Artificial Intelligence In Healthcare - Are We Ready For A Paradigm Shift?

Dr. Sanjay Dhawan, Group Director (Radiology), Clearmedi HealthcareHaving completed his FRCR ­ Medicine from The Royal College of Radiologists, London, Dr. Sanjay has over a decade of experience in the healthcare industry. He is also a Fellow Royal College of Radiologists (FRCR) member and holds the Certificate of Business Excellence from Haas School of Business, UC Berkeley.

Digital revolution in healthcare has been fired by a variety of economic and demographic factors, most of all by technological developments. This fact has been born by the huge investments which giants like Google, IBM and Microsoft Corporation have made in this space. In this domain of digital revolution in healthcare, AI is the newest kid in the block.

Artificial Intelligence (AI), by definition, means to substitute the cognitive and analytical capabilities of the human brain. Today, it is the buzzword across the spectrum of all businesses, from manufacturing to retail. This has been fueled by advancements in computing power and availability of huge volumes of digital data. The healthcare industry is no exception. While it has silently become an integral part of patient data archiving, analysis and interpretation, it is now making inroads into direct patient care. It is therefore not surprising that AI in healthcare is expected to grow at CAGR of 52 percent by the year 2024.

The Methodology
Availability of large volumes of digital data is the single most important spur to development of AI. Sophisticated algorithms are used to train the machines in extracting and analyzing information from structured data - a technique broadly referred to as Machine Learning (ML). On the other hand, Natural Language Processing (NLP) techniques are used to process unstructured data like clinical notes/medical literature. Neural Networks (NN) are a set of algorithms, interconnected and structured loosely like neurons and human brain, which obtain insights by pattern recognition. Lastly, Deep Learning (DL) is a broader term encompassing ML and NN techniques, together facilitating artificial intelligence. AI can be constantly improved by validating it over more data and user feedback.

Healthcare data is generated from both non-clinical and clinical activities, such as patient demographics, medical notes, electronic recordings from medical devices, physical examinations, laboratory data and images. There is a lot of interest in using AI in the diagnosis stage, particularly in Radiology and Imaging.
Applications of AI in Healthcare

• Healthcare Organization & Delivery: AI can help in resource allocation in healthcare. It can drive cost efficiency by matching individuals with a care provider who meets their needs, within their allocated care budget. The patient is a consumer and seeks great service. AI in combination with technology of smartphone apps can be used to improve patient experience.

• Public Health Screening: This is an area where AI has the potential of huge impact. Given geographies with large population and limited resources, AI helps by cutting-down on costs. The biggest impact is envisaged in detection of common diseases like tuberculosis as well as mass screening for cancer.

AI can drive cost efficiency by matching individuals with a care provider who meets their needs, within their allocated care budget

• Diagnostics: Machine learning works best in extracting clinically useful information from large volumes of imaging and laboratory data. It can pick-up salient diagnostic information, thus driving accuracy and productivity. AI algorithms for nodule detection on Chest X-rays and for detection of brain hemorrhage on CT scan are already available.

• Clinical Care: AI can help in immediate detection of critical events in a patient, thus enabling early therapeutic intervention. Other clinical uses under evaluation include detection of retinopathy in retinal scans, automated drug delivery sub-sequent to biochemical and metabolic alterations, detection of heart attacks/cardiac rhythm disorders, analyze speech patterns to predict psychotic episodes and identify & monitor symptoms of neurological conditions such as Parkinson's disease. In addition, robotic tools are being developed and are in use to assist surgery.

• Medical Research: AI can be used to analyze and identify patterns in a large and complex dataset, faster and more precisely. It can also be used to search the scientific literature for relevant studies, and to combine different kinds of data.

Limitations & Challenges of AI

• Inconsistencies with Digital Data: AI is heavily reliant on digital data. There are practical challenges such as incomplete digitalization of medical records, as well as the lack of standardization in IT systems, digital record keeping, and data labelling.

• Is it Really HUMAN-TECH: There are human attributes like compassion and empathy, so very important in clinical medicine, which AI systems lack. Clinical practice often involves complex judgments and abilities that AI currently is unable to replicate, such as combining knowledge, experience and psychosocial cues available from patient interaction. The doctor uses sensitivity and sensibilities in communication with a patient, not really amenable for machine learning. Therefore, AI threatens to disrupt the trust based relationship between the patient and doctor, as well as the ritual of a interaction with the doctor, which inspires confidence in a patient.

• Reliability & Patient Safety: Errors made by AI tools could be devastating, and care needs to be taken that all such algorithms are checked and certified with respect to patient safety.

Artificial Intelligence is here to stay and will definitely play a huge role in healthcare. It has limitations which may not allow it to independently deal with a patient in entirety. The human mind with its intelligence, learning and experience would probably always be the best bet to deal with intricate complexities of the human body and the myriad of diseases affecting it. However, AI tools will play a role in increasing accuracy, efficiency and productivity, across all aspects of healthcare.