The goal is to facilitate disease detection, increase early diagnosis and transform the management of medical conditions in the future.
A joint project between Einstein and Siemens Healthineers aims to develop a prototype capable of offering decisive support to radiologists and neurologists in helping to identify multiple sclerosis. The initiative, which has been in place for about three months, should make it possible to refer patients for appropriate treatment in the early stages of the disease.
Multiple sclerosis is a chronic neurological condition caused by inflammation and degeneration in the neurons of the central nervous system. In Brazil, there are approximately 40,000 cases, an average of 15 cases per 100,000 inhabitants, according to the International Federation of Multiple Sclerosis and the World Health Organization. The disease has no cure and usually affects young people, mainly women, between the ages of 20 and 40. Treatment consists of alleviating symptoms and slowing the progression of the disease, making early detection a fundamental factor.
The initiative, which is expected to last fourteen months, focuses on developing a multimodal artificial intelligence model that integrates data from multiple sources, including magnetic resonance imaging, radiology reports and clinical information from electronic medical records. The tool will generate a probabilistic disease score that will help doctors identify potential cases of multiple sclerosis early.
In this project, organizations are using multimodal machine learning, in which data from different modalities is combined before being used to train artificial intelligence models. The process is similar to the method used by doctors, who use information from multiple sources and modalities before offering a diagnosis. After receiving ethical approval, pre-existing Einstein databases and other open databases will be used, ensuring the diversity and quality of data for training the model.
“This project incorporates the advanced use of artificial intelligence to revolutionize the approach to preventive and diagnostic medicine, especially in complex diseases such as multiple sclerosis,” explains Gilberto Szarf, Manager of Research, Innovation and New Business in Einstein’s Imaging Diagnostics Department. “By improving the accuracy and speed of disease detection, we are not only bringing forward the start of treatments, but also democratizing access to high-quality healthcare. This represents an immense benefit for patients and sets a new standard for the healthcare sector,” adds Rodrigo Demarch, Einstein’s Executive Director of Innovation.
In addition to artificial intelligence, a developed interface will allow the import of manual data and the generation of detailed reports, facilitating clinical decision-making by neurologists. Einstein will be responsible for testing all the functionalities of the software solution within the institutional environment, ensuring that the tool is highly effective and safe.
“The initiative between Siemens Healthineers and Einstein aims to integrate cutting-edge solutions in artificial intelligence and data analytics to improve medical diagnoses and treatments, making them more accurate and effective. The collaboration is a model of what we can achieve when we combine medical expertise with technological innovation,” said Patrick de Faria, Head of R&D LAM at Siemens Healthineers.
Since 2017, Einstein Innovation has been creating customized, high-impact software solutions for the healthcare system, with a commitment to adding value to its partners and the healthcare ecosystem. Also with Siemens, in 2019, “Clinical Auto Coding” was developed, a software that uses natural language processing and machine learning to automatically reference ICD-10 codes based on patient clinical data, especially in oncology. The goal was to reduce errors and costs and increase the efficiency of the process, which is traditionally manual and prone to errors. The project covers the 50 most frequent ICD domains in the oncology area.
EN 
PT 
