AI can predict your future health – as is the weather

Artificial intelligence can predict human health problems over the decade in the future, scientists say.

This technology has learned to identify models in people’s medical documents to calculate the risk of more than 1000 diseases.

Researchers say it is like a weather forecast that provides 70% of the chances of rains – but for human health.

Their vision is to use the II model to identify patients with high risk patients to prevent illness and help hospitals understand the demand in their area, years ahead of time ahead of time.

Model-called Delphi-2M-Users similar technology with famous AI chatbots like chatgpt.

AI chatbates are trained to understand the models of language so that they can predict the word sequence in the sentence.

Delphi-2M has undergone training samples in an anonymous medical map so that it can predict what will happen next and when.

This does not predict the exact dates as a heart attack on October 1, but instead assesses the likelihood of 1231 diseases.

“Such as the weather where we could have 70% of the chances, we can do it for healthcare,” Prof. told me.

“And we can do it not only for one disease, but also all the diseases at the same time – we have never been able to do it before. I am excited,” he said.

Initially, the AI ​​model was developed using anonymous data in the UK – including a hospital reception, a general practice doctor and a lifestyle, such as smoking – assembled with more than 400,000 people within the framework Research project BIOBANK UK BIOBANK.

The model was then checked to find out if its projections were made using other BIOBANK data, and then with 1.9 million medical documentation in Denmark.

“That’s good, Denmark is really good,” says Professor Biri.

“When our model says it is a risk of one 10 next year, it really seems like it turns out to be one of 10”.

The model is best predicting diseases such as type 2 diabetes, heart attacks and sepsis, which have accurate progression of the disease rather than more casual events such as infections.

People are already offered statin that lowers cholesterol, based on the risk of heart attack or stroke.

The AI ​​tool is not ready for clinical use, but the plan is to use it in a similar way to identify patients with high risk as long as it is possible to intervene early and prevent illness.

This may include medicines or specific lifestyle tips – for example, people who may develop some liver disorders, which are more profitable from reducing alcohol consumption than the total population.

Artificial intelligence can also help report disease check programs and analyze all health care records in the area to anticipate demand – for example, how many heart attacks a year will be in Norwich in 2030 to help plan resources.

“This is the beginning of a new way to understand human health and the progression of diseases,” said Professor Moritz Gerstang, head of the II oncology department at DKFZ, the German Cancer Research Center.

He added: “Generative models, such as ours, can once help personalize care and anticipate health care needs.”

Model AI, described in a scientific journal Nature.

There are also potential prejudices because it was built from BIOBANK data in the UK, which is mostly consisting of people between the ages of 40 and 70, not the entire population.

The model is now upgraded to account for additional medical data, such as visualization, genetics and blood tests.

But Professor Biri says: “Just emphasize, this study-all you need to check and go well and reflect before it is used, but the technology is here to make such types of forecasts.”

He believes that he will go the same way to the use of health care genomics, where it took a decade to move from scientists who are confident in health care, the opportunity to use it regularly.

The study was a cooperation between the European Laboratory for Molecular Biology, the German Cancer Research Center (DKFZ) and Copenhagen University.

Professor Gustavo Sudre, Neurovisual and researcher II in King London, commented: “This study looks like a significant step towards a scalable, interpreted and most importantly responsible form of prognostic modeling in medicine.”