Machine Learning Algorithms Used in Medical Field with a Case Study
K. Gayathri Devi, Kishore Balasubramanian, Le Anh Ngoc in Machine Learning and Deep Learning Techniques for Medical Science, 2022
There are other machine learning algorithms, such as Feature Selection Algorithms, Performance Measures, and Optimization Algorithms. Various other sub-categories of machine learning algorithms are:Natural Language Processing (NLP)Computational intelligenceComputer VisionRecommender SystemsGraphical ModelsReinforcement Learning
Use of Machine Learning in Healthcare
Punit Gupta, Dinesh Kumar Saini, Rohit Verma in Healthcare Solutions Using Machine Learning and Informatics, 2023
Machine learning algorithms are broadly split into supervised and unsupervised learning. Both categories comprise a variety of algorithms that are used to implement mathematical models.Supervised learning comprises labeled training data. They mainly focus on classification and regression problems. Some examples of supervised learning algorithms are: random forest, decision trees, naïve Bayes models, and support vector machine (SVM).Unsupervised learning uses unlabeled data for model training. The most common algorithms for unsupervised learning are K-means clustering and deep learning.
AI/ML in Medical Research and Drug Development
Wei Zhang, Fangrong Yan, Feng Chen, Shein-Chung Chow in Advanced Statistics in Regulatory Critical Clinical Initiatives, 2022
The ITR reframed the traditional subgroup identification problems and shades new lights on personalized medicine. It connects with machine learning through a weighted classification problem. This approach has also been studied independently in computer science and it was referred to as contextual based bandit problem [Li et al., 2010]. The solution belongs to a single step off policy reinforcement learning [Sutton and Barto, 2020]. It is worth to note that the reinforcement learning algorithms are key algorithms in the field of artificial intelligence. As mentioned in Section 6.2, Google's DeepMind has used reinforcement learning algorithm developed Alpha Go and Alpha Zero to beat the best human Go game player in 2016. Applying those algorithms into medical field is not straightforward, and there are a few challenging issues. One question is how we can continue to improve the recommendation engine. Once the ITR algorithm is obtained from a training dataset, it will be a deterministic function conditional on patient's covariate information. To continue to improve the algorithms, some randomness for treatment exploration must be introduced. The epsilon-greedy algorithm, Thompson sampling, and upper confidence bounds are three popular choices. However, in medicine field, it may not be ethnical to allow patients to try some solutions which are known to be risky. Therefore, research on how to balance and quantify individual risk, and build it into exploration phase are needed.
Potential value and impact of data mining and machine learning in clinical diagnostics
Published in Critical Reviews in Clinical Laboratory Sciences, 2021
Maryam Saberi-Karimian, Zahra Khorasanchi, Hamideh Ghazizadeh, Maryam Tayefi, Sara Saffar, Gordon A. Ferns, Majid Ghayour-Mobarhan
Two commonly used categories of machine learning algorithms are supervised and unsupervised algorithms (Figure 2). In supervised learning, a model is developed using a labeled dataset with inputs that are related to a known outcome [6]. An example is the development of a model that relates the characteristics of a person, like height and weight, to a specific outcome, such as the onset of a disease within three years. Once trained, the algorithm will then be able to predict the known outcome in new datasets [6]. Supervised learning algorithms can be further categorized into classification or regression algorithms. Classification uses predefined classes in a training dataset to divide the data into categories (e.g. whether a tumor is cancerous or benign), while clustering uses a mining algorithm to identify similarities between data and cluster the data based on these similarities [2]. The classification method can be derived using one of several algorithms, such as the Nearest neighbors and Naïve Bayes algorithms [2]. In the regression method, the relationship between the variables is obtained by analyzing the data and this is then used to predict the trend in the future [2]. Unsupervised learning uses self-organized algorithms that do not require a labeled dataset, and thus, there are no predefined outcomes. Without any user input, unsupervised learning is used to find unknown patterns or clustering within a dataset [7]. Unsupervised learning can be categorized into clustering or association rule algorithms.
Machine learning algorithms for integrating clinical features to predict intracranial hemorrhage in patients with acute leukemia
Published in International Journal of Neuroscience, 2023
Quanhong Chu, Wenxin Wei, Huan Lao, Yujian Li, Yafu Tan, Xiaoyong Wei, Baozi Huang, Chao Qin, Yanyan Tang
Among the enrolled acute leukemia (AL) patients, only 75 patients had ICH, there were 873 patients without ICH. As shown in Table 1, comparisons of 42 features between AL patients with ICH and without ICH were performed using Mann–Whitney U test, chi-square test or Fisher’s exact test, as appropriate. Unfortunately, those traditional statistical methods have some disadvantages such as imbalanced sample distributions, which seriously affect the statistical power [36]. However, machine learning is a fast-growing field that generates predictive or descriptive models by learning from training data rather than by being rigidly programmed [16]. Machine learning algorithms can be divided into three categories: supervised, unsupervised and semi-supervised. Supervised learning algorithms specialize in two types of problems: regression problems and classification problems, which become more practical to apply in disease risk prediction area [36]. Supervised learning algorithms include several common algorithms such as RF, LR, SVM, KNN, GNB, DT and AdaBoost. In order to identify the precise classifier to reasonably fit the 42 different features, we used different machine learning algorithm models.
Artificial intelligence: improving the efficiency of cardiovascular imaging
Published in Expert Review of Medical Devices, 2020
Andrew Lin, Márton Kolossváry, Ivana Išgum, Pál Maurovich-Horvat, Piotr J Slomka, Damini Dey
ML can be categorized into three learning types: supervised; unsupervised; and reinforcement (Table 1). Supervised learning algorithms use a labeled dataset to predict a desired outcome. This involves the iterative selection, processing, and weighting of individual features to learn the underlying patterns within the data that best fit the given outcome [7]. Supervised learning can be sub-categorized into regression and classification algorithms, which use continuous and categorical output variables, respectively. Regression is a well-known and widely applied supervised ML algorithm, whereby stepwise models are effectively learned. Classification methods test data using support vector machines, random forests, and artificial neural networks to identify the best fit differentiation [7]. Limitations of supervised learning include the need for large training and validation datasets, and the manual labeling of large amounts of data which can be time-consuming. Furthermore, supervised algorithms can only predict known outcomes.
Related Knowledge Centers
- Artificial Intelligence
- Bioinformatics
- Data Mining
- Speech Synthesis
- Learning
- Quantum Machine Learning
- Health Informatics
- Automatic Summarization
- Almeida–Pineda Recurrent Backpropagation
- Dehaene–Changeux Model