Wearable devices fitted with various sensors are increasingly being used for the automatic and continuous tracking and monitoring of patients. Only first steps have been taken in the field of psychiatric care, where long term tracking of patient behavior holds the promise to help practitioners to better understand both individual patients, and the disorders in general. In this paper we use topic models for unsupervised analysis of movement activity of schizophrenia patients in a closed ward setting. Results demonstrate that features computed on the basis of this analysis differentially characterize interesting sub-populations of schizophrenia patients. Positive-signs schizophrenia sub-population was found to have high motor richness and low typicallity, while negative-signs patients had low motor richness and lower typicality. In addition we design a classifier which correctly classified up to 80% of the clinical sub-population (f-score=0.774) based on motor features.

The Real-Time Signal Processor (RSP) is a programmable signal processing architecture that was created to provide a quick and economical way to implement signal processing applications. The objectives that were chosen to meet these goals were that the RSP be easy to program, suitable for LSI implementation, and conveniently connectable into distributed systems. It was also intended that the RSP would be able to capitalize on the Reduced Computational Complexity (RCC) algorithms [1], [2], in order to achieve increased performance. Fabrication of the initial RSP chips was recently announced [3]. In this paper, the basic RSP architecture is given and interesting features of that architecture are highlighted.

In this paper we describe a frequency domain data transmission method to be used for digital data transmission over analog telephone lines which exploits recently derived reduced computational complexity algorithms, such as the Winograd Fourier Transform, to achieve a significantly lower computational rate than comparable time domain QAM modems implemented digitally using signal processing techniques. In addition to the lower computational rate, the proposed method also allows for better channel bandwidth utilization by allowing optimal signal power allocation based on the channel's signal to noise versus frequency characteristics. Experimental results on this method are presented, indicating that it may be possible to send over 10,000 BPS over an unconditioned telephone line while maintaining a 10-5BER.

We present an approach to perform the conversion between two widely used multiplexing techniques in telephony, time division, multiplex (TDM) to frequency division multiplex (FDM), using digital signal processing techniques. By exploiting some results from the theory of computational complexity we reduce considerably the number of computations required. Furthermore the use of only nonrecursive (FIR) filters whose coefficients are expressed in the canonical signed digit code permits a 16 bit implementation requiring no multiplications and only additions with concurrent shifts which further simplifies the required circuitry.