A new AI model is capable of predicting how 80% of ovarian cancer patients respond to therapy. The model can do so with an accuracy level of 80%, which is strikingly better compared to existing clinical approaches.
AI Model Can Predict Ovarian Cancer Therapy Outcomes
The AI model is called IRON, which stands for Integrated Radiogenomics for Ovarian Neoadjuvant therapy. It was initially developed by the group of Professor Evis Sala, who serves as the Chair of the Catholic University's Diagnostic Imaging and Radiotherapy at the Faculty of Medicine and Surgery and as the Director of the Policlinico Universitario A. Gemelli IRCCS' Advanced Radiology Center.
The model looks into different clinical features among patients, including blood-circulating tumor DNA, tumor markers, and images of the disease obtained via CT scans. It then looks into the likelihood of positive therapy response.
A study then looked into the capacity of the AI model. It covered 134 ovarian cancer patients with high-grade conditions. High-grade ovarian cancer constitutes for roughly 70% to 80% of all ovarian tumors. It is quite aggressive and typically chemotherapy-resistant. At present, predictions for therapy response for this specific tumor only have an accuracy level of 50%.
Moreover, because of its high levels of heterogeneity, there are only a few useful clinical biomarkers for it. This spurred the development of the novel AI model.
As part of the study, a total of 134 patients were covered in two different and independent datasets. The clinicians gathered clinical data and blood biomarkers for all the patients. They also gathered quantitative characteristics of the tumor via CT scans of the metastatic and primary sites of the tumor.
Initially, pelvic, ovarian, and omental locations represented most of the burden of the disease. Omental deposits were revealed to respond better to neoadjuvant therapy when compared to pelvic conditions. The blood biomarket CA-125 and tumor mutations were both found to correlate with the general disease burden prior to therapy and treatment response.
Based on advanced CT scan image analysis, there were six patient subgroups that were found to have characteristics that indicate positive therapy outcomes. All the features of the tumor were used as input data for AI algorithms of the tool. The AI model was further trained and then its effectiveness was tested.
Professor Sala notes that from a clinical standpoint, the framework caters to the need to pinpoint patients that are not likely to have a positive response to neoadjuvant therapy and that could fare better with surgical intervention. The professor concludes that the model could be used for stratifying each patient's risk in further clinical research.
ALSO READ: Women at Higher Risk of Ovarian Cancer Have This Genetic Mutation, Scientists Reveal
Ovarian Cancer
Ovarian cancer generally refers to cancer cell growth that builds up in the ovaries. These cells quickly multiply and attack body tissue that is healthy.
Though early symptoms may not be initially noticeable, they are typically attributed to other conditions that are common when they surface. Symptoms of ovarian cancer include weight loss, abdominal swelling or bloating, feeling full quickly, fatigue, pelvic discomfort, back pain, bowel habit changes, and a need to urinate frequently.
The condition ranks fifth in deaths due to cancer among women. It accounts for more deaths compared to any other cancer involving the reproductive system of women.
Check out more news and information on Medicine & Health in Science Times.
![Earth's Quasi-Moon Kamo‘oalewa Could Originate From Lunar Surface Not Asteroid Belt [Study]](https://1721181113.rsc.cdn77.org/data/thumbs/full/53275/89/56/50/40/earths-quasi-moon-kamo-oalewa-could-originate-from-lunar-surface-not-asteroid-belt-study.png)
