Our societies are dealing with the fast increasing of the elderly population and its consequences in terms of public health sustainability. Elderly tends to suffer from different diseases that affect their autonomy. An early detection of risk factors and the correspondent medical intervention could be determinant to maintain the Quality of Life (QoL) of this social group and retrieve part of the economic impact to public health systems. We rely on the hypothesis that some impairments/diseases can affect to the ambulatory skills of the elderly. This paper presents a methodology for detecting common walking patterns in the elderly according to their physical and cognitive conditions that could help to the early risk detection.