PanGIA Analysis System, a novel machine learning platform for non-invasive diagnosis of multiple cancers through urine

The PanGIA Analysis System (PAS) represents a novel machine learning-driven platform for the interrogation of complex biological systems through biochemical signature profiling. Analogous to advanced language models such as Google Gemini or ChatGPT, PAS employs trained algorithms to interpret multidimensional data derived from biological samples. The system utilizes proprietary hydrogel-based microarray substrates, termed NuTec Slides, designed to capture unbiased biomolecular profiles from diverse liquid matrices. Among these, urine offers a particularly informative yet underutilized medium for assessing physiological and pathological states. In this study, a commercialization-ready prototype of PAS was evaluated for its ability to discriminate urine samples containing cancer-associated analytes from non-spiked controls. First-morning urine from healthy volunteers was pooled and spiked with literature-validated analyte panels representing hematological cancers, breast, bone, and brain cancers. Following incubation of NuTec Slides with both spiked and unspiked samples, heat-based signal development, and scanning, extracted image feature data were analyzed by principal component analysis (PCA). This proof-of-concept study indicates that PAS can distinguish between control and spiked human urine samples containing literature supported analytes. Furthermore, we observe distinct clustering between individual cancers. Conclusion: These findings demonstrate the feasibility of PAS as a non-invasive diagnostic tool for cancer detection using urine-based biomolecular profiling. Continued clinical validation is warranted to establish its broader utility in diagnostics, prognostics, companion diagnostics, and monitoring of minimal residual disease.