Algorithms with a Spanish seal to combat rare diseases

0
100
Algorithms with a Spanish seal to combat rare diseases
Algorithms with a Spanish seal to combat rare diseases

Algorithms with a Spanish seal to combat rare diseases

Algorithms with a Spanish seal to combat rare diseases

Understandable Concepts, a Microsoft partner firm, collaborates with Foundation29 to apply new technologies in ailments that are difficult to treat and cure

A worker of the giant Microsoft with a son who has a rare disease. The creation of a foundation (Foundation29) specialized in technology and health. A partner company of the technology giant Microsoft, Plain Concepts, specialized in artificial intelligence (AI) and one of whose partners is a friend of the Microsoft above employee. All this together will give us the answer to why AI algorithms have not been used until now to investigate rare diseases.

It is estimated that those with one of these rare conditions (they do not usually affect more than one in 2,000 people) take, on average, about five years to be correctly diagnosed. Something that leads them to see an average of eight specialists on their journey. And since they are less known disorders, one in four cases ends up being misdiagnosed. According to Foundation29 data, all this makes 40% of rare disease patients see how their condition worsens from all those causes.

Therefore, the objective of applying artificial intelligence in this field is to try to reduce all these numbers and have better knowledge, success and diagnosis of these types of ailments.

Understandable Concepts’ involvement with the project is charitable. To function, a genetic study of the patient is needed and, in parallel, the doctor in his routines will ask the patient about the symptoms. “Our algorithm takes those signals that the doctor has noted, they are entered into the tool and, through Natural Language Processing (PLN) techniques to clinical data, it extracts the phenotypes (list of symptoms)”, explains Pablo Peláez, CEO of the firm above. Also, it analyzes the genomic information of the person.

SEE ALSO  WhatsApp shows how exchanging messages with other apps will work and no, it is not as we expected

AI thrives on data to learn and process this information. Therefore, the more cases the tool has, the better it will behave. For this reason, Foundation29 has its doors open to the collaboration of doctors and patients themselves. The more of them contribute their data, the more they will learn the algorithm and will be able to give more valuable information to doctors. Any of them may have access to said system, upon request for authorization.

According to the CEO of PlainConcepts, this tool does not diagnose the patient’s disease. However, it does allow the doctor to know in what percentage the symptoms and genetic data are identical to other pathologies so that the doctor can rule out some diseases and approach others. “It will help to focus diagnostics much more in a field where it is tough to do so,” he clarifies.

Similarly, the more genomic studies of people with rare diseases have been uploaded to this new platform, the more data you will also have to collate and carry out your analyzes.

A personal implication

Peláez points out that these algorithms are complicated, and there need to be people who are specialized in them, who know them and can work them for the tool to fulfil its function. «Our only objective with this project is to help, collaborating in the investigation of rare diseases. And we do it at no cost to Foundation29 », he stresses.

It does recognize, however, that the personal proximity that existed between the creator of the foundation and them was essential to accept this collaboration. It all started when “they described the problem that rare disease patients have. From there, we tried to find the best algorithm that would help us improve the diagnosis that doctors need, “he explains.

SEE ALSO  This is Z-Library, the largest digital library to read for free

PlainConcpets already had that algorithm from other different projects. That is, it is not created out of anything for rare diseases, although it was modified for the specific needs of the case. “We selected the processing of clinical natural language because we have many projects in this field to understand what people describe when they talk about their symptoms,” says Peláez, who adds that each year the algorithm is improved and significantly. Furthermore, he says that the adaptation of their method to the idea of ​​the Foundation29 was “relatively simple” because it was not the first time that they operated with the health sector.