Introduction
Deep Learning applications in the sector of healthcare have been used in the ChatBots, solutions of medical imaging which can easily recognise patterns in the symptoms of patients. However, the deep learning-based algorithm can also further be quite useful in order to recognise cancer types, tumours and other rare diseases nowadays. The most essential Deep Learning applications that have been utilized in healthcare are virtual assistants, ChatBots, image colouring, advertising, robotic and image captioning in the multiple healthcare sectors. The concept of deep learning has been delivering an ability to the healthcare sector in order to analyse the data and information without compromising the core accuracy considering the swiftness. This deep learning in healthcare has been considered as the elegant blend that has been using the architecture of layered algorithmic in order to shift by the data and information at quite an astonishing rate.
The core benefits of this deep learning in the area of healthcare are quite plentiful, accurate, swift and efficient. Here the multiple layers of the network, as well as technology, have been enabling the capability of computing which is quite unprecedented. The deep learning networks have been solving multiple complex issues from the reams of information and data around the healthcare sector. The deep learning applications remain as the core field bursting considering the remarkable innovation as well as a high possibility. It has also been observed that Google has also spent multiple amounts of time that are quite significant for examining the models of deep learning which further can be utilized in order to make stronger predictions. This has been happening around the patients who are hospitalized and supporting the multiple clinicians in order to manage both the data of the patients and their possible results. However, this study will superficially show the initial literature review where the different works of literature provided by the multiple authors will be discussed, and then there will be a brief ethics relevance as well as progress based discussion. There will also be some discussion regarding the technologies, resources utilized here in this paper. Then the research method and work plan will be significantly discussed. Finally, there will be a brief discussion of the overall content and a brief presentation regarding Deep Learning in the healthcare sector.
Aim & objectives:
Aim
The aim of this paper is to specifically show the effective and efficient applications of Deep Learning in the healthcare sector and the way it has been utilised for the benefit of this sector.
SMART Objectives
S (Specific): Deep learning usage and its approaches in the sector of healthcare
M (Measurable): This is measurable as the concept of Deep Learning has been successfully utilising in the healthcare sector
A (Achievable): This is achievable as the concept is quite easier and more convenient to be discussed in this paper
R (Realistic): This is quite realistic as this has been largely used in the healthcare sector for multiple benefits
T (Time-bound): This is time-bound and the decided time that it will take is 7 weeks to be accomplished
Literature review
According to (Miotto et al., 2018), healthcare has entered into a new era where biomedical data plays a vital role and ensures that the patient will get appropriate treatment. This research paper mainly focuses on deep learning and how it can provide opportunities as well as challenges to the healthcare industry. The modern biomedical system includes multiple types of data such as sensor data, images, electronic health records, etc. These types of data are very complex, generally unstructured, and poorly annotated. The traditional statistics learning and data mining approaches incorporate some robust features in order to cluster or predict models. Deep learning technologies integrate new effective paradigms in order to retrieve end-to-end learning models through complex data.
Fig: ANN and deep learning model
Based on the figure mentioned above, an artificial neural network is divided into three layers: input, hidden, and output. In contrast, the deep learning model is divided into n number of layers, which means apart from output and input layers, there can be multiple numbers of feature layers. The Healthcare industry started using deep learning models to handle multiple types of data such as clinical images, electronic health records, Genomics, etc. Deep learning is applied to electronic health records for handling both structured as well as unstructured data. Here structured data indicates data regarding laboratory tests, medications, and diagnosis, whereas unstructured data refers to free-text clinical notes. The main reason for applying deep learning in the place of conventional machine learning is to achieve better results.
According to (Faust et al., 2018), physiological signals are a very important data source that helps in disease detection, treatments, and rehabilitation. These signals are retrieved from sensors that are placed on the skin on an implant to the patient’s body. This research paper incorporates important information regarding physiological signals and also integrates deep learning. Deep learning is one type of machine learning method which contains successive hidden layers. The term “deep” is incorporated because of these hidden layers. Researchers of this research paper analyze 53 papers that are based on deep learning methods and healthcare systems. Deep learning algorithms play a vital role in managing large and diverse datasets and help predict health-related data.
