Artificial intelligence (AI) refers to computers programmed in a way to mimic the human brain. It finds applications in speech and image recognition, robotics, smart homes and our very own smart phones to name a few. Machine Learning (ML) which is another AI related terminology is the study of computer algorithms to improve automatically through experience much like an archer trying to hit the “bulls eye” with repeated attempts and improving at every attempt until the target is achieved. Like in other fields, AI and ML have slowly started invading the medical world. A good example is Watson which is rather increasingly used by oncologists to manage almost everything related to cancer.
Today a cardiologist is flooded with biometric data from live streaming mobile devices, genetic profile of cardiac diseases, clinical data form Electronic Health Records (EHR) and ongoing trials and lab research results. Piecing the data together and prioritising the variables for better patient outcomes is a Herculean task. AI and ML can better solve this puzzle and augment the cardiologist’s effort to provide a more meaningful and personalised care.
ML strategies are broadly classified as supervised and unsupervised learning involving algorithms like regularised regression, tree based methods, support vector machines, neural networks and deep learning. Earliest commercial application of deep learning in medicine was for analysis of images (e.g. to detect diabetic retinopathy from a database of 1.28 lakh images). Similar success was also achieved in detecting skin cancer. Deep learning in cardiology has shown potential in hearts pumping function assessment and image assistance in percutaneous coronary angioplasty which is programmed into the machinery. Their use in such niche areas is so common today that we fail to appreciate them. Another algorithm called “reinforced learning” which is more nascent help tailor the treatment based on patient characteristics. It can be incorporated into the electronic health records of patients and help them get a more personalised care. In the near future, AI and ML in cardiovascular medicine will positively influence the following in a big way:
Research and development: Novel therapeutic agent discovery and precision disease stratification.
Clinical practise: Aiding in accurate diagnosis and assisted therapy selection by integrating multi-omic data.
Public health: By way of optimising manpower and economic resource allocation for uplifting the general health of masses. AI can help in continuous remote monitoring and corrective action which will reduce the time taken to get medical help when needed and may be even avert a medical emergency by early predictive intervention.
Currently the developed world employs trained personnel to incorporate AI methodology in medical practise but soon they will be increasingly easy and commoditised much like installing the windows operating system on your desktop. This will lead to automation of interaction with medical records helping the doctor to make more meaningful treatment decisions. AI and ML should not be considered as a substitute for doctors but rather a powerful tool for the health delivery system to provide better medical care. Today we live in an age of information with mounting pressure on doctors and health infrastructure for efficient, effective, quicker and personalised care for patients and AI has come by as a boon.