English
Doctors + AI = Superheroes
Artificial intelligence (AI) has developed rapidly. Today, AI software beats human experts in many areas, from complex Go games to the optimization of advertising strategies to driving a car.
AI has also found its way into medicine. Computer systems diagnose heart problems or lung cancer better than experienced doctors. And doctors who are supported by AI can significantly improve their diagnoses in order to detect diseases earlier and more accurately. (see http://www.bbc.com/news/health-42357257)
AI tools will give physicians capabilities that seem like superpowers compared to today’s medical standards.
More data for better AI
Data are the decisive factor for the progress of AI systems. Although algorithms have been slightly improved and computing power has become cheaper and cheaper, the main reason why AI has been improving is the availability of big data. Why is this so?
Machine Learning (ML) is the most successful branch of AI. In ML, neural networks are digitally simulated and trained with huge amounts of data. Training data could be e.g. diagnoses, therapies and their results.
Just like (human) animals, ML algorithms “learn” by recognizing the patterns in the input data that lead to the desired or undesired results. For example: patient with fever → Aspirin treatment → patient no fever. However, with their purely mathematical approach, they are able to recognize patterns that the human mind cannot recognize. And they can digest exponentially larger amounts of input than humans – the more, the better.
Once trained, the AI can be easily copied and is then available wherever a computer is available, even on smartphones.
Training data difficult to access
A lot of data that could be used to improve medical CI’s are not accessible. On the one hand, it is technically difficult to merge and share data from thousands of incompatible systems. On the other hand, data protection regulations in most countries prevent hospitals, doctors and other medical stakeholders from easily sharing patient-related information.
Data generated from new sources, such as the Apple Watch, is locked away at a few big American companies. Their goal is to generate profits for their shareholders, not primarily to improve people’s lives.
In order to equip doctors with AI superpowers, we need access to the data that is currently stored in incompatible data silos. In addition, patients should regain control over their own information.
For example, patient data could be stored in a blockchain. This would allow each individual to control who can use the information in their patient file for what purpose.
With the blockchain, we could build a medical database that spans the globe and at the same time gives individual patients back control over their data.
A more detailed version of this article can be found at https://blockruption.com/2018/03/artificial-intelligence-big-data-and-blockchain-turn-doctors-into-superheroes/?lang=en.