Deep Learning (DL) is the most popular subset of machine learning, with many applications in healthcare and other areas. On the other hand, electroencephalography has effectively solved different brain activity-related healthcare applications. This non-invasive data collection method can monitor and diagnose brain-based health conditions as it is painless and safe. This chapter discusses different smart healthcare applications of DL. Different DL models applied in the existing literature have been described also, from which future researchers will benefit by having a clear insight into the concerned area. Then a smart health framework has been proposed where step by step-by-step process of designing the system has been presented using the Internet of Things (IoT) combined with DL.
Top1. Introduction
Nowadays, Deep Learning (DL) is the prevalent technique applied in a wide variety of domains, including speech recognition, object recognition, disorder forecasting, bioinformatics, biomedicine, etc. Programs relating to fitness care and health technology are rapidly expanding. Incredible development of data science, Internet of Things (IoT) related gadgets, and good-performing processors operating graphical and tensor processing units are the foremost reasons behind the introduction of DL (Bolhasani et al., 2021).
The adage “Health is Wealth” is very much appropriate in the current world. The fast-paced lifestyle, rising pollution, and the emergence of epidemic and pandemic diseases have resulted in a poor and unhealthy quality of human life. Mental health in this concern is as important as physical health. Brain activity analysis is typically regarded as a crucial area in neuroscience. Keeping mental and cognitive health in consideration, Electroencephalography (EEG) data is gathered noninvasively from people in diverse behavioral circumstances (Nahmias et al., 2020). EEG is a kind of brain signal recording device, activated during electrical activities by neuronic firing and taken by electrode channels located on the participant's scalp. The taken signal is further processed through different stages with the final intention of categorizing the disease (Das & Bhattacharya, 2021; Das et al., 2020).
One of the demanding situations in recognition tasks is understanding powerful representations with consistent performances from EEG signals (Das et al., 2022). DL carried out on EEG data to find patterns associated with different disorders are being utilized in developing diagnostic technique and predicting disease (Aghaeeaval, 2021). To keep away bias, the captured signal can be classified by a DL technique. The quantitative EEG and scientific information comprise prognostic data for patient outcomes. Human emotion should significantly contribute to human-computer interaction with promising programs in Artificial Intelligence (AI). EEG is broadly applied to analyze tense situations like epilepsy, neurodegenerative illnesses, and sleep-related issues (Aqeel et al., 2022). Establishing semantic information sharing among heterogeneous data is a crucial problem in dealing with the incredible capacity of various medical systems (Chong & Ali, 2021). The affected person's status tracking and the EEG processing consequences are shared with healthcare carriers, who can verify the affected person's condition and provide emergency assistance as per the need (Amin, Hossain, Muhammad, Alhussein, & Rahman, 2019).
The next section analyses some existing smart healthcare applications inducing DL techniques. Section 3 step by step presents the proposed work. Section 4 discusses the corresponding experimental results. Section 5 finally concludes the work.