A major strength of AI is its ability to see small variations in large amounts of digital information, called pattern recognition, which is a skill honed by training on massive amounts of data with the help of complicated algorithms. Therefore, it is not unexpected that a major application of AI in healthcare is in the fields of radiology and pathology. Currently, as in the past, human experts methodically evaluate and look for small changes that might signal disease on x-rays and scans and in body tissues obtained at biopsy. Most diagnostic decisions in medicine are based on such imaging and biopsy results.  AI technology has already been developed and tested that can equal or exceed the accuracy of human reviewers. These include screening for lung cancer on CT scans, screening for breast cancer on mammograms, and differentiating between benign and cancerous tissue under the microscope. Analytics are being developed that will tell us if a cancer is going to progress rapidly or slowly and what treatment might be most effective.

In the past several years we have seen many AI-based technologies move from academia to general application. Robotic surgery has allowed for precise intervention in critical body regions and virtual colonoscopy has detected colon polyps and cancers without the bothersome bowel prep and insertion of a colonoscope. Tools are now available to identify a variety of skin and eye disorders, such as diabetic retinopathy, and the detection of heart beat irregularities (arrhythmias).

Rapid diagnosis of infectious diseases is being enhanced by using AI technology similar to facial recognition to identify pathogens such as COVID, influenza and Ebola from a nose or throat swab. AI has also allowed for the development of antibodies that recognize and block the proteins that viruses use to infect cells and to produce more potent and stable mRNA vaccines for diseases such as COVID.

AI is reshaping the drug-discovery process by boosting our understanding of drug targets and synthesizing new drugs. There is potential for developing new and more effective drugs by matching molecular structural features for targeting therapy to an individual’s unique genetic makeup and to specific receptors found in cancers. AI analytics show promise to assist in optimizing immunotherapy, such as CAR-T cell (Chimeric Antigen Receptor T-Cell) therapy, in the treatment of certain cancers by stimulating the body’s own immune system. AI also has shown promise to predict response to cancer treatments and to identify risks for certain treatment toxicities

Electronic health records (EHRs) have played a major role in the digitalization of medical care documentation. However, this trend has brought with it information overload, stress of endless documentation and burnout for healthcare professionals. By creating more intuitive human-computer interfaces and by automating major tasks, such as clinical documentation, order entry and addressing email requests, we can relieve much of this stress. Voice recognition can improve the clinical documentation process and bring virtual medical assistants and virtual patient navigators into the clinic to assist with more routine tasks.

AI can also help support difficult decisions, whether to continue care for critically ill or comatose patients by analyzing data from many similar cases. Doctors have started to use chatbots, such as GPT-4, to write kinder responses to patients’ emails, provide compassionate replies for difficult medical questions, and have seen better success in appealing to insurance companies for expensive drug coverage. Although these many current and future promising benefits from AI will revolutionize healthcare, there is consensus that for AI to achieve a successful role in healthcare, it will require a blend of human experience and virtual augmentation, both working together to improve the delivery of compassionate care.