Work

Multi-Objective Optimization for Breast Cancer Risk Prediction Models with Particle Swarm Optimization

Year: 2023
Type: article
Abstract: Millions of women around the world are afflicted with breast cancer, which can be fatal if left untreated. The best way to improve patient outcomes is by early identification and precise risk prediction. Through multi-objective optimisation with Particle Swarm Optimisation (PSO), we describe a new method for improving the performance of breast cancer risk prediction models.The richness and diversity of the factors impacting breast cancer risk may not be well captured by the single-objective optimisation techniques used by many traditional risk prediction models. To overcome this shortcoming, we present a multi-objective optimisation system that optimises a number of different metrics all at once. These metrics include sensitivity, specificity, and AUC-ROC. The strategy tries to establish a compromise between model sensitivity and specificity, which is a crucial aspect in clinical decision-making, by optimising various objectives.We test our PSO-based approach to multi-objective optimisation on a dataset with a wide range of clinical, genetic, and lifestyle characteristics, and we compare its performance to that of conventional single-objective optimisation methods. Our experimental results show that the suggested method beats the state-of-the-art methods by a wide margin, as measured by its superior AUC-ROC and comparable sensitivity and specificity.In addition, our method makes it possible to provide a collection of Pareto-optimal solutions, giving doctors multiple options for diagnosing a patient based on their preferences and comfort levels with risk. This leeway allows doctors to make better decisions about their patients' breast cancer risk, which improves both patient care and outcomes.Finally, we show that PSO may be used as a robust and flexible multi-objective optimisation strategy for breast cancer risk prediction models. The findings of this study may lead to more precise and helpful breast cancer risk assessment tools, which could increase diagnosis rates and treatment options for this dreadful illness.
Language: English
Open Access status: diamond
OpenAlex ID: W7218290652
FWCI: 0.802
Cites: 34
Cited by: 7
Related to: 9