The effects of AI are evident across industries and sectors. With its incredible applications, AI is slowing paving its way to the healthcare industry. The healthcare industry accounts for 11% of global GDP or $9 trillion annually. AI in the healthcare market is predicted to grow from $14.6 billion in 2023 to $102.7 billion by 2028. According to data by NASSSCOM (National Association of Software and Service Companies), AI has the potential to add about $25 billion to India’s GDP by 2025. According to a report by World Economic Forum, the expenditure on AI in India is expected to reach $11.7 billion by 2025 and add $1 trillion to India’s economy by 2035.
Mr. Niraj Garg, Head of Digital and Automation of Siemens Healthineers, says, “AI aids in saving time by automating mundane and routine tasks within the daily clinical routine. These time-saving measures allow healthcare professionals to focus more on critical decision-making and delivering quality patient care.”
Types of AI and their respective relevance to healthcare
Machine Learning- neural network and deep learning
Machine learning is primarily used in precision medicine to predict what treatment protocols are likely to succeed on a patient based on various patient attributes and the treatment context. Neural Networks have been used for determining whether a patient will acquire a particular disease or not. Deep learning can be helpful in the recognition of potentially cancerous lesions in radiology images. Its applications can also be traced in radiomics, or the detection of clinically relevant features in imaging data.
Natural Language Processing
The dominant application of NLP in healthcare is the creation, understanding, and classification of clinical documentation and published research. NLP systems can analyze unstructured clinical notes on patients, prepare reports, transcribe patient interactions, and conduct conversational AI.
Rule-Based Expert Systems
Rules-Based Expert Systems are widely used for clinical decision support in the healthcare industry. However, they are slowly being replaced by approaches based on data and machine learning algorithms.
Robots
Surgical robots improve the surgeons’ ability to see, create precise and minimally invasive incisions, stitch wounds, and more. However, important surgical decisions are made by human surgeons. Robots are mostly preferred for surgical procedures including gynecologic surgery, prostate surgery, and head and neck surgery.
Robotic Process Automation
In the healthcare industry, they are used for repetitive tasks such as prior authorization, updating patient records, or billing. When combined with other technologies they can be used to extract data to input it into transactional systems.
Additionally, these technologies are being combined and integrated for the purpose of process optimization. Perhaps in the future, these technologies will be so intermingled that composite solutions will be more likely or feasible.
Applications of AI in the Healthcare Industry
Clinical Applications
AI is considered trustworthy in the interpretation of cardiac echocardiograms and it can diagnose attacks better than human physicians in the emergency settings. AI has largely been used in dermatology using contextual images, micro images, and macro images of deep learning. It has applications in diabetic retinopathy. According to a research by the Centers for Medicare & Medicaid Services approved Medicare reimbursement for the use of Food and Drug Administration approved an AI algorithm ‘IDx-DR’ for detecting more-than-mineral diabetic retinopathy, given that it demonstrated 87% sensitivity and 90% specificity. The AI-based InnerEye open-source technology can cut the preparation time for head and neck and prostate cancer by up to 90%.
Precision Therapeutics
The use of AI tools on multimodal datasets it easier to better understand the cellular basis of disease to provide more targeted preventive strategies. The recent announcements of DeepMind and AlphaFold would lead to a better understanding of disease processing, predicting protein structure, and developing targeted therapeutics. Many more drug manufacturing processes and combinatorial optimizations are possible through the use of AI.
Precision Medicines and Drug Development
AI will help in drug development by better predicting early which agents are likely to be effective and also better anticipate adverse drug effects thus avoiding the further development of otherwise effective drugs at a costly late stage in the development process. This will ascertain access to novel advanced therapies at a lower cost.
Patient Engagement and adherence
Messaging Alerts and other relevant targeted content at the right moment is an excellent way in which AI can be used to ensure patient compliance with prescribed treatment plans and drugs. Further, the information provided by EHR systems, biosensors, watches, smartphones, and conversational interfaces can be used to tailor recommendations by comparing patient data to other effective treatment pathways for similar cohorts.
