THANK YOU FOR SUBSCRIBING
As the technology evolves, newer testing, screening and treatment applications are being explored that could have a material impact on improved survival rates. The real value of AI investment in the fight against cancer tends to fall into two categories - diagnosis and treatment. Let’s review each in a little more detail.
Early diagnosis to increase survival rates
Whether or not AI algorithms are smarter than human doctors is debatable, but they’re certainly faster when it comes to pattern recognition and computation. In 2018, Google’s DeepMind trained a neural network to detect more than 40 different types of eye disease just through analyzing 3D retinal scans. The same kind of technology can be applied to scanning images for signs of cancerous anomalies. Neural networking means that the more data an algorithm has access to, the more accurate its calculations become - which, in this case, would be pattern recognition.
Take melanoma for instance. An oncologist would tell you that this is one of the most stubborn forms of cancer and notorious for being difficult to spot. Instead of a team of tired, overworked doctors pouring over images of MRI scans looking for something that’s incredibly easy to miss, an AI algorithm could take over and be far more accurate. In fact, an AI algorithm could scan biopsy images or MRI scans more than a thousand times faster than even the best doctor, with an accuracy rate of nearly 90%. And increased accuracy means better health outcomes.
A recent Google Health and Imperial College London joint research study also revealed that AI was more accurate than doctors in diagnosing breast cancer. Additionally, an initiative between Northwestern University and Google is improving the rate of lung cancer detection and is now moving towards clinical adoption.
High precision treatment for better patient outcomes
Cancer drugs and treatments aren’t a one-size-fits-all affair. Individuals react differently to treatments based on their biological and behavioral characteristics, and not every patient’s recovery time will be the same. That makes choosing the right course of medication for a cancer patient and mapping out their recovery roadmap incredibly important. To err is human, and there’s simply too much data to consider for doctors to make the best call each and every time. Again, this where AI can have an invaluable impact.
An AI algorithm is capable of analyzing vast quantities of patient data which allows it to balance the best course of treatment with an optimal recovery period. Such data could include anything from medication history and allergies, right through to hereditary traits and their lifestyle choices. These are of course considered by doctors before recommending treatment, but with diagnoses and image scanning, there is only so much data that a human doctor can process. This has led to the rollout of AI-based prescription services in a number of healthcare establishments, giving doctors valuable assistance when it comes to determining courses of treatment.
Data first and the rest will follow
By applying AI technologies to tackle some of the most complex medical challenges, the healthcare industry is changing the way patients are treated and their subsequent health outcomes. In fact, the healthcare industry is close to seeing multiple AI-centered cancer diagnosis and treatment systems and methodologies break through to clinical readiness. That is not in question. But, there will be missteps along the way, and the time is right for any organization considering AI in cancer diagnosis and treatment to steer the ship in the right direction.
Ironically, the creation of the AI algorithm itself is typically the easier (but not easy) part of delivering useful AI applications. The more challenging aspect is accessing, aggregating and organizing the data, so that the AI can learn and deliver results. AI is nothing without data. Without it, neural networks can’t learn, and algorithms have no rulebook from which to identify patterns or anomalies. The success of AI in the diagnosis and treatment of cancer lies in high-quality data engineering.
It’s not enough to simply collect data; it must be standardized, organized and labeled before it can be useful. A case in point is that of IBM Watson Health, the high-profile health AI and analytics division of IBM, that is now reportedly being sold due to difficulties stemming from data collection and data sets. Any healthcare organization seriously looking to improve health outcomes with AI-based cancer diagnosis and treatment should heed this cautionary example. The harnessing of AI-powered systems may hold the key to future health breakthroughs, but the battle will be won by those who take to heart the mantra of “data first, AI second.”