Volume 2 Issue 2: October 2023

Endometriosis is a prevalent disease of the female endometrium which affects women of all ethnicities and has been seen to be most common in the 25–35 years age group. The disease does not have a definitive cure, hence care and management are the essential components towards dealing with the disease. At present, the predominant means towards the diagnosis of the presence of the disease involves different imaging modalities alongside laparoscopy, where the instrumentation is expensive to acquire and requires clinical expertise. Recently, work has been done by an author who leveraged Raman blood spectroscopy alongside machine learning towards an affordable high throughput means towards the prediction of endometriosis.

This work utilises the Raman blood spectroscopy dataset alongside advanced signal processing, machine learning and clinical cybernetics, towards the design of a prediction machine which sits within a clinical framework to facilitate Human-Machine interaction for an enhanced care strategy for patients with endometriosis. The prediction machine is designed to initially predict whether a patient has the disease, and is then followed by the use of unsupervised learning to form an inference means towards predicting the extent of the disease. The results showed that a combination of the adopted methods could allow for a high prediction of the endometriosis disease. Subsequent work in this area would now include further optimisation of the prediction machine in order to potentially maximise the prediction accuracy.

The variation of the contraction force associated with the phantom motion used for the actuation of a bionic upper-limb prosthesis represents a scenario encountered regularly by amputees, while prior research appears to not have been able to succinctly address this problem. In this study, an extended prosthesis control system is proposed which is able to recognise gesture intent motions alongside the prediction of an associated contraction force as part of an advanced pattern recognition system. As part of this research topic, this paper introduces the proposed control architecture and is based on the solving of the gesture recognition problem amidst varying contraction forces for a transradial amputee with a seemingly faint phantom motion.

The work involves the application of a novel decomposition algorithm and the use of a set of computationally effective features, alongside the contrast of the recognition capabilities of the proposed approach using various classification models. The results show an enhanced recognition of gesture motion intent with the use of the decomposition method, despite the faint phantom motion signal from the amputee.