Diagnosis of diseases
Client:
Medical organizations. Details are under the NDA.
Medical organizations. Details are under the NDA.
Problem:
Late detection of diseases is one of the reasons complicating the treatment process. For example, according to WHO, about 9 million people die of cancer each year. The main problem is that many cases of cancer are diagnosed too late. Even in countries with optimal health systems and services, many cases are detected late, when treatment success is more difficult.
This problem has to be faced not only in cases of diagnosing cancerous tumors, but also in a number of other tasks, such as: detection of pneumonia, various skin diseases, and others.
Often the diagnosis of these diseases is done on the basis of visual analysis.
Late detection of diseases is one of the reasons complicating the treatment process. For example, according to WHO, about 9 million people die of cancer each year. The main problem is that many cases of cancer are diagnosed too late. Even in countries with optimal health systems and services, many cases are detected late, when treatment success is more difficult.
This problem has to be faced not only in cases of diagnosing cancerous tumors, but also in a number of other tasks, such as: detection of pneumonia, various skin diseases, and others.
Often the diagnosis of these diseases is done on the basis of visual analysis.
Goals
The working hypothesis is that systems based on machine learning and neural networks can analyze images.
Solution:
The introduction of neural networks for diagnosing diseases can provide effective assistance in making medical decisions, improve the quality and accuracy of diagnosis, and reduce the time for examining a patient.
Technologies can be applied when working with various types of information. The most widespread use of neural networks in medicine is in the field of working with images.
Medical organizations provided anonymized data of both sick and healthy people. The data were labeled, namely: they were classified into images with and without disease, areas of anomalies were marked and their classification was given. As a result of the training, the trained neural network learned to detect diseases in images.
The working hypothesis is that systems based on machine learning and neural networks can analyze images.
Solution:
The introduction of neural networks for diagnosing diseases can provide effective assistance in making medical decisions, improve the quality and accuracy of diagnosis, and reduce the time for examining a patient.
Technologies can be applied when working with various types of information. The most widespread use of neural networks in medicine is in the field of working with images.
Medical organizations provided anonymized data of both sick and healthy people. The data were labeled, namely: they were classified into images with and without disease, areas of anomalies were marked and their classification was given. As a result of the training, the trained neural network learned to detect diseases in images.
In addition to working with images, neural networks can analyze the textual description of the problem. Thus, using data on the symptoms of diseases, it is possible to recognize the symptoms of the disease and make a diagnosis.
Anonymized datasets received from medical institutions were used for training.
We were given descriptions of symptoms and diagnoses based on them by professional doctors.
As a result of training, the system revealed a pattern between the wording of symptoms and the diagnoses.
Anonymized datasets received from medical institutions were used for training.
We were given descriptions of symptoms and diagnoses based on them by professional doctors.
As a result of training, the system revealed a pattern between the wording of symptoms and the diagnoses.
