由Mass General Brigham开发的AI工具精确预测了亲密伴侣暴力风险年数,然后使用医疗记录进行诊断。 An AI tool developed by Mass General Brigham accurately predicts intimate partner violence risk years before diagnosis using medical records.

Mass General Brigham的研究人员在NIH的资助下开发了一个AI工具,利用机器学习,通过分析电子医疗记录来预测亲密伴侣暴力的风险。 Researchers at Mass General Brigham, funded by the NIH, developed an AI tool using machine learning to predict intimate partner violence (IPV) risk by analyzing electronic medical records. 聚变模型HAIM的精确度达到88%,并使用结构化和非结构化数据,查明了80.5%的IPV病例,在临床演示前3.7年。 The fusion model, HAIM, achieved 88% accuracy and identified 80.5% of IPV cases up to 3.7 years before clinical presentation, using both structured and unstructured data. 它发现了一些风险因素,如心理健康问题、经常接受急诊检查以及社会贫困,而预防性护理则与较低风险相关联。 It detected risk factors like mental health issues, frequent ER visits, and social deprivation, while preventive care was linked to lower risk. 这个工具不是用来诊断的,而是用来支持早期的预防性筛查,特别是因为许多受害者并不透露虐待情况。 The tool is not for diagnosis but aims to support early, proactive screening, especially since many victims don’t disclose abuse. 限制因素包括潜在的假阴性以及需要更多样化的数据。 Limitations include potential false negatives and the need for more diverse data. 这一方法显示出改善早期干预和健康成果的前景。 The approach shows promise for improving early intervention and health outcomes.