Machine Learning for Lung Nodule Malignancy Detection. This study, published in JAMA Network Open in January 2026, investigates the use of machine learning algorithms to detect whether lung nodules are malignant, combined with molecular imaging-guided surgical approaches for improved patient outcomes.
Institutional Collaboration. The research was conducted through Penn Medicine and the University of Pennsylvania, utilizing electronic health records from patients who provided consent specifically for research use by Penn Medicine investigators.
Proprietary Algorithm Development. The machine learning algorithm developed in this study represents significant intellectual property - it has a patent pending with the USPTO (US Provisional Patent Application No. 63/291,179) and is currently undergoing FDA approval review before it can be used in routine clinical practice.
Clinical Translation Pathway. The study is notable for its clear pathway toward clinical deployment. Before the algorithm can be used in standard medical care, it must receive FDA approval. The current submission and review process with both USPTO and FDA reflects the rigorous standards required for AI tools in clinical oncology settings.
Electronic Health Record Data. The primary data supporting this research comes from electronic health records at Penn Medicine, which were consented for research use specifically by Penn Medicine investigators. This careful consent structure reflects the sensitivity of patient health data used in machine learning research.
Data Privacy Protections. The research team chose not to make the underlying data publicly available to protect patient privacy and honor the original ethical approval under which consent was obtained. Making the data public without additional consent or further ethical review would compromise patient confidentiality.
Collaborative Access Model. Rather than open data sharing, the team established a collaborative access model. Other researchers interested in performing additional analyses can contact the corresponding author, and studies will be conducted in collaboration with Penn Medicine and the University of Pennsylvania, preserving data security while enabling scientific progress.
Algorithm Availability for Research. While the algorithm cannot be made fully public due to its proprietary status, it is available upon request for legitimate research studies. This balanced approach supports scientific collaboration while protecting the intellectual property developed through this work.
Combined AI and Surgical Approach. The study integrates machine learning-based malignancy detection with molecular imaging-guided surgery, representing a dual-technology approach to improving lung cancer diagnosis and treatment. This combination addresses both the diagnostic challenge of identifying malignant nodules and the surgical challenge of precisely locating and removing them.
Molecular Imaging Guidance. Molecular imaging techniques allow surgeons to visualize tumors with greater precision during surgery, using targeted contrast agents or tracers that highlight cancerous tissue. When combined with AI-based pre-surgical malignancy assessment, this approach could improve the accuracy and completeness of tumor removal.
Clinical Significance for Early Detection. Lung cancer has the highest cancer-related mortality rate globally, largely because it is often detected at late stages. Accurate ML-based identification of malignant nodules on imaging could enable earlier, more confident decisions to proceed to surgery, potentially saving lives through earlier intervention.
Pathway to Standard of Care. The combination of a pending patent, active FDA review process, and publication in a high-impact journal suggests this approach is being developed with a clear intention to integrate it into standard clinical practice once regulatory approval is obtained.