From C-3PO to Surgical Robots: The History of Artificial Intelligence in Health
A look back at AI’s journey into the healthcare field
When you think about artificial intelligence (AI), you may conjure images of non-physical entities vying for world domination (think Terminator), robots becoming self-aware and developing a consciousness (think Westworld, Ex-Machina), or a trusted friend with all the answers (think J.A.R.V.I.S. from Iron Man). Although some of these reel-life examples may be outside the scope of current reality, AI now definitely has a stronghold in various aspects of real life as well.
Simply put, AI enables computer systems and machines to mimic human cognition to perform tasks. These tasks could include visual perception, speech recognition, decision making, and predictive analyses. AI-powered tools are being employed by all the big tech giants worldwide — Netflix knows what you like and subsequently, would like to watch, and Amazon can predict your future purchasing habits based on what you have already bought. No doubt, AI is omnipresent in many facets of daily life, and its potential is also growing in the healthcare sector.
AI has opened the doors to unprecedented opportunities in accelerating research and improving healthcare delivery. It is being used in drug discovery and development, making accurate clinical predictions, imaging analyses for diagnoses, and intelligent robots that can perform surgeries. One of the most recent AI-related scientific breakthroughs was the determination of a protein’s 3D structure by Google-owned DeepMind’s deep learning algorithm (AlphaFold) with a high accuracy. This revolutionary breakthrough solved one of biology’s most pressing mysteries, and it was all achieved through AI.
But before such AI-driven scientific discoveries, how did AI actually step into and cement its place in the healthcare sector? AI’s journey into the healthcare industry is an interesting one, which can help in demystifying its future in the industry as well.
Trying to answer the age-old question: “Can machines think?”
During the early and mid-20th century, Hollywood attempted to answer this question. Early examples include HAL 9000, a computer with a human personality in Stanley Kubrick’s 2001: A Space Odyssey; and C-3PO, a droid that was “fluent in 6 million forms of communication” in the Star Wars franchise. In real life, Alan Turing began to explore whether machines can use reason to solve problems and make decisions, like how the human mind does. His 1950 paper, Computing Machinery and Intelligence, theoretically discussed how intelligent machines could be built and how their intelligence could be tested. But the practical challenges were undeniable. Early computers could only carry out commands and not store them, which is a necessity of the learning process. Moreover, the costs of computers were exorbitant. These factors impeded further exploration into the intelligent machine.
A conference in 1956 changed everything. John McCarthy coined the term “Artificial Intelligence” at the Dartmouth Summer Research Project, which aimed to bring top researchers for an open-ended discussion on this new research field. This conference set the stage for AI for the two decades. General Motors developed the first industrial robot arm that could perform die casting in 1961. In 1966, the Stanford Research Institute invented “the first electronic person”, which was called Shakey. Shakey was able to interpret complex instructions to perform actions — a key milestone in robotics and AI. Thus, AI established its place in driving innovations in the engineering field.
AI’s (slow) entry into the healthcare field
The origins of AI in healthcare can be traced back to Dendral, the first problem-solving program or expert system developed at Stanford University in the 1960s. It was originally intended to help organic chemists identify a substance’s molecular structure using mass spectroscopy and chemistry knowledge. Interestingly, a curiosity in extraterrestrial substances and species due to space explorations at the time drove the development of the program. So although not directly related to healthcare, the Dendral program laid the foundation for more healthcare-focused AI systems.
MYCIN was an AI-based expert system derived from the Dendral program used to identify the causative agents behind bacterial infections, such as bacteremia and meningitis, in the 1970s. The physician running the program would answer a series of yes/no or textual questions. Based on these responses, MYCIN would then produce a list of possible causative bacteria, and also recommend antibiotic dosages adjusted to the patient’s weight. AI pioneer Allen Newell considered this program “the granddaddy” of all expert systems, and “the one that launched the field”. However, the program was never used in real-world clinical settings.
The CASNET (causal-associational network) model in the 1970s was one of the earliest examples of computer-assisted medical decision-making for glaucoma. It comprised three components: observations of a patient, pathophysiological states, and disease classifications. The model draws from the clinical expertise of a network of glaucoma specialists to aid in the diagnosis and treatment of complex glaucoma cases. The CASNET model served as a prototype to test the feasibility of AI-powered tools in the clinical field.
AI in healthcare during the 1960s also went beyond curative medicine. ELIZA is considered the first chatbot and demonstrated the use of AI-driven language processing. The program attempted to recreate the conversation between a psychotherapist and patient. ELIZA served as the basis for more sophisticated chatbots that we encounter today, such as the “WHO Health Alert” on WhatsApp and Duolingo’s chatbot characters for language learning.
Since then, AI has tremendously evolved with the development of machine learning, deep learning, and big data science. Now, machines can continuously learn from the data they collect and make informed decisions and predictions, all without human involvement. Deep learning goes even further — algorithms, inspired by the complex network of neurons in our brains, learn from huge amounts of data. Deep learning enables machines to solve complex problems and is the most human-like form of AI used today.
With the large volumes of healthcare data at our disposal, these sophisticated subsets of AI are being leveraged to gain clinical insights and make better healthcare decisions. AI is now also playing a greater role in preventive medicine, especially with the recent shift in the global disease burden to chronic diseases. Predicting repetitive patterns of adverse events in chronic diseases can save significant healthcare costs and aid in risk stratification. This means that AI can be used to detect at-risk patients and develop tailored disease management plans to ensure positive patient outcomes. For example, SELENA+, an AI software system developed in Singapore, uses deep learning to detect eye damage from the images of retinas of patients with diabetes. This can help in timely intervention and thus, reduce the risk of permanent blindness, a major adverse consequence of type 2 diabetes.
AI now — only onwards and upwards
What was once a figment of scriptwriters’ imaginations has now become commonplace. From Amazon’s Alexa to self-driving cars, AI-powered technologies are changing the faces of many industries, including the healthcare one. AI has tremendous potential in the healthcare delivery sector, particularly in predictive medicine, drug development, streamlining the patient experience, robot-assisted surgeries, and in easing the load on healthcare systems. AI applications can slash annual healthcare costs by USD 150 billion by 2026 in the US. AI can also herald in a new age of personalized medicine due to its ability to analyse massive amounts of genomic data. Although a hostile takeover by intelligent cyborgs won’t be happening anytime soon (thankfully), the continuous evolution of AI is set to dramatically transform several facets of the healthcare sector.
Schenelle Dlima is a Scientific Content Writer at Saathealth, an AI powered, chronic care digital health platform.
