Evolving Predictive Analytics in Healthcare: New AI techniques for real-time interventions
2: College of Computer Sciences and Information Technology, King Faisal University, Saudi Arabia
3: Meta Platforms, Inc., USA
4: Department of Computer Science, Faculty of Information Technology, Haiphong University, Vietnam
A major use of practical predictive analytics in medicine has been in the diagnosis of current diseases, particularly through medical imaging. Now there is sufficient improvement in AI, IoT and data analytics to deal with real time problems with an increased focus on early prediction using machine learning and deep learning algorithms. With the power of artificial intelligence alongside the internet of 'medical' things, these algorithms can input the characteristics/data of their patients and get predictions of future diagnoses, classifications, treatment and costs.
Evolving Predictive Analytics in Healthcare: New AI techniques for real-time interventions discusses deep learning algorithms in medical diagnosis, including applications such as Covid-19 detection, dementia detection, and predicting chemotherapy outcomes on breast cancer tumours. Smart healthcare monitoring frameworks using IoT with big data analytics are explored and the latest trends in predictive technology for solving real-time health care problems are examined. By using real-time data inputs to build predictive models, this new technology can literally 'see' your future health and allow clinicians to intervene as needed.
This book is suitable reading for researchers interested in healthcare technology, big data analytics, and artificial intelligence.
Inspec keywords: diseases; medical image processing; health care; Internet of Things; deep learning (artificial intelligence)
Other keywords: cancer; real-time interventions; medical computing; AI techniques; medical image processing; data analysis; medical information systems; image classification; deep learning (artificial intelligence); health care; Internet of Things; evolving predictive analytics
Subjects: Handbooks and dictionaries; Patient diagnostic methods and instrumentation; Textbooks; Biology and medical computing; Monographs, and collections; Medical administration; General and management topics; Machine learning (artificial intelligence); Computer vision and image processing techniques
- Book DOI: 10.1049/PBHE043E
- Chapter DOI: 10.1049/PBHE043E
- ISBN: 9781839535116
- e-ISBN: 9781839535123
- Page count: 420
- Format: PDF
-
Front Matter
- + Show details - Hide details
-
p.
(1)
-
1 COVID-19 detection in X-ray images using customized CNN model
- + Show details - Hide details
-
p.
1
–19
(19)
Around the middle of December 2019, the outbreak of an epidemic triggered by a virus gripped the city of China, Wuhan. In the first week of January 2020, the World Health Organization (WHO) identified and named the virus SARS-CoV-2, a new type of coronavirus popularly later known as COVID-19. The genesis of COVID-19 traces the fact that it was discovered as a coronavirus in 2019. COVID-19 soon spread worldwide through millions of passengers traveling from Wuhan to other parts of Europe and the USA. COVID-19 assumed the proportions of a pandemic and soon spread across the world. The virus has a typical characteristic of changing or mutating as it infects people. The latest variant, Omicron, is now affecting people and spreading in the starting quarter of 2022. This is the most concerning aspect of its characteristic. Previously, researchers built models using machine learning, such as k-nearest neighbor, support vectors machine, multilayer perceptrons, decision trees, and random forests, and many deep learning models, such as CNN-LSTM, CNN VGG-16, VDSNet, RES NET50, UNET, CSEN, CheXNet, Inception V3, and DenseNet. This paper will include the work done for detecting COVID-19 patients by building customized deep learning models, such as artificial neural network (ANN) and convolutional neural network (CNN). We applied the ANN model after extracting the gray-level co-occurrence matrix features from images, whereas the CNN model was directly applied to the X-ray image dataset. The CNN models are used to analyze the images in which mathematical convolution is performed to extract the features. Various parameters, such as accuracy, sensitivity, recall, precision, and F1-score, were calculated to predict the model's efficiency.
-
2 Introducing deep learning in medical diagnosis
- + Show details - Hide details
-
p.
