cases where AI can execute healthcare duties as well as or better than humans, implementation issues will prohibit widespread automation of healthcare professional positions for some time. Ethical concerns around the use of AI in healthcare are also addressed.

INTRODUCTION:-

Artificial intelligence (AI) and related technologies are becoming prevalent in business and society and they are starting to be used in healthcare. Many parts of patient care, as well as administrative operations within provider, payer, and pharmaceutical companies, have the potential to be transformed by these technologies.

A number of research studies have already found that AI can perform as well as or better than humans in crucial healthcare activities such as illness diagnosis. Algorithms are already surpassing radiologists in detecting dangerous tumours and advising researchers on how to build cohorts for expensive clinical trials. However, we predict that it will be several years before AI replaces humans in wide medical process areas for a variety of reasons. In this essay, we discuss the potential for AI to automate portions of treatment as well as some of the challenges to rapid AI application in healthcare.

TYPES OF AI OF RELEVANCE TO HEALTHCARE:-

AI technologies for healthcare are outlined and detailed here.

• MACHINE LEARNING – NEURAL NETWORKS AND DEEP LEARNING:-

1. Machine Learning in General:

1. A statistical approach used to fit models to data and learn from it.
2. Common in AI, with a Deloitte poll revealing that machine learning1 was used by 63% of organisations researching AI in 2018.

2. Precision Medicine and Traditional Machine Learning:

1. Precision medicine is the practice of predicting therapy outcome based on patient characteristics and treatment environment.2

2. The majority of healthcare applications employ supervised learning, in which a known outcome variable (e.g., illness onset) is utilised for training.

3. Neural Networks:

1. Neural networks, which have been around since the 1960s, are a more complicated type of machine learning.3

2. Frequently used in healthcare research for classification applications like as illness prediction.

3. Processes inputs, outputs, and weights of variables or ‘features,’ analogous to neuron signal processing but with a poor similarity to the brain.

4. Deep Learning:

1. The most advanced type of machine learning uses neural network models that predict outcomes using many layers of information.
2. Frequently used in healthcare for activities such as identifying possibly malignant tumours in radiological pictures.4
3. Used in radiomics to find clinically significant patterns in imaging data that are not visible to humans.

5. Speech Recognition and Natural Language Processing (NLP):

1. Deep learning is increasingly being employed for voice recognition, which falls within the purview of NLP.

2. Because each characteristic in a deep learning NLP model may be meaningless to a human observer, model results might be difficult to comprehend.

1Deloitte Insights State of AI in the enterprise. Deloitte, 2018. 

2Lee SI, Celik S, Logsdon BA, et al. A machine learning approach to integrate big data for precision medicine in acute myeloid ­leukemia. Nat Commun 2018 3Sordo M. Introduction to neural networks in healthcare. OpenClinical, 2002. 

4Fakoor R, Ladhak F, Nazi A, Huber M. Using deep learning to enhance cancer diagnosis and classification. A conference ­presentation The 30th International 

Conference on Machine Learning, 2013 

6. Applications Beyond Healthcare:

1. While your focus is on healthcare, it’s worth noting that deep learning and machine learning have broad applications across various industries, including finance, manufacturing, and entertainment.

RULE-BASED EXPERT SYSTEMS:-

Expert systems based on collections of ‘if-then’ rules were the dominant AI technology in the 1980s, and they were widely used in the business sector. They were widely used in healthcare for ‘clinical decision support’ reasons throughout the previous several decades5 and are still actively used today. Today, many electronic health record (EHR) providers provide a set of regulations with their systems.

Human experts and knowledge engineers are needed to build a set of rules in a certain knowledge domain for expert systems. They operate effectively up to a point and are simple to grasp. However, when there are a huge number of rules (typically several thousand) and the rules begin to clash with each other, they tend to break down. Furthermore, altering the rules might be complex and time-consuming if the knowledge domain changes. They are gradually being supplanted in healthcare by more data-driven and machine-learning-based techniques.

ROBOTIC PROCESS AUTOMATION:-

Robotic Process Automation (RPA) is a technology that performs structured digital tasks for administrative purposes, mimicking human actions in information systems. It is cost-effective, easy to program, and transparent. Despite the name, RPA involves computer programs on servers rather than physical robots. It relies on workflow, business rules, and integration with information systems’ presentation layers to act as a semi-intelligent user. In healthcare, RPA is used for repetitive tasks like prior authorization, updating patient records, and billing. When combined with technologies like image recognition, RPA can extract data from faxed images for input into transactional systems.5

The trend is toward integrating these technologies, giving rise to AI-driven “brains” for robots and combining image recognition with RPA. The future may see even more intermingling, leading to composite solutions that leverage multiple technologies simultaneously.

