In a move that could transform the future of patient monitoring, Cleveland Clinic is partnering with startup Piramidal to develop a massive artificial intelligence model trained on brainwave data to monitor brain health in intensive care units instantly and accurately.
Article contents:
- Introduction: A revolution in brain monitoring
- The problem with current electroencephalogram monitoring
- The AI solution: Instant monitoring
- Piramidal brain foundation model
- Data and training: Building an artificial mind
- Rollout plan and challenges: Accuracy above all
- Future applications and ethics
- Conclusion: The future of healthcare relies on data
Introduction: A revolution in brain monitoring
Cleveland Clinic, one of the world’s leading medical institutions, is collaborating with San Francisco-based startup Piramidal to develop a large-scale artificial intelligence model that will be used to monitor patients’ brain health in intensive care units. Instead of being trained on text like ChatGPT, this system relies on electroencephalography (EEG) data, which records the brain’s electrical activity. Changes in this activity can indicate a serious problem, and this new system aims to detect them in seconds rather than hours.
The problem with current electroencephalogram monitoring
In the ICU environment, doctors search EEG data for evidence of seizures, altered consciousness, or deteriorating brain function. Currently, clinicians rely on continuous EEG monitoring, but they cannot monitor every individual patient in real time. Instead, EEG reports are typically generated every 12 or 24 hours and then analyzed. Manually reviewing a single day’s worth of brain wave data can take anywhere from two to four hours.
Dr. Imad Najm, a neurologist and director of the Epilepsy Center at the Cleveland Clinic Neurological Institute, says: «This type of thing takes a long time. It’s subjective and relies on experience and specialization».
The AI solution: Instant monitoring
The system developed by Cleveland Clinic and Piramidal is designed to interpret continuous streams of EEG data and flag anomalies in seconds so physicians can intervene sooner. Chris Bahuga, chief product officer at Piramidal, says: «Our model plays the role of continuously monitoring patients in the ICU and letting doctors know what is happening to the patient and how their brain health is evolving in real time».
Piramidal brain foundation model
Bahuga and Dimitrios Fotis Sakellariou founded Piramidal in 2023 with the goal of building a brain foundation model—an AI system capable of broadly reading and interpreting neural signals across different individuals. This concept is similar to language foundation models, but instead of understanding language, this model seeks to understand the “language of the brain”.
Data and training: Building an artificial mind
The company built its ICU brain model using publicly available EEG datasets, alongside proprietary data from Cleveland Clinic and other partnerships. Sakellariou says the model incorporates nearly one million hours of EEG monitoring data from “tens of thousands” of patients, both neurologically healthy and unhealthy. Brain activity patterns vary significantly from person to person, so building a foundation brain model requires massive amounts of data to capture common patterns and features.
Sakellariou says: «The beauty of a foundation model is in the same way ChatGPT can generalize text, it can adapt to your tone, it can adapt to your writing style—our model is able to adapt to different people’s brains».
Rollout plan and challenges: Accuracy above all
Currently, the team is using retrospective patient data to fine-tune the model. In the next six to eight months, they plan to test the model in a tightly controlled ICU environment with live patient data and a limited number of beds and clinicians. From there, they aim to slowly roll out the software across the entire ICU. Ultimately, the program will allow a hospital system to monitor hundreds of patients simultaneously.
The slow rollout is intended to minimize false positives and false negatives—instances where the system incorrectly identifies patients who do not have a serious event or fails to catch someone who does. Dr. Najm says the latter scenario in particular is “a big problem that keeps us up at night”. Piramidal did not comment on the model’s current accuracy but said it achieved “human-like” performance.
Future applications and ethics
While Piramidal’s immediate focus is on applying its brain foundation model in the ICU, Sakellariou and Bahuga also want to use it for epilepsy and sleep monitoring. Meanwhile, medical and consumer applications of foundation brain models raise questions about how brain data is used and stored, as well as how and when it is deployed.
Caroline Montojo, president and CEO of the Dana Foundation, a private philanthropic organization dedicated to neuroscience research, says: «Developments like this highlight the need for proactive ethical frameworks that support the responsible development and use of these technologies. It is crucial to bring many different perspectives into the early stages of technology design from multiple disciplines, including ethicists, sociologists, and legal scholars, as well as the lived experiences of patients».
Conclusion: The future of healthcare relies on data
This project represents an exciting intersection of advanced medicine and artificial intelligence. If successful, it will not only save lives in intensive care units but also pave the way for a new era of personalized, data-driven healthcare where artificial intelligence can serve as a digital sentinel constantly monitoring our most complex vital signs: our brain activity.
