London: A team of international researchers has developed an artificial intelligence (AI) tool that can estimate how much extra oxygen a Covid-1 patient may need. In collaboration with NVIDIA, more than 20 hospitals around the world (leaders in AI technology) have tested a new AI-based technology known as federated learning using data from five continents.
This technology uses algorithms to analyze chest X-rays and electronic health information of hospitalized patients with Covid’s symptoms.
The results, published in the journal Nature Medicine, show that it predicts the amount of oxygen needed within 24 hours of a patient’s arrival in the emergency department, with sensitivities greater than 95 percent and 88 percent.
To maintain strict patient confidentiality, patient data was completely anonymized and an algorithm was sent to each hospital so that no data was shared or its location excluded.
Usually in AI development, when you build an algorithm on data from one hospital, it doesn’t work well in another hospital, says Dr. Ittai Dayan of U.S. Mass. General Bingham. By creating test models using federated learning and objective, multimodal data from different continents, we have been able to build it. Generalized models that could help frontline physicians around the world, where experimental algorithms were created.
The study analyzed the results of about 10,000 Kovid patients worldwide.
Professor Fiona Gilbert, a lead researcher at Cambridge University, said federation learning has the transformative power to bring AI innovation into the workflow.
NGIDIA’s Global Head of Medical AIG Flores Mona said the federation allows learning researchers to harness the power of AI to set and collaborate on a new standard for what they can do on a global scale. This will try to make AI a strong model not only for healthcare, but in all industries without sacrificing privacy.