21
–39
(19)
In various fields, DL, which is a subcategory of artificial intelligence, has made an impact. Medical imaging is among the most significant application fields for DL, in which DL algorithms could be used for the classification of the dataset and prediction of datasets. The discriminative and generative structures of DL are discussed in detail in this chapter. Also discussed is the significance of CNN and its variants in the treatment of neurodegenerative disorders. The use of GPU-automated machines for DL allows medical professionals to analyze images. CNN and its variants, FAST-RCNN, are used to track the progression of disease symptoms in humans. In other words, various CNN algorithms are used to model and predict the possibility of new illness progression due to the patient's current clinical illness state. FAST-RCNN is also used to classify new brain disorders based on the patient's gait features, and various CNN techniques are also used to detect the development of diseases with high accuracy.
-
3 Intelligent approach for network intrusion detection system (NIDS) utilizing machine learning (ML)
- + Show details - Hide details
-
p.
41
–54
(14)
In this vociferous twenty-first century, we are surrounded by technology which has a huge impact on the lives of every generation from a small kid to an old-aged man, all are busy using technology. This amount of human involvement in technology motivates rapid advancement in technology where researchers and engineers put their all efforts into solving problems that will impact the lives of billions of users. These problems are either conspicuous or abstract. Any new advancement for conspicuous problems can be noticed by end-users but the advancement for abstract technology may not be noticed by end-users with such ease but they also play a major role in solving problems that need to be addressed. The diverse concept of future virtual systems, which includes various access systems, recurrence groups, and cells - all of which include inclusion areas in order to help remote administrators with the task of organizing and transmitting. Artificial intelligence (AI) and machine learning (ML) can help remote administrators to beat these difficulties by examining the geographic data, designing boundaries and memorable information to figure the pinnacle traffic, asset use and application types, enhancing and calibrating organize boundaries for limit extension and dispense with inclusion gaps by estimating the obstruction, and utilizing the between site separation data fifth-generation (5G) can be a key empowering influence to drive the ML and AI incorporation into the system edge. The figure underneath shows how 5G empowers synchronous associations with different Internet of Things (IoT) gadgets creating gigantic measures of information. The coordination of ML and AI with 5G multi-get to edge processing empowers remote administrators to offer an elevated level of robotization from the conveyed ML and AI engineering at the system edge, application-based traffic guiding and accumulation across heterogeneous access systems, dynamic system cutting to address shifted use cases with various quality of service necessities and ML/AI as an administration offering for end clients. In this chapter, we address all 5G technology, challenges associated with 5G technology, and the role of AI and ML in 5G technology. IoT will become more fruitful with AI, ML and 5G technology. All these technologies work in an interconnected system over a network because of this their reliability on the network increase. Therefore, in this chapter, we will also discuss how network intrusion detection can be done with the help of AI in a much faster way with better accuracy.
-
4 Classification methodologies in healthcare
- + Show details - Hide details
-
p.
55
–73
(19)
Predictive analytics in healthcare has attained peculiar growth due to the advent of machine learning (ML) techniques. ML and the capability to handle veracious and voluminous datasets produced from different sources contribute to the same. The data is acted upon by various techniques to get it cleansed, normalized, and converted into a form that can be inferred. Once the data is ready for inference, it needs to be classified to get relevant results. On this ground, the classification of medical data is essential for the diagnosis of medical conditions and for treating them appropriately. From a narrow perspective, it could be understood that the most used classification methodologies are logistic regression, naïve Bayes, stochastic gradient descent, K-nearest neighbor, decision tree, random forest, support vector machine, and kernel approximation. Many variants in classification methodologies have also been introduced in recent days by hybridizing them with other optimization techniques. This chapter provides a brief insight into the application of classification algorithms and their relevant variants in classifying diverse medical data based on requirements. This chapter also recommends which classification method shall facilitate a good diagnosis and decisive results in healthcare scenarios. It also provides information on how voluminous data are handled by healthcare applications, which are acted upon by ML approaches. It also discusses the pros and cons of using ML approaches in medical diagnosis. It is intended to discuss specific case studies derived from various reputed research organizations so that the reader shall receive a better understanding of the algorithms discussed in this chapter.
