Biostatistiek versus machine learning: van traditionele predictiemodellen naar geautomatiseerde medische analyse

Biostatistics versus machine learning: from traditional prediction models to automated medical analysis Machine learning is increasingly applied data develop clinical models. This paper discusses the application of in comparison with biostatistical methods. well-suited for structured datasets. The selection variables a model primarily knowledge-driven. A similar approach possible learning. But addition, allows unstructured datasets, which are e.g. derived imaging and written texts patient records. In contrast biostatistics, mainly data-driven. Complex able detect nonlinear patterns interactions data. However, this requires large datasets prevent overfitting. For both external validation developed comparable setting required evaluate model’s reproducibility. not easily implemented practice, since they recognized as black boxes (i.e. non-intuitive). purpose, research initiatives ongoing within field explainable artificial intelligence. Finally, development decision support systems discussed.