Two newly-published studies highlight some of the ways that artificial intelligence may help with the diagnosis and treatment of malignant mesothelioma.
The first study focuses on machine learning algorithms for mesothelioma diagnosis. Researchers used data from Turkey, which has a high rate of mesothelioma among families.
A separate study from New York’s Mount Sinai Medical Center demonstrates the role that artificial intelligence can play in planning mesothelioma radiotherapy.
Both studies suggest that machine learning may be a valuable tool to help doctors tackle asbestos cancer.
How Does Machine Learning Work?
Artificial intelligence refers to intelligence generated by a computer. Like people, computers have to “learn” before they can provide intelligent answers to medical questions.
In both of the new studies, researchers fed data about mesothelioma patients into computers. They analyzed the answers from the computer program and compared them to answers from human doctors.
Artificial intelligence is not a replacement for a doctor’s knowledge about mesothelioma. But computers can analyze large amounts of data quickly. Using their knowledge to help diagnose or treat mesothelioma could help speed up the process.
In an aggressive cancer like pleural mesothelioma, speed can have a major impact on outcomes.
Putting Machine Learning to Work in Mesothelioma
Avishek Choudhury is the author of the diagnostic study. He researches artificial intelligence in healthcare at the Stevens Institute of Technology. The study appears in the journal Technology and Health Care.
“Diagnosis of mesothelioma takes several months and is expensive,” writes Coudhury. “Given the risk and constraints associated with pleural mesothelioma diagnosis, early identification of this ailment is essential for patient health.”
The study used mesothelioma patient data from Dicle University in Turkey. Choudhury fed the data into multiple artificial intelligence systems. The goal was to confirm the most important factors for the earliest possible diagnosis of mesothelioma.
“C-reactive protein, platelet count, duration of symptoms, gender, and pleural protein were found to be the most relevant predictors that can prognosticate mesothelioma,” writes Choudhury.
Artificial intelligence confirmed that relying on these factors delivers the fastest, most accurate mesothelioma diagnosis.
Planning Mesothelioma Radiotherapy with Artificial Intelligence
The second study focused on a radiotherapy treatment planning tool called RapidPlan.
Pleural mesothelioma grows on the lining around the lungs. These misshapen tumors lie close to critical organs like the heart, stomach, liver, and lungs. Precision is critical to avoid serious side effects and produce the best results.
But precision is challenging. The complexity of radiotherapy planning can delay mesothelioma treatment. RapidPlan uses artificial intelligence to speed up the process.
Mount Sinai researchers used data from 57 mesothelioma patients who had volumetric modulated arc therapy (VMAT). They fed the data into the RapidPlan artificial intelligence system to help it “learn” about mesothelioma. Then they tested the educated system on 23 new mesothelioma patients.
The RapidPlan system reduced radiotherapy planning time for mesothelioma to just 20 minutes. That compared to at least four hours for an experienced treatment planner. Just as importantly, the artificial intelligence plan was less risky and more powerful.
“The RP model for malignant pleural mesothelioma showed improved sparing of critical organs with a reduced treatment planning time and increased prescription dose,” writes lead author Vishruta Dumane.
The study appears in Practical Radiation Oncology.
Choudhury, A, “Predicting cancer using supervised machine learning: Mesothelioma”, June 19, 2020, Technology in Health Care, https://content.iospress.com/articles/technology-and-health-care/thc202237
Dumane, VA, et al, “RapidPlan for Knowledge-Based Planning of Malignant Pleural Mesothelioma”, June 17, 2020, Practical Radiation Oncology, Epub ahead of print, https://www.practicalradonc.org/article/S1879-8500(20)30158-2/pdf