-
5 Introducing deep learning in medical domain
- + Show details - Hide details
-
p.
75
–91
(17)
The buzz words for today, namely artificial intelligence (AI), machine learning (ML), and deep learning (DL), are slowly entering the medical industry over the past decade, which brings technologies and solutions to a change in the structure of the medical field. These technologies are connected, and each one offers something different in the medical field by showing a difference in how medical professionals treat the patient. DL provides the power to transform and deliver a more decadent layer of medical technology solutions. It is progressively available with innovative technologies that have broad applications in the natural world medical field. DL plays a vital role in providing insight to medical professionals, which helps identify diseases at an early stage, thus delivering better tailored and most effective patient care. DL has become a well-known initiative that everyone has an idea about. The AI-DL industry has been developing quickly, which provides some sufficient development opportunities to the medical industry to bring a significant change. According to Gartner, almost 37% of all sectors surveyed use DL in their profession. It has been foreseen that by 2022, around 80% of modern developments will use AI and DL. It has been observed that the year 2021 would bring the most powerful deep understanding and AI trends that might reshape the country's economic, social, and medical domains.
This chapter describes DL, the history of deep knowledge in the medical field, the barriers to deep understanding, and DL opportunities in the medical industry. It also focuses on the various methods or algorithms of DL such as convolutional neural network, deep autoencoder, deep Boltzmann machine, deep belief network in biological systems, medical imaging, and health record and report management. It also discusses the various applications of DL in healthcare and how deep understanding is used in medical image analysis.
-
6 Deep-stacked autoencoder for medical image classification
- + Show details - Hide details
-
p.
93
–116
(24)
Image classification is the most important process in the computer-aided diagnosis system. Feature learning is the major challenge in the classification task where the representation features are refined from the high-dimensional input data. Autoencoder (AE) is a self-supervised neural network that maps the input data with the target output data through the sequence of encoder and decoder layers. It automatically learns the abstract-level representation features with the bottleneck feature of encoder-decoder based neural networks. The encoder layers are trained on various levels in extracting the essential features to represent the image. At last, all levels of encoders are stacked together to form a stacked network. The decoder part is replaced by the final softmax layer for classification. The complete stacked network is fine-tuned for classification in a supervised fashion. In this chapter, a stacked AE model is developed to classify the medical images, particularly skin cancer images, into benign or malignant. From the experimental results, it is noted that the classification accuracy of a stacked AE with fine-tuning is higher than that of a stacked AE without fine-tuning. The experimental results also confirm that the stacked AE model with fine-tuning provides improved results as compared with other classification methods.
-
7 Comparison of machine learning and deep learning algorithms for prediction of coronary heart disease
- + Show details - Hide details
-
p.
117
–142
(26)
The primary aim of this work was to compare the performance of the ML and DL algorithms for the CHD dataset. The dataset holds 4,240 with 16 features. We have almost done the entire process step by step, starting from visualizing the dataset, removing the missing values, followed by selecting the best performing features using the Bourta algorithm, which identified AGE, BP and BMI as the best. Since the number of features was too low we took the top eight features. The next step was to balance the class in the dataset as the class was identified to be imbalanced and hence applied SMOTE method to balance the class, and finally, standard scalar was applied to normalize the dataset. Once all these data preparation and pre-processing methods were done, ML and DL methods, namely, logistic regression, decision tree, k-nearest neighbours, random forest, naïve Bayes, support vector machine and feed-forward neural network was deployed, to evaluate the performance of the models. Several metrics were calculated and compared, from which it was identified that the random forest and neural network methods gave the same results. Although the working methodology and the assumptions made by the algorithms also differ, with these results it could not be concluded that a particular classification method is the best. One of the major drawbacks is that the dataset available does not hold all the features which are mandatory to confirm the disease. The dataset was not approved by any doctors in that field and hence the model developed could not be used for real-time implementation.