Fig: block diagram of offline decision support system and online decision support system
According to the image, the offline decision support system is divided into three parts such as physiological signals, feature extraction & assessment, and classification & assessment. Similarly, physiological signals are sent to feature extraction, and then it moves on to decision support. Feature extraction of offline systems has a connection with feature extraction of online systems. The classification & assessment section of the offline decision support system is related to the decision support of the online system. According to the researcher, in the future deep learning algorithms can be applied to healthcare systems using physiological signals in order to detect disease in the early stages.
According to (Esteva et al., 2019), deep learning is a subfield of ML or machine learning that enhances computational power and analyzes massive datasets effectively. This research paper provides some relevant information about deep learning and its implementation in the healthcare sector. This innovative technology is able to manipulate and understand data, including speech, languages, and images. The Healthcare sector archives huge benefits of deep learning, and it generates a sheer volume of data. It enhances the capacity of digital record systems and medical devices. Deep learning models handle large datasets by taking the help of specialized computing hardware. These models can accept different data types from heterogeneous healthcare data as input and process those data through n numbers of layers. After the data processing, the system is capable of predicting information. Innovative healthcare devices apply supervised learning to train the dataset. Reinforcement learning is another essential agent which helps to adopt deep learning in the healthcare system and archives remarkable feats. Medical images benefit from object detection and image classification systems of deep learning algorithms. It also plays a vital role in the diagnosis of ophthalmology, radiology, pathology, and dermatology. CNN-based methods need to be implemented in Image-level diagnostics to achieve a high level of success.
According to (Norgeot et al., 2019), machine learning is able to supplement clinical medicines and utilizes electronic health records more effectively.
Fig: Deep learning-based healthcare system
This research paper provides a model which indicates a deep learning-based healthcare system. This system develops a digital knowledge base by considering the patient’s personal, demographic, and clinical data in the first step. It is important to add data as well as outcomes to the knowledge base. The digital knowledge base should also consider past clinical decisions. After developing the digital knowledge base, it is effective to incorporate artificial intelligence analyses. It helps in the patient diagnosis system and plays a vital role in treatment selection. The third and final step of a deep learning-based healthcare system is a clinical decision. It helps to make important clinical decisions and provides recommendations to patients.
According to (Zhou et al., 2020), deep learning plays a vital role in enhancing the capacity of human activity in the healthcare system. Apart from deep learning, this research paper also focuses on internet of things technology to develop the internet of healthcare things.
Fig: semi-supervised deep learning framework
The researchers of this research paper develop an appropriate framework for semi-supervised deep learning, which takes labelled input data. The modern healthcare system can incorporate this framework to diagnose patients and predict healthcare data. This framework is able to take both labelled and unlabeled data. It is essential to pass the input data to the data cleaning process and apply reinforcement learning. Reinforcement learning and distance-based reward generation come under an auto-labelled module. This module passes the data to the LSTM-based classification module.
According to ( Bote-Curiel et al., 2019), in this research paper, the researchers share the information about the application of healthcare in biomedical information. The past decades are the witness to massive development in biomedical information like electronic health records, genomic sequence, as well as biomedical images and signals, which change the healthcare system majorly and increase the development of healthcare. For the biggest advantages of the healthcare system, deep learning provides lots of advantages to handling the large amount of medical data shown in the figure below. However, the physical system can not change the algorithm of the process in the future; the deep learning procedure in the medical field provides better decision strategies for the physician. Deep learning adapts importance for their performance and provides the barriers between diseases and risk factors. Also, the technology provides its capability to gather medical data with minimum time. However, electronic health records are generated across the world, but most can not be considered because of their methodological issues like data inconsistency, data quality, multiple scales, instability, legal issues, and incompleteness. Moreover, it is a very multidisciplinary and complex field evolving signals, physics, system, math, computer science, physiology, biology, instrumentation, medicine ( Bote-Curiel et al., 2019).