5Hussain A, Malik A, Halim MU, Ali AM. The use of robotics in surgery: a review. Int J Clin Pract 2014

DIAGNOSIS AND TREATMENT APPLICATIONS:-

Since the 1970s, AI has been focused on illness diagnosis and treatment. Early rule-based systems like MYCIN had promise but were not widely adopted due to limitations in performance and integration with clinical workflows. IBM Watson, combining machine learning and NLP, initially gained attention for precision medicine in cancer but faced challenges in handling specific cancer types and integration into hospital operations. Open-source programs like Google’s Tensor Flow also posed competition.

AI implementation challenges persist in healthcare, with rule-based systems lacking precision and struggling to keep up with evolving medical knowledge. Research labs and tech companies are making strides in AI applications for disease diagnosis and treatment, often based on machine learning models and big data. Challenges arise in medical ethics and patient/clinician interactions due to the shift towards evidence- and probability-based medicine.

Tech companies and startups are working on various AI solutions, including big data prediction algorithms, picture interpretation algorithms, and genetic profiling for tumor detection and treatment. Population health machine learning algorithms are being adopted by both care providers and payers to forecast populations at risk. However, integration of AI into clinical workflows and EHR systems remains a significant challenge, hindering broad implementation. Some EHR companies are beginning to incorporate AI functionalities, but widespread integration is still in its early stages.

PATIENT ENGAGEMENT AND ADHERENCE APPLICATIONS:-

Patient engagement and adherence are often considered the ‘last mile’ challenge in healthcare, crucial for achieving effective health outcomes.

Role of AI and Big Data:

Big data and AI are increasingly being used to address patient engagement and adherence challenges in healthcare.

Noncompliance Issue:

Noncompliance, where patients do not follow prescribed treatments, poses a significant problem in healthcare.

Survey Insights:

A survey of clinical leaders and healthcare executives revealed that over 70% reported having less than 50% of their patients highly engaged and 42% noted that less than 25% of their patients were highly engaged.

Personalizing Care with AI:

Machine learning and business rules engines are utilized to personalize and contextualize care plans along the care continuum.

Messaging Alerts and Targeted Content:

AI-driven messaging alerts and targeted content are explored to provoke actions at crucial moments in patient care.

Utilizing Patient Data:

Information from various sources, including electronic health records, biosensors, wearables, smartphones, etc., is used to tailor recommendations.

ADMINISTRATIVE APPLICATIONS:-

There are several administrative applications in healthcare. In this arena, the application of AI is less potentially revolutionary than in patient care, although it can generate significant efficiencies. These are required in healthcare since the average US nurse spends 25% of her work time on regulatory and administrative tasks.6RPA is the most probable technology to be applicable to this goal. It may be utilised in a range of healthcare applications, such as claims processing, clinical recording, revenue cycle management, and medical records management.7

Chatbots have also been used in some healthcare settings for patient contact, mental health and wellbeing, and telehealth. These NLP-based apps may be useful for simple operations such as medication refills or appointment scheduling. Patients reported anxiety about providing sensitive information, addressing complex health concerns, and poor usability in a poll of 500 US users of the top five chatbots used in healthcare.8

IMPLICATIONS FOR THE HEALTHCARE WORKFORCE:-

Job Displacement Concerns:

AI raises concerns about job automation, with studies suggesting significant job losses.9

Factors Limiting Loss:

External factors like technology costs, labor dynamics, and regulatory aspects may limit job loss to 5% or less.

6 Berg S. Nudge theory explored to boost medication adherence. Chicago: American Medical Association, 2018. 7Commins J. Nurses say distractions cut bedside time by 25%. HealthLeaders, 2010. 

8 Utermohlen K. Four robotic process automation (RPA) applications in the healthcare industry. Medium, 2018. 

Limited Impact in Healthcare:

Currently, no reported job losses in healthcare due to AI, attributed to slow integration into clinical workflows.

Automation Likely in Digital Roles:

If automation occurs, roles dealing with digital information (e.g., radiology, pathology) may be impacted.

Slow Penetration in Radiology:

Despite AI advancements, slow integration into roles like radiology is expected due to complex tasks.

