Artificial Intelligence has found a new home and is already causing waves, barely a decade into the latest technology. AI in healthcare companies was predicted to be one of the biggest revolutions the modern world has seen, and from the look of things, they may just have been right.
Healthcare organizations, healthcare providers, researchers and technology specialists are teaming up to ensure the best clinical support, scientific decision-making, high-level care delivery, timely triage and diagnosis, promotion of self-care and prevention wellness and promoting chronic care management all backed by artificial intelligence (AI).
A team of scientists from China tested an Artificial Intelligence system developed in five years, and the results were nothing short of astounding. The system was designed over five years at the Chinese Academy of Science, and upon testing, it seemed the doctors’ diagnosis was at odds with what AI predicted.
Diagnosing the patients with AI - how is it done?
Diagnosis of coma patients is made by an international system referred to as the Glasgow Coma Scale. Patients are given scores ranging from NT (not testable) to 6 in different tests, including motor response, verbal response and eye-opening.
Lower scores imply the patients have little chance of ever waking up again, and below a certain point, the families have the blessing of the law to unplug them from the machines.
Some of China’s best neurologists first diagnosed the AI coma patients, then diagnosed them by the system. The team gave the patients very low scores – they would likely never wake up again. The AI algorithm had other ideas. It indicated that the patients would wake up within a year of the scan. In all seven instances, it turns out the AI was right.
How accurate is the prediction of AI?
One of the most common questions people have online is the number of false negatives that AI in healthcare companies has. It’s all well and good that it could help seven fortunate people, but it wouldn’t be so useful if it stated all a hundred people in a pool would recover, and only three did.
The first patient was a 19-year-old that had spent six months almost entirely unresponsive. He was given a score of seven out of 23 and virtually had no hope. The system gave him more than 20 points. The second was a 41-year-old female who had been vegetative for three months. The team gave her a recovery score of 6; the computer said 20.23. Out of a pool of three hundred, the system achieved an accuracy of 90%.
However, just like human beings, AI is also capable of making mistakes. The most publicized of these has been the case of a 36-year-old man who suffered brain damage as a result of a stroke. He was given low scores by both the AI and doctors but was able to recover fully in less than a year. But, just like humans learn from mistakes, artificial general intelligence can also be developed further by learning from previous mistakes to tackle new challenges.
The uphill task artificial intelligence healthcare companies have is to weigh in variables that are not directly measurable by their systems. Things like their insurance policy, whether the family can afford continued treatment and the immediate risk of death all pose a genuine challenge.
Regarding the latter, patients often contract diseases like pneumonia, and by the time it’s diagnosed, it’s too late. AI in healthcare isn’t some magic bullet, but it does have incredible implications for the future.
Benefits of AI in healthcare sector
With over 500,000 patients suffering from different forms of brain damage in China alone, artificial intelligence healthcare applications are almost unbounded.
Some Chinese hospitals have already integrated the system directly into their diagnosis procedures. It weighs in as much as 50% in the decision-making of whether the patient should be kept on life support or not. While it’s not anticipated to take over doctors’ jobs any time soon, it plays a significant role in helping doctors make their decisions.
The system is so advanced that it can just be reprogrammed to diagnose different diseases in patients. Since it can perceive smaller details not readily apparent to the human eye, such as small changes in blood flow, it has applications ranging from diagnosing diabetes to cancer.
Compared to AI, conventional systems aren’t too reliable
As mentioned before, current scoring systems such as the GCS rely on simple, testable signs to be able to diagnose the severity of traumatic brain injury. It’s an incredibly useful system that can be relied on almost immediately. For instance, in the event of a car crash, it would be pretty challenging to wheel in a giant machine to diagnose a patient. That said, though, it’s not a system without its flaws.
More ‘invisible’ details must be considered that human beings normally cannot perceive. The patient might be fully conscious, but his ability to communicate with the outside world is impaired. Conventional methods can’t possibly wrap their way around this.
The scope of AI in healthcare companies
With a high success rate, there is a lot more hope for artificial intelligence coma patients than we could ever have possibly conceived. AI in healthcare is moving forward to become ubiquitous with a wider scope that ultimately leads to faster recoveries for patients. AI startups in healthcare are continuously innovating to enable the implementation of AI on a larger scale.
