No abstract

Hospital admissions, readmissions, and other cost of healthcare are significantly impacted by the growing number of patients with chronic conditions and the accumulation of healthcare resources. The hormonal imbalance PCOS produced in women in their reproductive years also contributes to a number of chronic ailments, including irregular periods, excessive weight gain, acne, and the growth of facial hair etc. A delayed or missing menstrual cycle brought on by the hormonal imbalance results in infertility. The current approaches and therapies are insufficient for earlier stage diagnosis, especially from their home-centric environment. Till date, no technology has been able to independently identify the presence of PCOS in the ovaries which eventually affects fallopian tube and alert the patient, consulting doctor, or nurse so that the next course of action can be started as soon as possible. Hence, to solve the aforementioned issues, the proposed research processes the information gathered from PCOS patients using a cloud computing platform integrated with medical big data mining and machine learning (ML) algorithms. In this study, a conceptual design is proposed from the perspective of communications engineering. In order to detect Fallopian Tube (FT) activity, the architecture combines an intra-body-based nano-sensor with a body-area network. This network receives data from the intra-body networking and transmits it across the air to the relevant personnel (doctor, nurse, patients). The relationship between feasible information rates, and other key metrics has been investigated through preliminary simulations utilising a particle oriented stochastic simulator. Data from sensor are utilised by the Apache Kafka acts as ingestion tool, then given into cloud service computing architecture wherein Advanced Apriori (AA) algorithm is applied over the data to detect characteristics featuring with strong correlations between them before undergoing CatBoost Decision Tree model for optimised prediction of PCOS. The comparison analysis demonstrates notable results in terms of scalability and computation times with an ideal accuracy range.