A new method of predicting response to anti-PD-L1 checkpoint inhibitor immunotherapy in Non-Small Cell Lung Cancer stages III and IV patients.
By using Artificial Intelligence and Computer Vision technologies in combination with modern Bioinformatics approach, we developed a predictive biomarker capable of predicting responseto anti-PD-L1 checkpoint inhibitor therapy, with response understood in sense of Response Evaluation Criteria in Solid Tumors (RECIST 1.1). The biomarker is computed based on the data from several distinct sources:
Scanned image of tumor histologic specimen
Information from these sources is automatically combined within the deep-learning algorithm to produce a single-value biomarker score. The resulting value is then compared with the pre-set thresholds and thus the most probable response type is selected.
RNA-seq raw gene counts (.txt file)