Background and Objectives: Personalised medicine remains a major challenge for scientists. The rapid growth of Machine learning and Deep learning has made it a feasible alternative for predicting the most appropriate therapy for individual patients. However, the lack of interpretation of their results and high computational requirements make many reluctant to use these methods. Methods: Several Machine learning and Deep learning models have been implemented into a single software tool, SIBILA. Once the models are trained, SIBILA applies a range of interpretability methods to identify the input features that each model considered the most important to predict. In addition, all the features obtained are put in common to estimate the global attribution of each variable to the predictions. To facilitate its use by non-experts, SIBILA is also available to all users free of charge as a web server at https://bio-hpc.ucam.edu/sibila/. Results: SIBILA has been applied to three case studies to show its accuracy and efficiency in classification and regression problems. The first two cases proved that SIBILA can make accurate predictions even on uncleaned datasets. The last case demonstrates that SIBILA can be applied to medical contexts with real data. Conclusion: With the aim of becoming a powerful decision-making tool for clinicians, SIBILA has been developed. SIBILA is a novel software tool that leverages interpretable machine learning to make accurate predictions and explain how models made those decisions. SIBILA can be run on high-performance computing platforms, drastically reducing computing times.
翻译:个人化医学:个人化医学仍然是科学家面临的一项重大挑战。机器学习和深层学习的迅速增长使得它成为预测个人患者最适当治疗的可行替代方案。然而,由于对结果缺乏解释以及计算要求高,许多用户不愿使用这些方法。方法:一些机器学习和深层次学习模型已安装成单一软件工具,SIBILA。模型培训后,SIBILA应用了一系列解释性方法,以确定每个模型认为最重要的投入特征;此外,所有获得的特征都被置于共同之处,用以估计每个变量的全球归属与预测的对比。为方便非专家使用,SIBILIA还免费向所有用户提供SIBILA作为网络服务器在https://bio-hpcum.edubem.edu/sibila/。结果:SIBILA应用了三个案例研究,以显示其在分类和回归问题中的准确性能和效率。前两个案例证明,SIBILA可以准确预测每个变量的准确性。最后一个案例表明,SIBA的准确性解释是SILA的正确性,而SILA的正确性解释过程是用来解释。SILA的正确性数据。S-ILA的正确性,而不断演变的正确性数据是用于精确性能的精确性能的计算。