-
8 Revolution in technology-enabled healthcare: Internet of Things
- + Show details - Hide details
-
p.
143
–161
(19)
The Internet of Things (IoT) is a humongous network, designed through extended internet connectivity to a physical device. Recently, internet technology has revolutionized the medical sector. This renowned new technology offers new prospects for medical professionals to monitor their patients and collect health matrices for timely intervention and proper management. IoT device collects patient data and feeds the data to a software application where the healthcare professional can view it. IoT with sensors can also interpret the data based on the algorithm and generate an alert so that healthcare professionals can intervene. The implementation of information technology in the field of medicine is a growing niche in the healthcare sector and has vast applications in different subspecialties of medicine. It has huge potential in diagnostics, monitoring of treatment, and follow-up of patient care. Internet of Medical Things (IoMT) applications are constantly evolving as technology becomes more powerful and sophisticated.
Medical informatics reduces the costs involved in treatment, improves its efficiencies, and saves the lives of the people. Although there are many benefits to using Internet of Things (IoT) technology in healthcare settings, there are also challenges in its use, such as the security of patient data, adoption of new technology, and difficulties in implementation. Compared to cost and challenges there are more benefits in the usage of the Internet and its facilities by the patients. This revolution along with IoT is just beginning and many innovative IoMT solutions are on the horizon for the future.
-
9 Smart healthcare monitoring framework using IoT with big data analytics
- + Show details - Hide details
-
p.
163
–183
(21)
Wireless networking systems, including the Internet of Things (IoT), have effectively expanded through different aspects of our lives as costs have decreased and functionality has improved. Healthcare is one of the most rapidly evolving and challenging technology fields. From digital healthcare tracking facilities for elderly people and assisted treatment, IoT would have an effect on the medical environment, from treating and handling chronic conditions to delivering customized medicine. The IoT in medical uses, the different ways in which it has penetrated the healthcare industry, and future developments in its growth such as nano-IoT and bio-IoT are all discussed. When processing and executing data analytics on the large volumes of data generated by wearable body sensor networks, a fusion of cloud and IoT architectures is used to make sensible medical systems that can support real-time applications. Wireless body sensor networks (WBSNs) are wireless communicating instruments that are mounted in or on a patient's body. This chapter discusses the design and deployment of a particular medical application based on a WBSN. The suggested method relies on the patient's heartbeat, plethysmogram, and absolute oxygen ratio data. Using IoT technologies, the recorded data is transmitted from the WBSN to the centralized database. Data recorded are analyzed by exploratory data analysis and data mining algorithms. So, if the patient has any abnormal symptoms then the system network will send the data report to the physician. This has been tested and simulated using TensorFlow and the system's performance is determined on the basis of network resilience, as well as the accuracy of the recorded data in different communication networks, as well as system consistency and efficiency in the management.
-
10 Experimental analysis and investigation of dementia detection framework using EHR-based variant LSTM model
- + Show details - Hide details
-
p.
185
–205
(21)
This chapter introduces a deep learning model for forecasting the progression of dementia disease. Because the condition is fundamentally progressive, the model takes into account the temporal data gathered from the cases. In contrast to existing methodologies, our model can predict the disease's future condition rather than just categorise the state of a present diagnosis. Experimental results have displayed that our prototype outclasses the vast majority of existing techniques. Furthermore, this method contains a table for various data sizes. Simultaneously, the findings demonstrate that the TA characteristic is a substantial predictor of DD progression.
-
11 An intelligent agent-based distributed patient scheduling using token-based coordination approach: a case study
- + Show details - Hide details
-
p.
