How AI could predict cancer years in advance.

For a long time, AI has been foretold to be the destroyer of humanity. We have all seen canon Sci-Fi movies like Space Odyssey, Terminator, or I Am Robot. They have a common plot point: AI designed to allow us to achieve new heights goes mad and ends up doing the complete opposite.

Even though this genre has the word “Science” in it, there is more fiction than actual science. Real-world AI is far from being so technologically advanced, most of the time it can’t make conclusions that would seem obvious to us humans. We only need a couple of examples to understand the concept of a cat, while AI would require thousands of high-quality photos to reach ~60% prediction accuracy.

That is because most AI models, in short, and with a lot of simplifications, are designed to level things out. But some things are easier for AI’s to understand than others. Finding deviations from the norm in standardized images is not so difficult. Deviations in statistics or from graphs, different schemes, or almost identical images are very easy to analyze. On such data objects, any deviations from the median are shouting to be noticed.

If we asked you to divide the dots into 3 groups depending on how far they are from the median, the image on the right represents how you would most likely do it. So would AI but on a much bigger scale.

Medical data and research results would be a perfect fit to implement an AI. Even though every single human being is unique, our bodies kind of work in the same way. Most of us are in the green zone, but some cases leave a whole trail of deviations in patients’ medical history. Blood analysis here, an almost unnoticeable markings on x-ray there, one after another, these deviations stack and slowly put a patient in the red zone. Then doctors notice these changes and start troubleshooting. Unfortunately, because the detection of dozens of errors in time is unbelievably complex for humans, in some cases it might be too late for help.

But AI does that job just fine. The latest research made by the French has proved AI to be capable of spotting signs of lung cancer on CT scans a year before they can be diagnosed with existing methods. The researchers trained AI on CT scans from 888 patients, who were earlier examined by radiologists for suspicious growths. Then, the program was tested on a different set of 1179 patients, 177 of which were diagnosed with lung cancer. AI was able to identify 172 of the 177, or 97% tumours, present in those CT scans. The remaining 5 deviations the program didn’t notice were located near the chest area, where it’s harder to distinguish malignant tumours from healthy parts of the body.

To further test the capabilities of the AI, researchers repeated the experiment, but this time they used 1-year-old CT scans for the same set of patients. It was able to identify 152 suspicious areas that were later confirmed as lung cancer.

Researchers also mentioned a problem they came across. The program identifies too many suspicious areas that are not cancer (false positives) and this would need to be vastly improved before the program could be used in the clinic because investigating all these would result in unnecessary biopsies.

The number of datasets they had was just not enough for AI to perfect its predictions and many more CT scans will have to be made, reviewed by radiologists and “fed” to AI before its accuracy will be improved. We are talking thousands if not tens of thousands of datasets. That might take another year or two, but every contribution counts. Further improvement of AI technology in healthcare is slowed by a lack of sufficient data sets to work with.

Every contribution matters. You can help technology move forward or even save someone’s life. Your anonymized medical data is digital gold, more important than bitcoin or oil. Your test results before and after taking a medication will help people who do not have access to quality health care to get qualified help. Your CT scan will help find cancer, and your blood test can save a diabetic.

Record, analyze, compare, predict, prevent. Doctors worldwide will be able to do their jobs better with new data and you will literally get an ability to see your future by comparing your data with the data from people with similar records.

With DeHealth you can tokenize and monetize your personal data, as corporations will no longer be able to use it without your express permission. Use software and services aimed at diagnostics, health monitoring and treatment. Be prepared for any situation and always have your medical records in your pocket. Get Free access to medical goods and services provided by businesses and organizations around the world. Last but not least, help others predict and prevent unwanted conditions.

650+ medical institutions, 30,000 doctors and 2,000,000 patients have already registered and provided the world with more than 1,600,000 declarations and 2,500,000 patient card entries during 1,000,000 appointments.

Become a part of better healthcare at www.dehealth.world