Artificial intelligence can reveal rare diseases that are difficult to diagnose and enable people who have suffered from these diseases for many years to make a pilgrimage among experts.
The diagnosis of rare diseases is one of the most complex medical problems to be solved. People in distress often spend years wandering from one expert to another until they hope to get a diagnosis. The application of artificial intelligence can help find ghost diseases that are difficult to name.
The characteristics of rare diseases become the worst enemy of diagnosis. Their low frequency and the variability of their symptoms complicate clinical markers and give them a name for what happened.
Unlike other diseases that usually have more or less clear clinical manifestations, certain biomedical markers that are helpful for diagnosis, or supplementary exploratory tests, this situation does not occur in most rare diseases.
Clinicians usually do not have enough experience to make a diagnosis because a case of this disease has never been found before.
“Diagnostic signs” have left the patient for many years without knowing what is happening to him
Wandering from an expert for ten years
Currently, the average time from onset of symptoms to diagnosis of rare diseases is 5 years, although in many cases it can reach 10 years or more.
Because of the pilgrimage, through different consultations and doctor pilgrimages, many times are called “diagnosis journey”, therefore, countless patients and their families must do before naming what happened.
Rare diseases affect more than 300 million people worldwide
The first artificial intelligence for rare disease diagnosis
A project was launched for the first time that uses artificial intelligence to detect rare diseases.
The first stage focuses on detection Cardiac amyloidosisAlso called stiff heart syndrome. The artificial intelligence of the trial was able to identify 50 cases from a sample of 16,500 heart failure patients.
15% to 20% of affected people do not know that they have the disease
The main challenge in the first phase of the project is the low incidence of amyloidosis, which also has some difficulties in diagnosis, because its symptoms are often confused with other typical age-related diseases.
In fact, 15% to 20% of the affected people do not know that they have the disease.According to researchers FSJD, Cardiac amyloidosis accounts for 3% of total hospitalizations and 2.5% of medical care expenditures.
Artificial intelligence analyzed 10 years of anonymous medical records of approximately 10,000 patients.
Subsequently, big data and machine learning algorithms were used to improve data processing and analysis models, and samples of patients with heart failure (diagnostics related to amyloidosis) were studied.
The 50 cases found in the medical records will be evaluated by medical experts. The results of the algorithm (the ability to learn from new medical records, build predictive models and recognize patterns) must always be confirmed by doctors, who will verify the final opinions as experts. Artificial intelligence-based tools will become decision support systems.