AI and Medicine
The vast data capture and the technology now available have enabled the interest in AI to have practical applications for the first time in numerous fields. One of the most promising of these fields is that of medicine, despite being perhaps one of the more contentious.
A brief history of AI in medicine
The use of AI within medicine has had a long history in some shape or form, dating back to the early days of computer science. One of the earliest examples of AI in this field was a program developed in the 1950s that used machine learning algorithms to diagnose infectious diseases. Although these pioneering systems were bold they proved ineffective for multiple reasons. These far simpler computers with low processing power could not process large amounts of medical data in a manner that was timely or at a reasonable cost. In the decades that followed, AI continued to evolve and became increasingly integrated into everyday medical practice. For example, computer-aided diagnosis systems were developed in the 1960s and 1970s, such as MYCIN (Shortcliffe, 1974) which helped to make more accurate diagnoses of bacterial infections and to suggest the use of the most appropriate antibiotics. CASNET (Weiss, 1978) was used to help in the diagnosis of Glaucoma – both of the systems
(Figure 2-Timeline of important AI events with a focus on Medical application. Adapted from Kaul, Et al 2020)
used rule-based protocols and knowledge-based systems to support physicians in their work. However, the rigid criteria under which these systems functioned limited the accuracy of their diagnostic capacity, requiring a large amount of oversight before decisions could be validated. Given these limitations, perhaps it is of little surprise that they did not gain much of a foothold at this particular time. In the 1990s and 2000s, advances in machine learning algorithms and the increasing availability of large amounts of accessible medical data paved the way for the development of more sophisticated AI systems. For example, researchers created algorithms that could analyse medical images to detect cancer and other diseases through the system ‘ Watson ’ (Ferrucci, 2012). Also, natural language processing systems were developed to help physicians extract relevant information from electronic medical records. These systems were also steadily getting more powerful and more affordable. Above, in Fig. 2, there is a modified timeline showing various important milestones in the development of Medical AI.
33
Made with FlippingBook - PDF hosting