Administration
It has been reported that an average US nurse spends 25% of work time on regulatory and administrative activities. This can instead be replaced by the use of RPA and machine learning for a variety of purposes like claim processing, clinical documentation, revenue cycle management, and medical records management. Additionally, NLP-based applications can be used for making appointments and refilling prescriptions. Chatbots are also being widely used for purposes like patient interaction, mental health and wellness, and telehealth.
Diet recording
Diet recording is commonly operated by the 24-hour recall method. AI can be used to improve the accuracy of diet recording at a lower cost. This will provide better outcomes in the global health landscape.
Implications of AI in the Healthcare Industry
On Workforce
It seems that the healthcare jobs that are most likely to be automated would be those that involve dealing with digital information such as radiology and pathology. But the penetration of AI even in radiology and pathology is likely to be slow. Deep learning models in labs and start-ups are trained for specific image recognition tasks, such as nodule detection on chest computed tomography or hemorrhage on brain magnetic resonance imaging. In practice, thousands of such narrow detection tasks are necessary to fully identify all potential findings in medical images, and only a few of these can be done by AI today. Thus, it can be safely said that AI will not replace human clinicians rather the latter may move towards tasks and job designs that are dependent on human traits like empathy, persuasion, and big-picture integration. Perhaps, only those will lose the jobs who refuse to work in collaboration with AI.
Ethical Implications
Healthcare decisions have been made almost exclusively by humans in the past, and the use of smart machines to make or assist with them raises issues of accountability, transparency, permission, and privacy. Deep learning algorithms used for image analysis – are virtually impossible to interpret or explain. Substantial changes will be required in medical regulation and health insurance for automated image analysis to take off. There are also likely to be incidents in which patients receive medical information from AI systems that they would prefer to receive from an empathetic clinician.
Economical
As per reports around 63 million people face poverty every year due to their healthcare expenses and AI can successfully make healthcare services affordable and accessible. Mudit Dandwate, the CEO and co-founder of Dozee, says, “I foresee the costs are going to be much lesser. For example-Continuous monitoring is only available in ICU beds right now, which is only 1,00,000 beds out of close to about 20 lakh beds in the country. What AI can do is it can monitor any bed at 1/10th the cost of a normal ICU cost. So, in the cost of putting one ICU bed, now you can put 10. That is how the technology creates disproportionate returns when the newest technology comes in.”
Challenges to AI in the healthcare industry
Diagnoses and Treatment
Although rule-based systems incorporated with EHR systems are widely used they lack the precision of more algorithmic systems that can be developed with the use of machine learning. Further, they are difficult to maintain as medical knowledge changes. Also, they are often not able to handle the explosion of data and knowledge based on genomic, proteomic, metabolic, and other ‘omic-based’ approaches.
Data Confidentiality and Privacy
In a survey of 500 chatbot users in the US, for the top five chatbots used in healthcare, patients expressed concern about revealing confidential information, discussing complex health conditions, and poor usability. Thus, stringent privacy standards should be incorporated to ensure patients about safeguarding patients about their data and confidentiality.
Adoption in Daily Clinical Practice
AI will surely be helpful in the healthcare industry like it charms the other industries. However, the adoption of AI in daily clinical practices is considerably challenging. Therefore, to ensure widespread adoption AI systems must be approved by regulators, integrated with EHR systems, standardized to a sufficient degree, taught to clinicians, paid for by public or private payer organizations, and updated over time in the field. These amendments are likely to take longer than the technology itself to mature.
Integrating, the healthcare industry is bound to see some major changes in technology, laws and regulations, and ethics. Thus, it becomes important for healthcare institutions, government, and other regulatory bodies to establish a structure to monitor key issues and react responsibly to them, and establish stringent governance mechanisms to limit the negative implications that the potential introduction of AI could bring to the healthcare industry. Additionally, since AI is a consequential technology to impact society it would require continuous attention and thoughtful policy for many years.