207
–225
(19)
Patient scheduling is the process of sequencing and scheduling the patients for different resources in the healthcare domain. This problem is highly complex because multiple constraints and multiple objectives need to be achieved. The schedules may be more or less congested based on the number of available resources. In this Covid-19 pandemic situation, we could witness this. In countries like India in particular, demand often exceeds the availability of resources and greatly increases the waiting time of many patients with medical procedures. Based on the World Health Organization statistics, it is observed that only one doctor is available for every 2,500 people in India, which is far below the international standards. Also, the availability of diagnostic equipment like magnetic resonance imaging scan, computerized tomography scan, ultrasound, electrocardiogram, lab, and other facilities like beds, oxygen, and ventilators are not on par with its population size when compared with other developed countries. However, even in this case, effective scheduling could reduce the patient waiting time and improve hospital resource utilization. Hence, in this chapter, inspired by the real needs in hospital environments an attempt has been made to develop algorithms for effective scheduling and to reduce the waiting time of patients and hence saving their life.
This chapter discusses a new intelligent agent-based distributed patient scheduling using a token-based coordination technique to reduce the waiting time of the patients. This novel approach generates efficient scheduling with reduced message passing and computational complexity in a distributed pattern. JADE is used as an implementation tool and the performance is evaluated on the data set with 10 departments and a varying number of patients. It shows that the total weighted waiting time and the total weighted completion time of the agent-based patient scheduling (APS)-token are reduced by 57.91% and 12.57%, respectively, in comparison with APS-auction. The number of late patients is reduced by 35.71 - 39.02%. The total idle time of all the resources is less in the token-based approach. There is a drastic reduction in the execution time in the proposed method. and the average reduction factor of execution time is 34.89%. It clearly shows that the APS-token approach gives better performance in less computation time.
-
12 Internet of Things (IoT) for the efficient healthcare system
- + Show details - Hide details
-
p.
227
–241
(15)
Currently, we have major infections including Coronavirus (COVID-19), Chikungunya virus, Hantavirus, Malaria, blood glucose, chronic diseases like arthritis, hepatitis, cancer, AIDS/HIV, and Typhoid, which are currently widespread medical problems. Healthcare is a necessary module of human needs. It is energizing for personal growth as well as for the improvement of society. As a result, mental and physical well-being is critical because it has a significant impact on human society, public health, and the global economy, all of which are predictable to be pretentious in the near opportunity. Therefore, it is crucial to establish an additional intellectual and methodical application of non-pharmaceutical treatments to make the certain organization of illnesses such as COVID-19, Chikungunya virus, and Hantavirus with the least possible impact on our lives and society. The Internet of Things (IoT) concept necessitates the use of hardware strategies that screen or capture records and are associated with the open or private cloud, allowing them to naturally trigger specific events. Through smart cities, smart healthcare, and smart homes, the IoT technology is transforming our lives and resolving to construct a more appropriate and smarter community. In this chapter, significant emphasis has been given to IoT for an efficient healthcare system. The basic purpose of the chapter is to explore the numerous possibilities of IoT in the field of healthcare system.
-
13 Comprehension of melody representation and speed-up approaches for query by humming system
- + Show details - Hide details
-
p.
243
–260
(18)
Query by humming (QBH) music framework under music information retrieval (MIR) is one of the most vital research domains. Largely, it adopts eminent techniques to identify the songs from the music repository based on the user humming for music retrieval applications. The focus of the suggested research work presented in this chapter is to establish novel techniques for the QBH system to popularise it for MIR applications.
Establishing more generic techniques for music retrieval through the QBH framework is a trivial task because of the inherent challenges like an accurate projection of music melody content using discriminating feature set, design and development of prominent melody matching criteria, indigent humming queries and intricate musical structures.
In the present book chapter, novel techniques are recommended and discussed for representing melody with various melody extraction perspectives. Further, two-speed optimisation approaches are suggested and elaborated.
After analysing the different views of the recommended approaches, an attempt is made to find the proximity with preexisting work. Also, the comprehensive view of all the projected approaches along with preexisting work is illustrated. This effort has instituted many eminent specifics that were not taken into account so far.
