Unlocking the potential of AI in diagnosis and treatment

Artificial intelligence (AI) is the ability of a digital computer or a robot to perform tasks commonly associated with human beings, such as the ability to reason, generalise or learn from past experience. By processing vast amounts of data, AI algorithms can identify patterns and predict medical outcomes with unprecedented accuracy and speed. It has the potential to improve the practice of physicians and enhance patient outcomes for the world’s overburdened healthcare systems.

The World Economic Forum’s Digital Healthcare Transformation Initiative has recently identified AI as one of the possible means to tackle three of the most pressing challenges in healthcare — the increasing burden of chronic diseases; inequitable patient outcomes and healthcare access worldwide; and resource constraints.

The Indian government’s National Digital Health Mission will align closely with this initiative once it takes shape.

The application of AI into healthcare stems from industrial AI, the origins of which date back to the 1950s. John McCarthy, at the Dartmouth Conference in 1956, coined the term “Artificial Intelligence”. AI with demonstrable medical applications took off in the 1970s with the creation of INTERNIST-1, the first artificial medical consultant. It utilised a search algorithm to arrive at clinical diagnoses based on patients’ symptoms. The modern era of AI began in the early 2000s with the creation of voice recognition feature, followed by the IBM launching a question-answering system called Watson and then the GPT models, in 2020. The improvements in computing power, faster data collection and data processing, the growth of genomic sequencing databases and widespread implementation of electronic health record (EHR) systems have contributed to the the growth of AI in healthcare.

Currently, AI includes techniques such as machine-learning, deep-learning, natural language processes and computer vision and algorithms like large language models (LLM) which can analyse large data to understand, summarise, generate and predict new content. AI is widely tested and used in medical imaging such as X-Ray and photography and has shown great potential in diagnosing conditions like diabetic retinopathy, skin cancer and lung tuberculosis. Machine-learning models have been shown to predict and diagnose Alzheimer’s disease. AI has been explored for use in cancer diagnosis, risk stratification, molecular characterisation and drug discovery.

Using a commercially available endoscope with an integrated AI system, the device can point out a suspicious area to target biopsy and even suggest the probability of it being a cancer. AI can be used for personalised or precision medicine, using time-efficient analysis to accurately predict which treatment protocols will be best suited for each patient based on their individual genetic, molecular and tumour-based characteristics. The use of telemedicine to analyse an ECG or interpret imaging has been used effectively to guide treatment at remote locations and to triage patients for interventions.

AI algorithms are proving instrumental in diagnosing conditions like brain stroke, intracranial haemorrhage and pulmonary embolism. Wearables, smartphones and internet-based technologies can be used to monitor patients’ cardiac data points, potentially enabling actionable detection of cardiac events.

AI has an emerging role in teaching and training with the use of virtual assistant and virtual reality. AI has been used in designing clinical trials, drug discoveries and predicting drug-drug interactions and adverse effects. Some upcoming applications of AI include a health LLM by Google to give users personalised health coaching messages based on their patterns in sleep schedule, exercise intensity and changes in heart rate variability, etc. The other developments include IBM’s Watson Oncology, Google’s DeepMind platform and Microsoft’s Hanover project. A company called Medsol AI Solutions in South Africa has come up with a cutting-edge, wi-fi-enabled ultrasound probe designed for rapid breast cancer detection

However, a major concern of using generative AI in healthcare is the absence of ‘human touch’ and empathy. The other big problem is the unavailability of credible data. Its effective use needs massive amounts of data, which, however, can be flawed because of bias against different ethnic groups and minorities which are not adequately represented. Currently, most AI algorithms are based on data of white male subjects. Then, there is a concern about patient privacy and instances of monetisation of data. Algorithm dysfunction or misdirection and lack of transparency or interpretability, the ‘black box’ phenomenon, are other problems. Plus, there are ethical considerations for collecting genomic information and concerns about integrating AI with IT systems, gaining physician’s trust and ensuring compliance with government regulations.

There is a huge potential of AI in India. With gross inequalities in nutrition, education, infrastructure, trained manpower and healthcare facilities, it can be a boon. It can be used to predict epidemics based on disease trends and epidemiological data. For example, spurts of dengue fever, malaria and encephalitis can be predicted based on rainfall, humidity and past trends and by analysing web traffic and modelling mosquito movement patterns. More importantly, at the community level, predictive analytics can project the development of non-communicable diseases like diabetes, fatty liver and heart ailments. Screening for tuberculosis in vulnerable populations has been already carried out.

However, India is slow in catching up with the world. Though leading hospitals like the PGIMER, Chandigarh, AIIMS, New Delhi, the Asian Institute of Gastroenterology, Hyderabad, and Tata Memorial Hospital, Mumbai, have initiated India-centric research, much more needs to be done. While the government has initiated digital record-keeping through the National Digital Health Mission, currently, only a handful of private hospitals are using EHR, which is the primary source of all data. Initiating and maintaining digital records in the whole country is a mammoth task, requiring a huge investment. In India, primary care has to be the focus for the use of AI in the prevention, screening and surveillance of diseases, besides providing remote medical access.

The journey of AI from a novel concept to a fundamental aspect of healthcare signifies a technological revolution. India should not miss this transformative advancement.