In a new study, a team from the Australian National University (ANU), reported that the AI model — called DeepPT — was found to successfully predict responses to therapies for patients with multiple types of cancer, reported Xinhua.
By using DeepPT in combination with a second tool called ENLIGHT to select the most suitable therapy, the patient response rate to cancer therapy increased from 33.3 per cent to 46.5 per cent.
“We know that selecting a suitable treatment for cancer patients can be integral to patient outcomes,” Danh-Tai Hoang, lead author of the study from ANU’s Biological Data Science Institute, said in a media release.
DeepPT builds on previous work by the same researchers to develop an AI tool to help classify brain tumors. Both tools use microscopic pictures of patient tissue called histopathology images.
Compared to current methods of processing complex molecular data to identify suitable therapies — a process that can take weeks — Hoang said using histopathology images is more cost-effective and timely.
“Any kind of delay obviously poses a real challenge when dealing with patients with high-grade tumors who might require immediate treatment,” he said.
The AI model, which was developed in collaboration with scientists from the US National Cancer Institute and pharmaceutical company Pangea Biomed, was trained on over 5,500 patients with 16 common types of cancer, including breast, lung, head and neck, cervical and pancreatic cancers.
