Lantern Pharma Launches AI Platform for Blood-Brain Barrier Permeability Prediction, Completes Japan Trial Enrollment
Lantern Pharma has announced two major milestones: completion of patient enrollment in Japan for its Phase 2 HARMONIC trial of LP-300 in never-smokers with non-small cell lung cancer (NSCLC), and the public release of predictBBB.ai, a real-time AI platform for predicting blood-brain barrier (BBB) permeability of small molecules.
The HARMONIC trial enrolled ten patients ahead of schedule at five Japanese sites, including the National Cancer Center in Tokyo, marking continued progress in the trial’s global expansion across the U.S., Japan, and Taiwan—regions with higher prevalence of never-smoker NSCLC. According to Lantern, never-smokers represent 33–40% of NSCLC cases in Japan, compared to 15–20% in the U.S.
The company, based in Dallas, develops oncology therapeutics using its RADR platform, which leverages over 100 billion oncology-specific data points and machine learning to predict treatment responses and uncover viable drug candidates.
HARMONIC is a multicenter, open-label, randomized Phase 2 trial evaluating overall and progression-free survival in never-smoker patients with relapsed advanced NSCLC following tyrosine kinase inhibitor (TKI) therapy. The trial compares standard chemotherapy (carboplatin/pemetrexed) alone versus in combination with the investigational drug LP-300, which was approved by the U.S. FDA for investigational use in this study. LP-300 is designed to work alongside chemotherapy by targeting tyrosine kinase gene pathways implicated in tumor growth and spread. NSCLC in never-smokers often involves mutations in TK genes such as EGFR, ALK, ROS, and MET, which drive oncogenesis.
In addition to never-smokers, former smokers with NSCLC exhibiting similar genetic profiles may also be eligible to participate. Lantern’s RADR (Response Algorithm for Drug Positioning & Rescue) platform was used to advance LP-300 by confirming its mechanism of action and identifying combination synergies. RADR integrates oncology datasets and machine learning tools to predict drug response, drawing from public databases, commercial clinical trials, and proprietary 3D tumor models.
See also: How AI Enables Precision Oncology

Image credit: Lantern Pharma. The company’s AI-driven drug discovery platform RADR workflow
In a previously reported lead-in cohort of seven U.S. patients, the trial showed an 86% clinical benefit rate and 43% objective response rate. One patient—a 70-year-old never-smoker with advanced NSCLC—achieved a complete response in lung and adrenal lesions and has sustained that response for nearly two years across 21 treatment cycles, with no dose-limiting toxicities or significant adverse events reported.
Lantern plans to release additional trial data from both U.S. and Asian cohorts by the end of Q3 2025. The company estimates the market for treating never-smoker NSCLC at over $4 billion annually. Currently, there are no therapies approved specifically for this subset of lung cancer patients.
Launch of predictBBB.ai: Open-Access AI Module for CNS Drug Development
Alongside its clinical progress, Lantern has launched predictBBB.ai, a public AI tool for real-time prediction of blood-brain barrier (BBB) permeability in small molecules. The system achieves 94% accuracy, 95% sensitivity, and 89% specificity, powered by ensemble learning applied to a proprietary molecular features data lake containing billions of datapoints across millions of compounds.
Crossing the BBB remains a major bottleneck in CNS drug development—only 2–6% of small-molecule drugs typically succeed. Lantern’s new platform addresses this challenge with real-time SMILES-based analysis using multiple models (logistic regression, random forest, SVM, deep neural networks). The system has already been applied to CNS oncology programs at Lantern and its partner Starlight Therapeutics.
predictBBB.ai holds five of the top eleven positions on the Therapeutic Data Commons Leaderboard, outperforming academic and commercial benchmarks. The system’s release is part of a broader freemium strategy designed to encourage early adoption and collaborative use among pharma, biotech, and academic researchers. A white paper and model demo are available at www.predictBBB.ai.
The platform architecture supports expansion into dozens of future modules addressing additional drug development properties, drawing on the same ensemble learning framework and Lantern’s molecular data lake. These tools are expected to include both general-purpose property predictors and therapeutic area-specific analytics.
Lantern holds a PCT patent application (PCT/US2024/019851) for the system, with a favorable search report and 20-year coverage from filing. The company describes this module as the first in a suite of AI-driven drug development tools planned for release.
Topics: Clinical Trials