Fig:- Biomedical Information managed by deep learning to assists the physician
Citation:- ( Bote-Curiel et al., 2019)
The application of deep learning in healthcare contains different areas like disease prognosis, disease diagnoses, disease prediction, and prevention and also developed tailors health treatment based on the medical life cycle. One of the biggest examples of medical care is the âPrecision Medicine Initiative,â which was promoted in 2015 by US President Obama. Through this project, it is easy to map the human body and also detect the actual diseases of the human body. If the patient is drug-addicted, then the technology also helps to identify the drug level in the patient body. The importance of deep learning technology helps to manage the medical field to achieve clinical information in hospital management. This research paper is appropriate for this project and shares the relevant information, and the information is up to date. The research paper also shares the key features of the present topic. For this reason, it is easy to understand the topic very deeply ( Bote-Curiel et al., 2019).
According to (Waring et al. 2020), this is quite an optimal usage of this deep learning in the sector of healthcare because of its core ability in order to decrease the impact admin at the time of allowing for the multifarious medical professionals so that they can focus upon the best health system. In the United Kingdom, the NHS is quite committed to becoming a great leader in the sector of healthcare that nowadays is powered by Deep Learning (Waring et al. 2020). NHS for certainly has been beleaguered by the effective and efficient cost cutting. It also has the core ability in order to further refine the patient care by the strong usage of an intelligent analysis of this Deep Learning in this current competitive market of all over the world (Waring et al. 2020). The appropriate usage of the Deep Learning solution can further potentially help the healthcare sector with efficiencies in streamlining the patient care. This has been proven as the needed help and support to the multifarious healthcare professionals. By the year 2019, UK Prime Minister Boris Johnson has put money for implementing as well as utilizing the Deep Learning for the healthcare initiatives with ÂŁ249.5 million for the NHS. As per (Waring et al. 2020), the Deep Learning applications in the sector of healthcare have been observed in the usage such as ChatBots, solutions of medical imaging, identification of the patient symptom pattern and so on (Waring et al. 2020). The Deep Learning based algorithm has been identifying the core specifications of the types of cancers, other rare diseases and some types of the pathologies as well. This Deep Learning has been utilizing with computer system aided detection of multiple diseases, research on medical, discovery of drug, diseases those are life threatening like diabetic retinopathy or cancer diseases by the core processes of a medical imaging nowadays (Waring et al. 2020). The medical imaging has been colorized by the usage of this deep learning. The generative network has been framed in order to colorize by both incorporating the semantic and perceptual understanding of the colors class distribution. The robotics have also been using in the healthcare sector with the strong help of this deep learning. These robots have been using in the healthcare sector for carrying goods, answering patientsâ questions, taking patients to the emergency room and so on.