In order to comprehend the performance of the approaches, empirical metrics like mean reciprocal rank, mean of accuracy, Top X hit rate and response time are employed. Further, exhaustive experimentation is carried out with the target music fragment datasets and humming query datasets [28].
The book chapter is prepared to present the regular and distinct features of recommended and preexisting techniques. Further, in the chapter detailed experimental analysis of the suggested approaches is illustrated.
-
14 Python for digital health solutions: elevated outcomes
- + Show details - Hide details
-
p.
261
–277
(17)
This chapter elaborates upon the current as well as potential future use-cases of Python-based solutions in various sectors of the healthcare industry such as research and diagnostics. The text discusses the improvements of such solutions made to the current healthcare system as well as the drawbacks of such implementations. Some of the implementations discussed are Python-based patient health portals, disease and hospital-workflow simulations, remote patient-diagnosis systems as well as medical image segmentation to aid in medical diagnosis. Wherever possible, examples of suitable real-world implementations have been included.
-
15 IoT-enabled healthcare - a paradigm shift
- + Show details - Hide details
-
p.
279
–294
(16)
There has been fast growth in the digitization of data, including health data which is termed as the "Internet of Things" or "IoT in Healthcare." In the wake of the COVID-19 outbreak, IoT in healthcare has grown exponentially. Having the correct medical equipment is a must for a successful medical procedure. The Internet of Things (IoT) is well-suited to both performing effective operations and assessing the results of such activities. Patient care during the COVID-19 Pandemic is improved because to the use of IoT. The IoT allows for real-time monitoring of a variety of health conditions. Using a smartphone, smart medical devices are able to transmit the necessary health data to the physician.
The future of IoT in healthcare is being shaped by a convergence of many fast growing technologies. Health care delivery and technology development face a number of obstacles. These include concerns relating to patient privacy and data security; the digital divide; governmental oversight; and the acceptance of new technologies by medical professionals and institutions. The chapter gives a worldwide overview of the IoT's impact on healthcare.
-
16 IoT-based cardiovascular prediction framework using deep learning algorithms
- + Show details - Hide details
-
p.
295
–319
(25)
Cardiovascular diseases (CVDs) are the biggest cause of mortality worldwide. The prediction of cardiovascular problems is a significant problem in the field of clinical data analysis. Heart disease is associated with various risk factors, and it is imperative that accurate, reliable, and reasonable methods be developed for early detection in order to accomplish timely treatment of the disease. Different machine learning (ML) approaches are available to predict CVDs. But they are not enough to predict the diseases effectively and efficiently. Healthcare generates a large amount of data, which makes deep learning algorithms increasingly viable tools for helping with decision-making and forecasting. We have taken note of the many features that have been included in recent versions of the deep learning model. This research framework proposes a new modified ConvNet for predicting cardiovascular problems with precise accuracy and reliability. The dataset contains 14 attributes for analysis. Several promising findings are obtained and verified using an accuracy and confusion matrix. To improve outcomes, the dataset is normalized and certain unnecessary attributes are isolated. The proposed Internet of Things-based framework achieves 95.1 percent accuracy in real-time analysis.
-
17 An intelligent approach using convolutional neural network (CNN) for early detection of melanoma and other skin diseases
- + Show details - Hide details
-
p.
321
–349
(29)
Owing to medical technology's rapid growth, new techniques and algorithms appear every day to enhance our quality of life. Admittedly, skin diseases are more common than other diseases because the skin is easily affected by the environment. Some skin diseases may be caused with no symptoms or only mild skin problems that make people not consider it; however, they may result in severe conditions. The most common problem is the lack of dermatologists and delays in getting appointments. Inaccurate diagnosis can be fatal as most skin diseases appear similar. One of the real-life examples is the skin diagnosis as a sign of someone's severe disease that person had a mole in his leg, but he did not consider that as a symptom. That mole indicated cirrhosis of his liver, which led to death. So, artificial intelligence (AI) can help make the diagnosis automated, faster, and more efficient. The application is based on a pre-trained convolutional neural network (CNN) model to classify skin diseases as well as healthy skin. This work is based on an AI application for skin disease classification using machine learning. The deep learning algorithm in machine learning called CNN was used to build a model for the classification of skin disease. With the advancement of AI, we will build a mobile-based application that captures the patient's skin symptoms to provide the perfect diagnosis. In this application, the user will first sign in or register for the application. The user will then take a photo of the symptom on his/her skin. After that, the image will be processed and classified using the CNN model, where it will be previously trained to a dataset of skin lesions images. The model will recognize the images trained with whereafter the processing; the output will be whether a healthy skin or a disease diagnosis.
