Background: The bone is the most frequent target of metastatic breast cancer.However, the underlying molecular mechanisms are unclear.Though many microarray studies have reported gene signatures that can predict metastasis in breast cancer, predicting cancer metastasis to a specific organ, such as to bone, remains a challenge in cancer study.Methods: Previously we proposed a concept CoMi (Context Specific miRNA activity), which could quantify the regulation activity of miRNA on specific context (CoMi activity) by calculating the statistical difference of the expression level between the miRNAtargets and non-targets in a given context (GO term biological process), and demonstrated its application on predicting prognosis.Here we extend this concept to CoMi module and use it to predict whether the cancer will metastasize to bone.First, CoMi features which are significant associated with the bone metastasis are identified by Cox proportional hazards regression.Based on the significant CoMi features, miRNAs regulate on the same context are regarded as a significant miRNA regulation module.Then each significant CoMi module is used to predict bone metastasis of breast cancer patients by sum up all the risk scores (Cox proportional hazards regression) of CoMi features in the module.Results: We tested the proposed method on a breast cancer cohort contain 855 samples gene expression profiles together with bone metastasis information.Some bone metastasis specific CoMi modules are identified, such as module hemopoiesis, which has a log rank p value of 9.05e-07 in train dataset, 0.0075 in test dataset and 0.0031 in independent test dataset.Considering the hematopoietic stem cell niche is a potential therapeutic target for bone metastatic tumors, our results might facilitate to identify the miRNAs regulation modules involved in hematopoietic stem cell niche.Conclusions: We proposed the concept of CoMi activity module, and identified some CoMi activity modules which could predict bone metastasis in breast cancer.These modules could be used to stratify breast cancer patients in bone metastasis groups and nonbone metastasis groups.Furthermore, the proposed method has the promising application on discover the bone metastasis mechanism of breast cancer .