Ethics relevance:
| Sl. No. | Topics learnt | Ethical relevance |
| 1 | The degree of understanding the paper | In this paper the relevant information sources in the section of literature review has provided numerous findings with respect to the given research topic and is appropriate for understanding deep learning in the context of healthcare |
| 2 | Data collection methods | For this paper the collected data sources are found to be relevant with the topic of âDeep learning in healthcare systemsâ. The collected reference materials have provided numerous information and findings in the section of literature review |
| 3 | Biomedical information | In this paper the role of biomedical information is found to be relevant in terms of monitoring the ECG status, daily clinical practices, medical imaging, and so on. All these characteristics can be determined as ethically relevant via deep learning techniques in the field of healthcare management |
| 4 | Data Protection | For the collected information sources with regards to this paper, a security policy will be adopted for the prevention of information misuse within the healthcare context along with the management of the overall biomedical information via deep learning systems approach |
| 5 | Degree of the appropriateness of the paper | This paper is a relevant understanding platform for the various techniques related to the healthcare system via deep learning modules. Also, this paper shares valuable information in terms of clinical information for the hospital management |
| 6 | Data privacy policy | For this paper no data is being altered to compromise the quality of healthcare management with regards to deep learning and ANN networking systems |
Technologies and resources:
This paper has extensively used the technology of âDeep Learning in Healthcareâ. The technology of deep learning can provide a vast range and robust applications via the CAD (Computer-Aided Diagnosis) within the healthcare settings (Qayyum et al, 2020). Hence this technology will be applicable for the vast field of HCI (Human-Computer Interventions) within the healthcare setting. Moreover the application of technology relating to Deep Learning (DL) consists of high-levels of performance including clinical pathology, radiology, ophthalmology, and so on. Medical imaging facilities have been developed by the applicable technology of deep learning, which is being widely used in this paper, and is indicated by the following characteristics as shown in the figure
(Source: Qayyum et al, 2020)
Apart from the application of Deep Learning (DL), this paper has also proposed the concept of ANN which is considered as an Artificial Neural Network having an input, and output. Between the input and output, there is a hidden layer, by which a deep learning model is developed in the field of a healthcare context. The role of CNNs (Conventional Neural Networks) has been proven as the state-of-art technology for the healthcare setting with regards to deep learning systems approach (Qayyum et al, 2020). For the achievement of best results the role of adversarial networks via ML or DL technologies, in the generative characteristics have been employed for the accurate prediction, and diagnosis of diseases within the healthcare setting. By the usage of Deep Learning (DL) technology, it is possible for reconstruction of medical imaging processes, and provides real-time and accurate data with respect to the condition of patients. Also, IoT in healthcare settings can be applied for best results.
Major resources to be used in this research paper are as under
| SL. No. | Resources used in this paper | Description |
| 1 | University Libraries | The relevant information regarding deep learning in the healthcare context can be found from numerous research papers and books from the University libraries. The university library can provide all the necessary information sources with regards to the usage of deep learning and the associated technologies with improved systems within the healthcare setting |
| 2 | Free Online Journals | Free online journals are readily available on the Internet. These journals can provide vast information sources regarding new trends and technologies in the field of healthcare with regards to the machine learning and deep learning systems for a healthcare context |
| 3 | Google Scholar | Google Scholar is considered as one of the most popular search engines for the free access to numerous research papers, and online articles. This paper has collected all of the information sources (section of literature review) from Google Scholar for the identifiable topic âDeep Learning in Healthcareâ |
| 4 | Technical resources | In this paper, the technical resources include the skilled people who are specialized in the field of Information and Communication Technologies (ICTs), Internet of Things (IoT), and so on in the context of the healthcare system via deep learning systems approach. |
| 5 | Project Gutenberg | Most of the relevant information sources, apart from Google Scholar can be found from Project Gutenberg, and is the oldest collection library of more than 50,000 books in various fields. Keyword-based searching can be applicable for searching the relevant topic in Project Gutenberg |
Research work & work plan
Research work
The research method is an important aspect in the research paper to gather information from different sources. Through this method, it is easy to organize the research paper. From the above research aim section, it is identified that the main purpose of this present topic is to understand deep learning technology and analyze the advantages of this technology in the healthcare field. Through the research methodology techniques, the research paper provides different types of evidence that deep learning technology actually works better in the medical field, and physical or hospital management takes different advantages from this technology. The methodology technique divides into two different ways: secondary methodology technique and primary research methodology technology.
Using secondary methodology technique gathers the relevant information from the previous research which are related with the topic, from those research paper it is understood that the deep learning technology provides different types of benefit in the medical field like help in to identify the diseases, drug manufacturing and discover, medical imaging diagnosis, personalized medicine, smart health record, data collection, better radiotherapy, smart electronic health record. The research plan of this research paper is to gather the research papers which are current and then analyze those research papers. The technologies are used in different medical fields for the biggest advantages, for drug discovery. The technology uses one type of sensor to understand the level of drug in the patient body, MRI scan, CT scan and ECG help to identify the dreadful diseases such as brain tumor, cancer, and heart diseases (LĂȘ et al., 2002).
