Pangaea Data

Pangaea Data operates as a health AI company, providing a software platform to biopharmaceutical companies and healthcare providers. The firm was established in 2018 by Dr. Vibhor Gupta and Professor Yike Guo, who now serve as Director and Chief Science Officer, respectively. The company maintains a global presence with offices in London, South San Francisco, and Hong Kong.

The founders' backgrounds are deeply rooted in life sciences, computer science, and academia. Dr. Gupta has over two decades of experience in the life sciences industry, including roles at Deloitte, Quantum Secure, and Seven Bridges Genomics. His academic work focused on molecular biology and building machine learning models for oncology and infectious diseases. Co-founder Professor Yike Guo is a Professor of Computing Science at Imperial College London and the founding Director of the Data Science Institute at Imperial. The technology behind Pangaea is based on the founders' work over the last 25 years, which has attracted more than $300 million in research funding.

Pangaea's business focuses on identifying untreated and undertreated patients for hard-to-diagnose conditions. These patients, who may be undiagnosed, misdiagnosed, or miscoded, can constitute up to 90% of the patient population for over 7,000 such conditions. The company's clients are pharmaceutical companies and healthcare organizations. Its business model involves providing subscription-based access to its software platform, allowing clients to try the product and demonstrate impact before committing significant budget. This is facilitated through partnerships, such as with Microsoft, which makes the platform available through the Azure marketplace.

The company's core product is an AI platform, named PALLUX, previously known as PIES (Pangaea's Intelligence Extraction and Summarization). This platform uses unsupervised Natural Language Processing (NLP) and Natural Language Generation (NLG) to analyze vast amounts of unstructured and structured text from electronic health records (EHRs), such as doctor's notes and discharge summaries. By emulating a clinician's manual review process against established clinical guidelines, the system identifies patients who meet specific criteria for diseases but are not currently in a treatment pathway. This process is privacy-preserving, as the platform is deployed behind the health system's firewall and does not transact patient data. The explainable nature of the AI, with results tied to clinical guidelines, has been a key factor in its adoption. Clinical studies have shown the platform can find significantly more eligible patients—up to 6 times more in the case of cancer cachexia—with accuracy rates of 85-90%.

Keywords: health AI, clinical data analysis, patient characterization, undiagnosed patients, biopharmaceutical services, healthcare data, electronic health records, natural language processing, unsupervised AI, disease diagnosis, patient identification, real-world data, clinical trial recruitment, precision medicine, federated learning, data privacy, value-based care, population health, care gaps, hard-to-diagnose conditions, PALLUX, Vibhor Gupta, Yike Guo