Emerging Technologies for Particle Engineering
Dilip M. Parikh in Handbook of Pharmaceutical Granulation Technology, 2021
Artificial intelligence (AI) is defined as the ability of a machine to learn and “think” for itself from experience and perform tasks normally attributed to human intelligence, for example, problem-solving, reasoning, and process understanding. AI also has many subsets like machine learning (ML), which describes the ability of an algorithm to learn with experience. It is the application of targeted statistical techniques that enable machines to improve upon tasks with experience. Machine learning has been used in combination with well-established techniques such as “fuzzy logic” to build a set of rules that allow the equipment to consistently improve its performance against a predefined set of objectives as it gathers data. Deep learning, another subset of AI, is composed of algorithms that permit software to train itself to perform tasks, such as speech and image recognition, by exposing multilayered neural networks to vast amounts of data. AI technologies are widely used in situations where tasks are multidimensional, and where relationships are nonlinear and extremely complex; for example, the underlying relationships between formulation ingredients, process conditions, and drug product quality. AI integrates many branches of statistical and machine learning, pattern recognition, logic, and probability theory as well as biologically motivated approaches, such as neural networks, evolutionary computing, or fuzzy modeling, collectively described as “computational intelligence” [80].
Plans for a school-based project to involve students as teachers, learners, and colleagues of Artificial Intelligence
John A. Bilorusky in Cases and Stories of Transformative Action Research, 2021
This is the story of a still unfolding project of transformative action-and-inquiry, with the ultimate aim of involving school age students in collaborating with Artificial Intelligence (AI). Artificial Intelligence, in this context, comprises computer algorithms that can learn how to make basic decisions. For this action research project, we are not talking primarily about AI “teaching” students, nor are we primarily concerned with students learning technology skills by “using” AI, although that is a secondary objective. Instead, our focus is on affirming and further developing the uniquely human capabilities of students through a collaborative, transformative process of action-and-inquiry. In this process, students will collaborate with one another, be guided by their teachers; and teachers and students, alike, will be guided and advised by an educator expert with the use of AI systems. The process of collaboration, especially between students, will also include one or more AI system(s) as “fellow collaborators” in this process.
Clinical Data Analytics
Arvind Kumar Bansal, Javed Iqbal Khan, S. Kaisar Alam in Introduction to Computational Health Informatics, 2019
The technique is based upon proteome analysis of the serum. Proteome is the entire set of proteins found in the human genome. The technique uses a combination of wet-lab techniques to identify amino-acid sequences, bioinformatics and data-mining techniques to identify probabilistically disease-associated proteins. Laser absorption-based mass-spectrometer based analysis identifies various proteins in the serum. Mass spectrometry can be applied a patient's tissue cells, blood, serum or other body-fluids for a similar analysis. To uncover the differences in mass-spectral pattern of proteins, data-mining techniques are used. The goal is to extract a protein pattern that corresponds to the presence of the disease based upon sensitivity and specificity analysis. The technique is useful for the detection of different types of cancer such as ovarian cancer, prostate cancer, breast cancer, liver cancer and colon cancer. The AI techniques used are decision trees, neural networks, clustering and statistical methods.
Artificial Intelligence (AI) Applications for Age-Related Macular Degeneration (AMD) and Other Retinal Dystrophies
Published in Seminars in Ophthalmology, 2021
Tatiana Perepelkina, Anne B Fulton
Artificial Intelligence (AI) is broadly defined as “hardware or software that exhibits behavior which appears intelligent,” capable of mimicking human cognition such as thinking, learning, and problem solving.1 The term was coined at a Dartmouth workshop in 1956, and since that time AI has emerged as a new branch of computer science. A few years later, machine learning was introduced as a subdivision of AI. Machine learning provides computer systems the ability to learn and improve without being explicitly programmed. These algorithms can access data, sort it, and learn without human intervention, while making predictions and inferences based on trends.1 Deep learning is a subset of machine-learning and refers to multiple layers and sets of algorithms (“neural networks”). Captivatingly, the deep learning approach was inspired by the principles of cognition and data processing performed by the brain. As early as 1943, deep learning was first proposed to model neuronal behavior in the central nervous system.,2,3 Just as a biological neuron receives, combines, and transmits a modified signal to the next pool of neurons, an artificial neural network consists of layers of small abstract units, called neurons, working in a similar way and capable of pattern recognition. The inputs of an outer layer of neurons are attached to the data, and the inputs of the internal layer neurons are attached only to the outputs of neurons in the preceding layers.
The fundamentals of Artificial Intelligence in medical education research: AMEE Guide No. 156
Published in Medical Teacher, 2023
Martin G. Tolsgaard, Martin V. Pusic, Stefanie S. Sebok-Syer, Brian Gin, Morten Bo Svendsen, Mark D. Syer, Ryan Brydges, Monica M. Cuddy, Christy K. Boscardin
AI is not one method but rather an approach to data mining (extracting and discovering patterns in data) and analysis using a wide range of statistical techniques on large and often complex data sets. As illustrated in Table 1, some statistical methods associated with AI have been used in medical education for decades – such as Principal Component Analysis (PCA) and regression models with advanced methods for penalizing multiple comparisons (Parsell and Bligh 1999; Reed et al. 2007). Other AI methods share little resemblance with traditional statistics, for example deep neural networks (DNNs), which are commonly used for data mining or analyzing imaging data. DNNs are the basis for advancements in computer vision and are responsible for remarkable breakthroughs demonstrating super-human diagnostic performance within radiology, dermatology, and pathology (Topol 2019).
Can a machine-learning model improve the prediction of nodal stage after a positive sentinel lymph node biopsy in breast cancer?
Published in Acta Oncologica, 2020
V. Madekivi, P. Boström, A. Karlsson, R. Aaltonen, E. Salminen
The current study presents supportive and new information on the use of statistical methods that could help overcome some of the challenges in nomogram development, such as inferior discrimination in different populations. Previous studies involving machine learning have not compared the results to logistic regression analysis [19–22] or have not found machine learning methods superior to it [9]. In this study, both the machine learning method and the logistic regression analysis performed well in the training set with excellent AUC values but the XGBoost model was better in the validation series. The self-learning features of XGBoost were able to compensate the small patient cohort better than the logistic regression model. This could indicate that the logistic regression model over-fits to the training set and does not generalize well to the validation set. XGBoost on the other hand seemed to learn more general features that worked better in both sets. The advantages of artificial intelligence techniques are especially notable in studies where the amount of data are even larger. The self-learning capabilities of these techniques could prove valuable in finding different kinds of interactions between the explanatory variables.
Related Knowledge Centers
- Artificial Neural Network
- Generative Artificial Intelligence
- Computational Creativity
- Deep Learning
- Knowledge Representation & Reasoning
- Artificial General Intelligence
- Operations Research
- Psychology
- Neuroscience
- Uncertainty