Abstract Background: The clinical intervention relies on molecular testing of primary biomarkers. We developed an AI enabled platform to predict NCCN guidelines recommended hot-spot genomic alterations from the image analysis using H&E slide. Methods: AI engine was trained using clinically observed hot spot genomic alterations present in lung cancer patients as a ground truth based on histological FFPE whole-slide images. The H & E stained histological images of the tumor having matching pathogenic mutations were downloaded from the TCGA database for the training and testing purpose. The program extracted histopathological features, followed by pattern mapping to predict genomics status based on the results of NGS testing from 480 samples. ResNet AI algorithm was trained to predict lung specific hot spot genomic alterations. An H & E stained slide was imaged with a 40x objective lens to obtain a brightfield image as an input data. The input image was segmented into 65 x 103 tiles, each of 64 µm2 in size for an individual pathological class. 80% of the data set was used for training the engine while validation and blinded study was respectively performed on remaining 10% data. Results: Using single blinded clinical samples. AI model analyzed whole slides. Table 1 summarize parameters for clinically significant variants. Higher sensitivity and specificity was observed for variants encoding mutated cell surface proteins compared to variants encoding intracellular mutant targets when compared on similar training and testing parameters. Conclusions: AI-enabled prediction of hotspot genomic alterations resulted comparable to NGS with shortest TAT, cost and resources. It predicted actionable gene alterations with high positive and negative prediction values. Table 1. Gene Sample (n = 480) Evaluation parameters for tested samples Training Testing Sensitivity (%) Specificity (%) PPV (%) NPV (%) F1-score (%) Accuracy (%) KRAS 248 62 100 94 94 100 97 97 ALK 16 4 100 100 100 100 100 100 EGFR 72 18 88 100 100 90 94 94 MET 16 4 50 100 100 66 66 75 ROS 16 4 100 100 100 100 100 100 STK11 88 22 82 73 75 80 78 77 RET 24 6 50 100 100 60 50 67 BRAF 14 2 100 100 100 100 100 100 NTRK 92 24 78 56 64 71 70 67 ERBB2 12 4 50 50 50 50 50 50 Citation Format: Gowhar Shafi, Shiva Shivamurthy, Anand Ulle, Chongtham Cha Chinglemba, Sumit Haldar, Fauzul Moubeen, Mohan Uttarwar. AI-enabled prediction of lung cancer specific hot spot gene alterations from histology images. [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2023; Part 1 (Regular and Invited Abstracts); 2023 Apr 14-19; Orlando, FL. Philadelphia (PA): AACR; Cancer Res 2023;83(7_Suppl):Abstract nr 5368.