If you search online for artificial intelligence, or AI, you’re likely to encounter a whole host of speculative questions such as these, and at the moment, no one really knows the answer to them. Thanks to TV series and films such as ‘Westworld’ and ‘Ex Machina’, the general public’s perception of AI may be slightly skewed. When we talk about AI in healthcare, we’re not generally talking about the level of technology that can create humanoid robots capable of overthrowing human society. Well, at least not yet.
Despite futuristic imaginings and the associated worries, the AI market is projected to reach $70 billion by 2020, and is predicted to have a transformative effect on a number of industries across the globe.1 But specifically in the healthcare field, there is still some level of confusion surrounding what AI really is, and why it matters to the industry.2
While no one seems to agree on a single definition for the term, the general consensus is that AI is any task a computer can perform equally as well as, if not better than, humans.2 But where there’s AI, there can be machine learning (where computers act upon information, without being explicitly programmed) and even deep learning (where software learns to recognize patterns in distinct layers). Currently in healthcare, most of the new AI-generated solutions analyse data and use them to recommend treatments based on algorithms created by humans, rather than through machine learning.
But deep learning is beginning to find its way into healthcare.2
With its Face2Gene platform, Boston start-up FDNA is attempting to create a process to aid healthcare professionals (HCPs) in determining what disease a patient has, simply by analysing their face, with no need for additional scans or testing. Utilizing a photo database of people with over 2,000 rare genetic diseases, HCPs can take a quick picture of their patients, upload the photo to the FDNA app, and in turn, receive a list of potential diseases that the patient might have based on the image. The list is generated through a combination of AI and deep learning, where the individual’s facial features are analysed, to see whether there are any indicative signs of rare disease. The start-up hopes to dramatically speed up the diagnostic process for those with rare diseases—who are typically seen by seven doctors before a correct diagnosis is made.3
Another area in which AI is showing promising potential is orthopaedic surgery. AI-assisted robotics can analyse data from pre-op medical records, and physically guide the surgeon’s instruments during an operation; using data from previous procedures to inform new surgical techniques.4 So is it effective? An AI-assisted robotic technique created by Mazor Robotics was studied in 379 people with degenerative lumbar spine disease, across nine different sites. Remarkably, it was found that there was a five-fold reduction in surgical complications when the robotic-guidance was used, in comparison to procedures carried out without it.4 According to the Harvard Business Review, when applied to the orthopaedic surgery field, this type of technique could generate a 21% reduction in people’s length of stay in hospital following surgery, due to fewer complications and mistakes, and subsequently create $40 billion in US savings, per year.4
AI-powered virtual health assistants are also emerging as a potential aid for patients. San Francisco start-up Sensilla’s has developed a platform, which they describe as “a cross between WhatsApp and Siri, that captures all the important signals about a person’s health”. Patients can talk to the platform’s nurse avatar, ‘Molly’, through their smartphone‘s microphone, and share how they’re doing once a day, or every few days. This information is then compiled into a medical record—along with data from other devices and wearables the patient may use—that only authorized healthcare professionals can access. The AI powering the platform enables ‘Molly’ to respond not only the patients’ symptoms, but also their mood, meaning she can speak to the patients with empathy, and detect whether they may need mental health counselling.5 Again, the Harvard Business Review estimated the potential savings this type of technology could have, and they found that AI-powered nurse assistants could save $20 billion per year in the US, by saving nurses significant amounts of time.4
Some would argue that AI has almost endless possibilities within the healthcare industry. Utilizing its full potential could mean improved diagnosis, faster drug discovery, early detection, and that healthcare professionals have more time to focus on providing better care.4,6 As it becomes increasingly sophisticated, many may fear that AI is advancing faster than our own (and they’re not necessarily wrong)—but most would argue that the opportunities of this progress far outweigh the dangers.2
This Let’s Talk Innovation article was created by respiratory_care v2.0.