According to Google CEO Sundar Pichai, artificial intelligence is as important as electricity or fire and has recently renamed the Google Research team into Google AI. The AI ?? also needs a lot of attention in its development, and even the AI ?? can have extremely different uses. One of these is in the medical field.
Google has a team for that too, it is called Medical Brain and has just developed an algorithm that can establish risks related to health conditions and the probability of death for each individual user.
This is one of the biggest challenges for the health sector today, and the statistics in these cases can help a lot by exploiting mountains of data, analyzing them and making them presentable. These operations, however, require considerable time and costs, but the system of the Google Medical Brain team can do it in a reasonable time, analyzing data not only of PDF and digital files, but also those written on old tables written in hand: the algorithm can then analyze data and make predictions quickly and, according to the document, even more, reliable.
In one of the examples cited in which the algorithm proved rather reliable a woman with breast cancer and fluid in the lungs has heard the opinion of two doctors undergoing a radiological scan. Hospital computers with traditional systems measured vital signs and previous health reports, resulting in a 9.3% chance of death. The Google algorithm has instead examined the woman, but using more than 175 thousand data has estimated a probability of death of 19.9%. And that’s what happened, unfortunately, a few days later.
It is not pleasant to hear from a computer that you have one in five chance of dying, but by becoming more accurate, artificial intelligence algorithms could save millions of lives. In fact, it is not just trivial to know when and if you will die once you get to the hospital, but a method to draw a more accurate picture of the patient based on any information about it, and others who have faced similar situations. This can allow doctors to establish different care, and potentially save the patient’s life.
Google claims to reach up to 95% certainty in determining if the patient will die within 24 hours of being admitted to the hospital. To do this, he used the anonymized data of 216,000 cases of 114,000 patients, which were indispensable for training the algorithm. The approach of Google not only allows you to have a higher accuracy than the predictive methods used today but avoids the “manual work” of the experts to make the data presentable. In fact, the software takes care of everything, saving about 80% of the time to make the forecasts.
The neural network behind Big G technology analyzes PDF documents or scans of paper documents and makes predictions about patient conditions immediately and in the near future. The accuracy of the results of the system is so high that Google has already planned experiments in different hospitals in order to verify with more samples its reliability and possibly determine if the experiment can become a commercial product.