Cleveland clinic He works with a startup based on San Francisco Pyramidal to develop a immense AI model, which will be used to monitor the health of patients’ brain in intensive care units.
Currently, doctors rely on constant monitoring of EEG in order to detect abnormal brain activity in a patient with the ICU, but they cannot monitor every patient in real time. Instead, EEG reports are usually generated every 12 or 24 hours, and then analyzed to determine if the patient experiences a neurological problem. Manual checking of the brain wave data can take two to four hours.
“These kinds of things are time consuming. It is subjective and depends on experience and specialist knowledge,” says Iwad Najm, a neurologist and director of the Epilepsy Center at the Institute of Neurology of Cleveland Clinic.
The system developed by Cleveland Clinic and Pyramidal was designed to interpret continuous EEG data streams and flagship irregularities in a few seconds so that doctors can intervene earlier.
“Our model plays this role of continuous monitoring of patients on the ICU and allowing doctors to know what is happening to the patient and how their brain health is evolving in real time,” says the main product officer of the Pyramidal Kris Pahuja product.
Pahuja and the general director of Dimitris Fotis Sakellariou founded the pyramidal in 2023 to build a foundation model for the brain – AI system, which can read and interpret neural signals for different people. Earlier, Sakellariou spent 15 years as a neuroingenior and scientist AI conducting EEG research. Pahuja worked on the product strategy in Google and Spotify. Their startup, which is supported by Y Combinator, collected $ 6 million for seed funds last year.
The company has built its ICU brain model, using publicly available EEG data sets, as well as reserved EEG data from Cleveland Clinic and other partnerships. Sakellariou claims that the model contains almost a million hours of EEG monitoring from “tens of thousands” of patients, both neurologically and unhealthy well. Brain activity patterns are extremely variable from person to person, so building a brain foundation model requires huge amounts of data to capture common patterns and features.
“The beauty of the foundation model is the same as chatgpt can generalize the text, it can adapt to your tone, it can adapt to your writing method – our model is able to adapt to the brains of different people,” says Sakellariou.
Currently, Cleveland Clinic and the Pyramidal Team are using the patient’s retrospective data to adapt the model. Over the next six to eight months, they plan to test the model in a strictly controlled ICU environment with live patients and a confined number of beds and doctors. From there, they are trying to slowly introduce software to the whole ICU. Ultimately, the software will allow the hospital system to monitor hundreds of patients at the same time, says Najm.