ABSTRACT
The advent of artificial intelligence (AI) and machine learning (ML) in healthcare has been a major value addition and enrichment of the potential of what can be achieved using electronic health records (EHR) including radiology images, laboratory test results, time series data (such as electrocardiogram (ECG), electroencephalogram (EEG), etc.), as well as data collected from sensors like photoplethysmogram (PPG) and pulse oximeter (SpO2). EHR can be more efficiently mined for different healthcare applications including but not limited to prognosis, diagnosis, and treatment. The machine learning technology in healthcare systems is most effective. There are many areas for the study in biomedical engineering, such as studies on genomics and proteins, biomedical imaging, brain– body–machine interface, and healthcare management based on publicly available data, i.e., EHR. This chapter presents a systematic approach on the use of machine learning techniques on large medical data (EHR) for classification, evaluation, and to improve decision making in healthcare system for the betterment of humanity. It may be noted that significant work has been done on wearable devices recently. PPG is widely used for 132various applications. This chapter discusses the use of PPG in biomedical engineering and healthcare applications. One can also investigate the study on data fusion, i.e., heart-rate detection from fusing various physiological parameters (such as those measured using ECG, PPG, and EEG, among others). Working with fused data is challenging, as sampling rate of each sensor varies, and the signal normalization and resampling are part of pre-processing. Machine learning can help address the complexity of prediction in healthcare analytics, whether it is classification or regression, as is evident from wide research in the area. Some of the optimal techniques are also discussed in the chapter.
