CONCLUSION: The individual risk of treatment-relevant toxicity can be predicted using machine learning algorithms. We demonstrate a promising way to improve efficacy and facilitate proactive toxicity management of systemic cancer …