Artificial Intelligence Algorithms in Predictive Factors for Hematologic Toxicities During Concurrent Chemoradiation for Cervical Cancer

• Published in: Curēus (Palo Alto, CA) Vol. 16; no. 10; p. e70665
• Main Authors: Petre, Ion, Negru, Serban, Dragomir, Radu, Bordianu, Anca, Petre, Izabella, Marc, Luciana, Vlad, Daliborca Cristina
• Format: Journal Article
• Language: English
• Published: 01-10-2024
• ISSN: 2168-8184; 2168-8184
• DOI: 10.7759/cureus.70665

The most recent research conducted for the International Federation of Gynecology and Obstetrics indicates that, depending on the stage of cervical cancer (CC), several therapies can provide similar overall survival and progression-free survival rates. To determine the hematologic toxicities during concurrent chemotherapy for cervical cancer, we evaluated these two therapies (cisplatin or carboplatin). Hematologic markers have been studied using statistical models and descriptive statistics. Artificial intelligence models were built using the treatment data and all the information gathered from each patient after one or more administrations to forecast the CC stage. The information was gathered from stage III cervical cancer patients and provided by Oncohelp Hospital from the West Region of Romania. Many traditional machine learning techniques, such as naïve Bayes (NB), random forest (RF), decision trees (DTs), and a trained transformer called TabPFN, were used in the current study to obtain the tabular data. The algorithms NB, RF, and DTs yielded the greatest classification score of 100% when it came to cervical cancer prediction. On the other hand, TabPFN demonstrated an accuracy of 88%. The effectiveness of the models was evaluated by computing the computational complexity of traditional machine learning methods. Early detection increases the likelihood of a good prognosis during the precancerous and malignant stages. Being aware of any indications and symptoms of cervical cancer can also help to prevent delays in diagnosis. These hematologic toxicities, which have been demonstrated to grow linearly with lowering hematologic markers below their normal expectations, would significantly impair patients' quality of life.The most recent research conducted for the International Federation of Gynecology and Obstetrics indicates that, depending on the stage of cervical cancer (CC), several therapies can provide similar overall survival and progression-free survival rates. To determine the hematologic toxicities during concurrent chemotherapy for cervical cancer, we evaluated these two therapies (cisplatin or carboplatin). Hematologic markers have been studied using statistical models and descriptive statistics. Artificial intelligence models were built using the treatment data and all the information gathered from each patient after one or more administrations to forecast the CC stage. The information was gathered from stage III cervical cancer patients and provided by Oncohelp Hospital from the West Region of Romania. Many traditional machine learning techniques, such as naïve Bayes (NB), random forest (RF), decision trees (DTs), and a trained transformer called TabPFN, were used in the current study to obtain the tabular data. The algorithms NB, RF, and DTs yielded the greatest classification score of 100% when it came to cervical cancer prediction. On the other hand, TabPFN demonstrated an accuracy of 88%. The effectiveness of the models was evaluated by computing the computational complexity of traditional machine learning methods. Early detection increases the likelihood of a good prognosis during the precancerous and malignant stages. Being aware of any indications and symptoms of cervical cancer can also help to prevent delays in diagnosis. These hematologic toxicities, which have been demonstrated to grow linearly with lowering hematologic markers below their normal expectations, would significantly impair patients' quality of life.