Orbit Genomics presented new data on its blood-based lung cancer diagnostic platform at the Labroots Precision Medicine: Genomics, Proteomics & Molecular Diagnostics Virtual Event Series 2026.
The poster, “AI-Enabled Blood Test for Accurate Diagnosis and Subtyping Lung Cancer,” describes Orbit Genomics’ work to develop a blood-based diagnostic test for indeterminate pulmonary nodules (IPNs) while expanding the platform to classify major lung cancer subtypes.
The approach combines genomic analysis of short tandem repeats (STRs) and single nucleotide polymorphisms (SNPs) with machine-learning and AI methods, including decision trees, gradient boosting, neural networks and AI-based image analysis.
In the data presented, the prototype diagnostic distinguished benign from malignant pulmonary nodules with 96% positive predictive value (PPV) and 99% negative predictive value (NPV). The testing set included 138 samples. Preliminary lung cancer subtype experiments classified three major subtypes – adenocarcinoma, carcinoma/squamous and small cell – with 87% PPV and 92% NPV.
The results demonstrate scientific feasibility while also identifying the need for additional samples representing a broader range of lung cancer subtypes to further train and enhance the AI models. Expanded analysis, training and validation are underway with clinical collaborators and support from an NIH Direct to Phase II SBIR grant.
Poster authors: Harold Ray Garner, Kyle Caltrider, Sarah Urfer and Dede Willis, Orbit Genomics, Inc.
View the poster: AI-Enabled Blood Test for Accurate Diagnosis and Subtyping Lung Cancer