Application of human mesenchymal stem cells represents a promising approach in the field of regenerative medicine. Specific stimulation can give rise to chondrocytes, osteocytes or adipocytes. Investigation of the underlying biological processes which induce the observed cellular differentiation is essential to efficiently generate specific tissues for therapeutic purposes. Upon treatment with diverse stimuli, gene expression levels of cultivated human mesenchymal stem cells were monitored using time series microarray experiments for the three lineages. Application of gene network inference is a common approach to identify the regulatory dependencies among a set of investigated genes. This thesis applies the NetGenerator V2.0 tool, which is capable to deal with multiple time series data, which investigates the effect of multiple external stimuli. The applied model is based on a system of linear ordinary differential equations, whose parameters are optimised to reproduce the given time series datasets. Several procedures in the inference process were adapted in this new version in order to allow for the integration of multiple datasets. Network inference was applied on in silico network examples as well as on multi-experiment microarray data of mesenchymal stem cells. The resulting chondrogenesis model was evaluated on the basis of several features including the model adaptation to the data, total number of connections, proportion of connections associated with prior knowledge and the model stability in a resampling procedure. Altogether, NetGenerator V2.0 has provided an automatic and efficient way to integrate experimental datasets and to enhance the interpretability and reliability of the resulting network. In a second chondrogenesis model, the miRNA and mRNA time series data were integrated for the purpose of network inference. One hypothesis of the model was verified by experiments, which demonstrated the negative effect of miR-524-5p on downstream genes.