DeepBio Unveils AI-Based Cancer Diagnosis and Prognosis Research Results at American Association for Cancer Research Meeting

Five Research Results on Deep Learning Applications for Prostate and Breast Cancer Announced

[Asia Economy Reporter Lee Gwan-joo] DeepBio announced on the 11th that it introduced five of its deep learning studies related to cancer diagnosis and prognosis prediction in an online session at the 2022 American Association for Cancer Research (AACR) Annual Meeting held in New Orleans, USA, from the 8th to the 13th of this month (local time).

The abstracts adopted by DeepBio at this year's AACR include ▲development of a common survival analysis model applicable to various cancer types ▲two studies related to survival analysis of breast cancer patients ▲diagnostic research on prostate cancer using frozen section specimens ▲and research on gene mutation association analysis of prostate cancer lesions.

DeepBio especially presented research results on a common survival analysis deep learning model applicable to various cancer types. In this study, a model trained on pancreatic cancer patient data accurately predicted the prognosis of rectal and breast cancer patients at a satisfactory level, confirming the existence of common histological features that can be used to predict prognosis across various adenocarcinomas. Additionally, the study classifying high-risk and low-risk recurrence groups through cancer pathology image analysis models suggested the potential use of molecular diagnostic tests as a screening tool to predict the necessity of chemotherapy in early breast cancer patients.

Updated research results were also shared in the field of prostate cancer. Using the existing AI for prostate cancer severity diagnosis, a study was conducted to verify performance by analyzing frozen section tissue images following prostate cancer biopsy pathology images. As a result, the company explained that the model detected the severity of prostate cancer within frozen section tissues with 98% sensitivity and selected high-severity cases with a high accuracy of 99%.

Kim Sun-woo, CEO of DeepBio, said, “Sharing the first research and meaningful achievements related to deep learning-based cancer diagnosis and prognosis prediction at this AACR provided an opportunity to establish international evidence for our technology,” adding, “Based on this clinical evidence, we will expand cooperative relationships with domestic and international hospitals, research institutes, and pathology platform companies to establish ourselves as a leading company in digital pathology.”