Automated electrohysterographic detection of uterine contractions for monitoring of pregnancy: feasibility and prospects.

BACKGROUND: Preterm birth is a major public health problem in developed countries. In this context, we have conducted research into outpatient monitoring of uterine electrical activity in women at risk of preterm delivery. The objective of this preliminary study was to perform automated detection of uterine contractions (without human intervention or tocographic signal, TOCO) by processing the EHG recorded on the abdomen of pregnant women. The feasibility and accuracy of uterine contraction detection based on EHG processing were tested and compared to expert decision using external tocodynamometry (TOCO) . METHODS: The study protocol was approved by local Ethics Committees under numbers ID-RCB 2016-A00663-48 for France and VSN 02-0006-V2 for Iceland. Two populations of women were included (threatened preterm birth and labour) in order to test our system of recognition of the various types of uterine contractions. EHG signal acquisition was performed according to a standardized protocol to ensure optimal reproducibility of EHG recordings. A system of 18 Ag/AgCl surface electrodes was used by placing 16 recording electrodes between the woman's pubis and umbilicus according to a 4 × 4 matrix. TOCO was recorded simultaneously with EHG recording. EHG signals were analysed in real-time by calculation of the nonlinear correlation coefficient H2. A curve representing the number of correlated pairs of signals according to the value of H2 calculated between bipolar signals was then plotted. High values of H2 indicated the presence of an event that may correspond to a contraction. Two tests were performed after detection of an event (fusion and elimination of certain events) in order to increase the contraction detection rate. RESULTS: The EHG database contained 51 recordings from pregnant women, with a total of 501 contractions previously labelled by analysis of the corresponding tocographic recording. The percentage recognitions obtained by application of the method based on coefficient H2 was 100% with 782% of false alarms. Addition of fusion and elimination tests to the previously obtained detections allowed the false alarm rate to be divided by 8.5, while maintaining an excellent detection rate (96%). CONCLUSION: These preliminary results appear to be encouraging for monitoring of uterine contractions by algorithm-based automated detection to process the electrohysterographic signal (EHG). This compact recording system, based on the use of surface electrodes attached to the skin, appears to be particularly suitable for outpatient monitoring of uterine contractions, possibly at home, allowing telemonitoring of pregnancies. One of the advantages of EHG processing is that useful information concerning contraction efficiency can be extracted from this signal, which is not possible with the TOCO signal.

First steps in adaptation of an evidential network for data fusion in the framework of medical remote monitoring.

This paper presents a medical remote monitoring application which aims at detecting falls. The detection system is based on three modalities: a wearable sensor, infrared sensors and a sound analysis module. The sound analysis is presented briefly. The multimodal fusion is made using the Dempster Schaffer theory through Evidential Network. A first evaluation of the use of data mining techniques in order to extract blindly data representatives is proposed. These representatives are used to continuously increase the system performances. The system is evaluated on a local recorded data base.

HoCoS: Home Companion Software. A service oriented solution for elderly home accompanying and remote healthcare monitoring.

The age of the population in all societies around the world is increasing. Elderly people prefer to maintain their independence, their autonomy and live at home as long as possible. We propose as a solution to this issue a Home Companion Software baptized HoCoS. This solution aims to help the elderly with daily life by providing an ergonomic and familiar interface. The second purpose is to integrate transparent remote healthcare monitoring service that ensures elderly security without disturbing the ergonomics of the application. We present service oriented architecture that offers extensibility and interoperability between heterogonous systems in order to combine several technologies and operators. We carried out ergonomic tests on this solution to evaluate its comfort and ease of use.

Audio based surveillance forcognitive assistance using a CMT microphone within socially assistive technology.

This work proposes a system for Acoustic Event Detection and Classification (AEDC) using enhanced audio signal provided by a CMT (Coincidence Microphone Technology) microphone. The CMT microphone through signal processing algorithm provides an enhanced signal in several azimuths with a step of 15 degrees . The AEC module exploits this technology to increase classification performance. The automatic detection system based on DWT uses an adaptive threshold for a different energy level and sampling rate quality. The classification system is based on an unsupervised order estimation of Gaussian mixture model adapted to the variability of sound event acoustic information and the representation cost.

Audio sound event identification for distress situations and context awareness.

In this paper the acoustic event detection and classification (AED/AEC) system developed under European Community's Seventh Framework Companionable project in awareness context is presented. The system relies on the use of Wavelet transform technique for detection and on an unsupervised order estimation of Gaussian mixture model (GMM) arranged in hierarchical form in the aim to improve the recognition accuracy. The results, measured in terms of two metrics (accuracy and error rate) are obtained applying the implemented system in off-line mode of audio analysis form of distress scenarios recorded in this fact.