Challenges in Clinical Processes:

Clinical processes for AI-based image analysis are not ready, with varied focuses and data labeling challenges.

Regulatory and Insurance Barriers:

Substantial changes in regulation and insurance are needed for widespread adoption of automated image analysis.

Unlikely Substantial Employment Change:

Due to complexities, substantial changes in healthcare employment due to AI are unlikely in the next 20 years.

Potential New Jobs:

New jobs may emerge to work with and develop AI technologies.

Impact on Costs:

Static or increasing human employment may mean AI won’t substantially reduce medical diagnosis and treatment costs in the near future.

ETHICAL IMPLICATIONS:-

Finally, there are a number of ethical issues associated with the use of AI in healthcare. In the past, practically all healthcare choices were made by people, and the use of smart computers to make or help with them poses questions of responsibility, transparency, consent, and privacy.

9 UserTesting Healthcare chatbot apps are on the rise but the overall customer experience (cx) falls short according to a UserTesting report. San 

Francisco: UserTesting, 2019 

Transparency is maybe the most challenging issue to overcome with today’s technologies. Many AI systems, particularly deep learning algorithms employed in picture analysis, are very hard to understand or comprehend. If a patient is informed that a photograph has resulted in a cancer diagnosis, he or she would most certainly want to know why. Deep learning algorithms, and even therapists with a thorough grasp of their operation, may be unable to provide an explanation.

AI systems will likely make mistakes in patient diagnosis and treatment, and it may be difficult to hold them accountable. There are also chances that patients will obtain medical information from AI systems that they would prefer to receive from a compassionate practitioner. Machine learning algorithms in healthcare may also be prone to algorithmic bias, such as forecasting a higher risk of disease based on gender or ethnicity when those elements are not truly responsible.10

With AI in healthcare, we may expect significant ethical, medical, professional, and technical developments. Healthcare institutions, as well as political and regulatory organizations, must build systems to monitor major concerns, respond responsibly, and establish governance measures to prevent harmful consequences. Because this is one of the most powerful and impactful technologies to have an influence on human society, it will need ongoing attention and smart policy for many years.

THE FUTURE OF AI IN HEALTHCARE:-

We think that AI will play a key role in future healthcare solutions. It is the key capacity driving the development of precision medicine, which is universally acknowledged to be a much-needed enhancement in treatment. Although early efforts to provide diagnostic and treatment suggestions were difficult, we anticipate that AI will eventually master that domain as well. Given the fast advancements in artificial intelligence for imaging processing, it is likely that most radiology and pathology pictures will be analyzed by a computer at some time. Speech and text recognition are now utilized for patient communication and clinical note recording, and their use is expected to increase.

The most difficult task for AI in many healthcare sectors is not determining if the technologies would be effective, but rather guaranteeing their acceptance in daily clinical practice. For widespread adoption to take place

AI systems must be authorised by regulators, connected with EHR systems, standardized to the point where comparable products operate in the same way, taught to physicians, funded by public or commercial payer groups, and updated in the field. These problems will eventually be addressed, but it will take considerably longer than it will for the technologies to mature. As a result, we anticipate limited adoption of AI in clinical practice during the next five years, followed by widespread usage within ten years.

10 Davenport TH, Dreyer K. AI will change radiology, but it won’t replace radiologists. Harvard Business Review 2018. 

It also becomes increasingly evident that AI systems will not completely replace human physicians, but rather will supplement their efforts to care for patients. Human therapists may eventually shift towards jobs and job designs that need distinctly human talents such as empathy, persuasion, and big-picture integration. Those healthcare practitioners who refuse to collaborate with artificial intelligence may be the only ones who lose their employment over time.

CONCLUSION

Artificial intelligence is indeed a crucial and valuable technology, particularly in addressing the evolving needs of the healthcare industry. It presents promising solutions that can revolutionize the approach to patient care by enabling personalized and tailored treatment strategies. This innovation holds several advantages over traditional analytics and other clinical decision-making tools.

One of the key benefits lies in the precision and accuracy that AI brings to the handling of healthcare data. The ability of AI algorithms to analyze vast datasets allows for more insightful observations in the diagnosis and treatment processes. This heightened level of precision has the potential to significantly enhance patient outcomes, contributing to a more effective and efficient healthcare system.

In essence, the integration of artificial intelligence in healthcare holds the promise of ushering in a new era of personalized medicine, where treatment approaches are finely tuned to meet the unique needs of individual patients. This transformative technology is poised to play a pivotal role in advancing the quality of healthcare delivery.