-
18 Self-organizing deep learning approach for controlling movements of wheeled apparatus through corneal connotation
- + Show details - Hide details
-
p.
351
–360
(10)
This chapter presents a novel approach for facilitating the patients suffering from developmental coordination disorder (DCD). The motivation for this work is to develop a wheeled apparatus through corneal connotation related to sternly disabled patients facing difficulties in operating their wheelchair despite control operations provided on the handles or any other parts of the wheelchair. The main focus is on patients suffering from a lack of motor skills, e.g. "Dyspraxia." An exemplary implementation described in this chapter relates to "Design and Development of Applications using Human-Computer Interaction," a mechanical and electronic control system in a wheelchair whose operating movements are controlled by corneal association or eyeball movements. More specifically our work relates to the field of image acquisition. The related data are processed through self-organizing deep machine learning of artificial neural networks.
-
19 Prediction of breast tumour outcome to chemotherapy using statistical MR images through deep learning approaches
- + Show details - Hide details
-
p.
361
–378
(18)
Generally, breast cancer is generally of the most common cancers for women nationwide. Several diagnostic imaging tests have been performed recently. As a result, multiple image enhancement methods for diagnosing the progression of tumour tissues during chemotherapy were created. The aim of neoadjuvant chemotherapy (NAC) is to reduce cancerous cells before therapy. The ability to predict NAC behaviour may assist to decrease toxicity and time to psychological recovery. Deep convolution neural network (CNN) computational study of dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI) seems to set a remarkable benefit in determining recipients and no recipient's patients. The objectives of this research are to introduce a new deep learning (DL) model that uses several MRI inputs to predict breast cancer reaction to NAC. For pre-processing and smoothing, an adaptive median filter has been preferred, whereas for segmentation K-means has been used. The best model for predicting pathological complete response (PCR) based on pre- and post-chemotherapy DCE-MRI was justified using a few external cases. The parameters such as accuracy, sensitivity, and specificity were used to evaluate the proposed model's efficiency. The graphical findings indicate that the peripheral area contains the much more significant extracted transforms from non-PCR tumours. Grey-level co-occurrence matrix and linear discriminant analysis have been suggested as feature extraction methods. Long short-term memory has also been suggested as the classification approach, which is more efficient than the other techniques. Use of the previous and first chemical pictures acquired with DCE-MR, the suggested and established CNN model was able to identify PCR and non-PCR patients with significant precision, even with a small training dataset. After further assessment based on additional data, this model may be used in clinical research.
-
20 Risk analysis and prediction of cancer associated with Type II diabetes: a review
- + Show details - Hide details
-
p.
379
–389
(11)
Diabetes has become one of the most fatal disorders owing to the change in lifestyle, dietary habits, and lesser physical activity. According to a recent World Health Organization report, it is claimed that 422 million people are suffering from diabetes. Having said that, the Type II category of diabetes is more fatal in nature as it is characterized by the insulin resistance developed by the body. Moreover, Type II diabetes is known to severely impact the kidneys, eyes, and the heart. Also, a large number of researchers are conducting studies to find out if there is any correlation between diabetes and cancer. In this present work, we present a review of such studies and also present our endeavor toward cancer risk assessment and prediction in diabetes Type II patients.
-
Back Matter
- + Show details - Hide details
-
p.
(1)