USP Electronic Research Repository

Deep Learning for Drawing Insights from Patient Data for Diagnosis and Treatment

Kumar, Dinesh and Sharma, Dharmendra P. (2020) Deep Learning for Drawing Insights from Patient Data for Diagnosis and Treatment. In: Artificial Intelligence Applications in Healthcare Delivery. Routledge, United Kingdom. ISBN 9780367321512

Full text not available from this repository.


The field of Artificial Intelligence (AI) has matured extensively over the last couple of decades. Given the increase in computing power and its availability researchers are able to develop models much easily and crunch data much quickly. Though research is still ongoing on development of newer AI algorithms, focus has shifted in the application of AI into real life problem domains for the challenge on improving quality of human life through incorporation of AI algorithms. For example, in Sourla et al., 2012 identifies the use of AI for monitoring and early notification for patients suffering from heart diseases. Similarly, several AI based systems have been designed as diagnostic tools for chronic illnesses for example diagnosing depression and diabetics. This had been made possible due to the availability of abundant clinical data within such domains. The enormous potential of data science – from improving data collection methods to data pre-processing to data analytics for a better understanding of science behind data and resulting in development and application of AI algorithms for predictive modelling - has enabled medical practitioners and researchers to collect various types of medical data alongside numerical and textual data and use AI’s great potential in the healthcare domain. These analyses include images and videos which are now a source of valuable information for research into areas such as skin lesion classification, depression analysis, MRI segmentation etc. This chapter investigates the application of deep learning algorithms on image based medical datasets particularly for the classification of skin lesion images. We will introduce the datasets available to the research community in this domain and their characteristics. We will also highlight the ongoing research and efforts in development of mobile applications using deep learning AI engines available to the community that can be used as an early warning system for detection of skin cancer. We also demonstrate using \textit{transfer learning} to incorporate newly available data into our classification model. Finally, we will present major technical challenges reasoning from such datasets and how are these addressed in a sample but intractable domain for application. As ongoing research, we will discuss the future directions of deep learning for skin cancer classification providing the approaches and some promising results.

Item Type: Book Chapter
Subjects: Q Science > QA Mathematics > QA75 Electronic computers. Computer science
Divisions: Faculty of Science, Technology and Environment (FSTE) > School of Computing, Information and Mathematical Sciences
Depositing User: Dinesh Kumar
Date Deposited: 27 Nov 2020 02:51
Last Modified: 05 Jul 2021 03:44

Actions (login required)

View Item View Item