Caris Life Sciences evaluated 3,958 lung squamous cell carcinoma samples using its proprietary GPSai algorithm, uncovering 123 cases where the tumors originated from other primary sites such as cutaneous, urothelial, head and neck, and thymic cancers. This study highlights the ability of AI-driven molecular profiling to detect misdiagnoses that traditional pathology may overlook.

Among the 123 reclassified cases, GPSai prompted guideline-preferred first-line systemic therapy changes for 88 patients (71.5%), underscoring the direct influence of accurate tumor origin identification on treatment selection and potential patient outcomes. Extrapolating these findings to US incidence rates suggests approximately 1,000 lung SCC misdiagnoses could occur annually without AI assistance.

GPSai is now embedded at no extra cost within Caris’ MI Cancer Seek and MI Tumor Seek platforms, following FDA approval in November 2024 for the first simultaneous WES/WTS tissue-based assay with companion diagnostic indications. This integration reinforces Caris’ strategy of combining high-performance sequencing and AI to enhance precision medicine.

Since January 2024, GPSai has overturned 3,857 diagnoses across various cancer types, demonstrating scalability and continuous improvement in diagnostic accuracy. Caris continues to expand its clinico-genomic database and computing capabilities to refine AI models and deliver actionable insights for clinicians worldwide.