Through this technology, deep learning provides benefits for physicians in the medical field. Using this technology, it is easy to gather relevant data like medical insurance fraud schemes. Through this technique, the medical field uses this technology and successfully achieves the aim. Through the research method, we understand the deep learning technology and the uses of the technology that can be effective for the future development of these research papers. However, it is true that deep learning technology is not perfect for the medical field because when analyzing the research papers, it is identified that the technology can not work all the time perfectly.
The technology provides an error in many cases of the medical field that is the complex situation for the physical. Privacy issues are a major risk in deep learning technology. For this technology to always need high quality of the data, most healthcare failures provide this type of data. Lastly, it is said that this type of risk creates barriers between aim and patients. The research method helps to identify this type of information from the relevant sources. All the information is genuine. Below is the work plan to develop this overall research paper (LĂȘ et al., 2002).
Work plan
Timeline
| Purpose | Week 1 | Week 2 | Week 3 | Week 4 | Week 5 | Week 6 | Week 7 |
| Choose the research topic | |||||||
| Set aim and objective | |||||||
| Review the research papers | |||||||
| Analysis the risk in the project | |||||||
| Mitigate the risk | |||||||
| Analysis the technologies and resources in the topic | |||||||
| Structure the work plan | |||||||
| conclusion |
| Milestones | Dates |
| Start the project | 10/7/2021 |
| Find aim and objectives | 12/7/2021 |
| Choose methodology | 19/7/2021 |
| Gather resources | 23/7/2021 |
| Start literature review | 27/7/2021 |
| End literature review | 10/8/2021 |
| Start developing framework | 11/8/2021 |
| End developing framework | 25/8/2021 |
| End the project | 27/8/2021 |
The timeline in the research paper is used to provide the current status and the time limit to finish the research process. On the above the timeline covers each and every point to complete the research papers. To complete the research paper needs 7 weeks and each week covers one or two purposes. The purposes are: Choose the research topic, Set aim and objective , Review the research papers, Analysis the risk in the project, Mitigate the risk, Analysis the technologies and resources in the topic, Structure the work plan, conclusion.
Conclusion
From the overall study, it can be concluded that this paper has entirely focused on the concepts of DL (Deep Learning) within a healthcare setting. A block diagram of the clinical and decision support system has been identified in this paper along with a deep-learning based clinical healthcare system. Also, this paper has provided a section of literature review with regards to the different findings. A conceptual framework of deep learning in the semi-supervised form has been depicted in this paper. The role of biomedical information will be managed by the relevant characteristics of DL in the healthcare system. This paper has also discussed the ethical relevance, technology and resources, and the detailed timeline. Future work of this paper will focus on the research for latest technologies in the context of healthcare setting.
