Piramidal, a startup founded by Dimitris Sakellariou and Kris Pahuja, aims to revolutionize the analysis of brain scan data using AI models. They have recognized the fragmented nature of EEG technology, which requires specialized knowledge to interpret and is inconsistent across different machines. This inconsistency poses challenges for nurses and doctors monitoring patients with brain disorders. The founders have spent years studying the feasibility of computational tools in neurology and believe that automating the analysis of EEG data can greatly improve patient outcomes.
Sakellariou highlights the need for a more streamlined approach, stating that each time an EEG device is used, the entire system needs to be rebuilt for a specific problem. The variation in EEG systems, hospital IT setups, and data formats further complicates the process. However, Piramidal believes that a foundational model for EEG readings could provide a standardized solution that works across different setups, electrode configurations, and patients.
The founders draw parallels to Meta’s Llama series, which focuses on creating a foundational capability for language understanding. Similarly, Piramidal aims to develop a foundational model for brain activity interpretation. While their progress is not yet published, they claim to have built the foundational model and are in the process of preparing it for deployment in hospitals.
Their first production version of the model will be tested in ICU settings through four pilot programs starting in Q1 of next year. This will serve as a proof of concept for the model’s effectiveness in diverse circumstances. Piramidal’s technology will complement existing patient monitoring, providing valuable insights and potentially improving patient care.
Although fine-tuning for specific applications will be required, Piramidal believes that their model, even in its current state, is incredibly valuable. Sakellariou emphasizes the significance of having a pretrained model when compared to starting from scratch. The company has secured a $6 million seed round led by Adverb Ventures and Lionheart Ventures, with participation from Y Combinator and angel investors. This funding will support the development of the model and the expansion of the team.
In terms of data, Piramidal acknowledges the abundance of open-source data but notes that it is siloed. They are in the process of aggregating and harmonizing this data to create a comprehensive integrated data store. Partnerships with hospitals will also provide vast amounts of training data, potentially surpassing human capability for pattern recognition.
While achieving superhuman capability is a long-term goal, the immediate focus is on improving the quality of care. The ICU pilots will allow for rigorous evaluation and documentation of the technology, further bolstering its credibility. Piramidal’s innovative approach has the potential to revolutionize the analysis of brain scan data, ultimately benefiting patients and healthcare professionals alike.
