A recent study conducted in China aimed to enhance the diagnosis of ovarian cancer, a challenging task due to the lack of effective biomarkers. The study systematically evaluated the predictive value of routine laboratory tests and developed an artificial intelligence (AI) model to assist in identifying patients with ovarian cancer. The study utilized data from three hospitals in China and employed a multi-criteria decision-making-based classification fusion (MCF) framework to integrate predictions from 20 AI classification models.
The MCF model, which included 52 features such as laboratory tests and age, accurately identified ovarian cancer. It achieved an area under the receiver-operating characteristic curve (AUC) of 0.949 in the internal validation set and AUCs of 0.882 and 0.884 in two external validation sets. The model outperformed traditional biomarkers like CA125 and HE4, especially in early-stage ovarian cancer detection.
The study highlights the potential of routine laboratory tests as biomarkers for ovarian cancer and underscores the importance of AI in improving diagnostic accuracy. The developed MCF model offers a low-cost, accessible, and accurate diagnostic tool for ovarian cancer.
Background
Ovarian cancer is a significant health concern, known for its high mortality rate and challenges in timely diagnosis. Traditional biomarkers like CA125 have limitations in sensitivity and specificity, prompting the exploration of alternative diagnostic approaches. Laboratory tests present a promising avenue due to their accessibility and potential relevance to ovarian cancer.
Methods
The study involved a retrospective multicenter analysis of laboratory tests and clinical features from women with or without ovarian cancer. Researchers looked at 98 different laboratory tests and used a method called multi-criteria decision-making-based classification fusion (MCF) to create an AI model for predicting ovarian cancer.
Findings
The MCF model achieved high accuracy in identifying ovarian cancer, surpassing traditional biomarkers like CA125 and HE4, particularly in early-stage detection. The model’s performance remained consistent across internal and external validation sets, demonstrating its robustness and generalizability.
It offers a valuable diagnostic tool for ovarian cancer, leveraging routine laboratory tests to improve accuracy and accessibility. The study emphasizes the contribution of various laboratory tests beyond traditional biomarkers to ovarian cancer prediction.
Conclusions
The study underscores the potential of AI and routine laboratory tests to enhance ovarian cancer diagnosis. The developed MCF model represents a significant advancement in diagnostic accuracy and offers a practical solution for early detection of ovarian cancer, potentially improving patient outcomes.
It also received funding from multiple sources, including the Ministry of Science and Technology of China and various scientific foundations.