Reference list
Bhatt, C., Kumar, I., Vijayakumar, V., Singh, K.U. and Kumar, A., 2020. The state of the art of deep learning models in medical science and their challenges. Multimedia Systems, pp.1-15. https://www.researchgate.net/profile/Chandradeep-Bhatt/publication/344628869_The_state_of_the_art_of_deep_learning_models_in_medical_science_and_their_challenges/links/5f85813692851c14bcc3a6d8/The-state-of-the-art-of-deep-learning-models-in-medical-science-and-their-challenges.pdf
Bote-Curiel, L., Munoz-Romero, SGuerrero-Curieses, A., & Rojo-Ălvarez, J. L. (2019). Deep learning and big data in healthcare: A double review for critical beginners. Applied Sciences, 9(11), 2331.https://www.mdpi.com/475502
Esteva, A., Robicquet, A., Ramsundar, B., Kuleshov, V., DePristo, M., Chou, K., Cui, C., Corrado, G., Thrun, S. and Dean, J., 2019. A guide to deep learning in healthcare. Nature medicine, 25(1), pp.24-29. https://www.researchgate.net/profile/Alexandre-Robicquet/publication/330203264_A_guide_to_deep_learning_in_healthcare/links/5db3782f92851c577ec35e39/A-guide-to-deep-learning-in-healthcare.pdf
Faust, O., Hagiwara, Y., Hong, T.J., Lih, O.S. and Acharya, U.R., 2018. Deep learning for healthcare applications based on physiological signals: A review. Computer methods and programs in biomedicine, 161, pp.1-13. http://shura.shu.ac.uk/21073/1/Deep%20learning%20for%20healthcare%20applications%20based%20on%20physiological%20signals%20a%20review.pdf
LĂȘ, J. K., & Schmid, T. (2020). The practice of innovating research methods. Organizational Research Methods, 1094428120935498.https://journals.sagepub.com/doi/abs/10.1177/1094428120935498
Ma, L., Zhang, C., Wang, Y., Ruan, W., Wang, J., Tang, W., Ma, X., Gao, X. and Gao, J., 2020, April. Concare: Personalized clinical feature embedding via capturing the healthcare context. In Proceedings of the AAAI Conference on Artificial Intelligence (Vol. 34, No. 01, pp. 833-840). https://ojs.aaai.org/index.php/AAAI/article/download/5428/5284
Miotto, Riccardo, Fei Wang, Shuang Wang, Xiaoqian Jiang, and Joel T. Dudley. “Deep learning for healthcare: review, opportunities and challenges.” Briefings in bioinformatics 19, no. 6 (2018): 1236-1246. https://academic.oup.com/bib/article/19/6/1236/3800524
Montagnon, E., Cerny, M., Cadrin-ChĂȘnevert, A., Hamilton, V., Derennes, T., Ilinca, A., Vandenbroucke-Menu, F., Turcotte, S., Kadoury, S. and Tang, A., 2020. Deep learning workflow in radiology: a primer. Insights into imaging, 11(1), pp.1-15. https://link.springer.com/article/10.1186/s13244-019-0832-5
Norgeot, B., Glicksberg, B.S. and Butte, A.J., 2019. A call for deep-learning healthcare. Nature medicine, 25(1), pp.14-15. https://drive.google.com/file/d/10NG64hSxBWU43G9MmPVy1WclPuVV-a_J/view
Qayyum, A., Qadir, J., Bilal, M. and Al-Fuqaha, A., 2020. Secure and robust machine learning for healthcare: A survey. IEEE Reviews in Biomedical Engineering, 14, pp.156-180. Available at https://ieeexplore.ieee.org/stamp/stamp.jsp?arnumber=9153891
Tuli, S., Tuli, S., Wander, G., Wander, P., Gill, S.S., Dustdar, S., Sakellariou, R. and Rana, O., 2020. Next generation technologies for smart healthcare: challenges, vision, model, trends and future directions. Internet Technology Letters, 3(2), p.e145. https://scholar.google.com/scholar?output=instlink&q=info:SLR3d6NjlWkJ:scholar.google.com/&hl=en&as_sdt=0,5&as_ylo=2020&scillfp=14701495733798009865&oi=lle
Waring, J., Lindvall, C. and Umeton, R., 2020. Automated machine learning: Review of the state-of-the-art and opportunities for healthcare. Artificial Intelligence in Medicine, 104, p.101822. https://www.sciencedirect.com/science/article/pii/S0933365719310437
Zhou, X., Liang, W., Kevin, I., Wang, K., Wang, H., Yang, L.T. and Jin, Q., 2020. Deep-learning-enhanced human activity recognition for Internet of healthcare things. IEEE Internet of Things Journal, 7(7), pp.6429-6438. https://ntnuopen.ntnu.no/ntnu-xmlui/bitstream/handle/11250/2650589/HAR4IoHT_final_2.pdf?sequence=2
