Double Deep Machine Learning

2022

Epistemic AI accelerates biomedical discovery by finding hidden connections in the network of biomedical knowledge. The Epistemic AI web-based software platform embodies the concept of knowledge mapping, an interactive process that relies on a knowledge graph in combination with natural language processing (NLP), information retrieval, relevance feedback, and network analysis. Knowledge mapping reduces information overload, prevents costly mistakes, and minimizes missed opportunities in the research process. The platform combines state-of-the-art methods for information extraction with machine learning, artificial intelligence and network analysis. Starting from a single biological entity, such as a gene or disease, users may: a) construct a map of connections to that entity, b) map an entire domain of interest, and c) gain insight into large biological networks of knowledge. Knowledge maps provide clarity and organization, simplifying the day-to-day research processes.

2019

This paper proposes a tentative and original survey of meeting points between Knowledge Representation and Reasoning (KRR) and Machine Learning (ML), two areas which have been developing quite separately in the last three decades. Some common concerns are identified and discussed such as the types of used representation, the roles of knowledge and data, the lack or the excess of information, or the need for explanations and causal understanding. Then some methodologies combining reasoning and learning are reviewed (such as inductive logic programming, neuro-symbolic reasoning, formal concept analysis, rule-based representations and ML, uncertainty in ML, or case-based reasoning and analogical reasoning), before discussing examples of synergies between KRR and ML (including topics such as belief functions on regression, EM algorithm versus revision, the semantic description of vector representations, the combination of deep learning with high level inference, knowledge graph completi...

Springer eBooks, 2015

Modern medicine and health care in all parts of our world are facing formidable challenges: exploding costs, finite resources, aging population as well as deluge of big complex, high-dimensional data sets produced by modern biomedical science, which exceeds the absorptive capacity of human minds. Consequently, the question arises about whether and to what extent the advances of machine intelligence and computational power may be utilized to mitigate the consequences. After prevailing over humans in chess and popular game shows, it is postulated that the biomedical field will be the next domain in which smart computing systems will outperform their human counterparts. In this overview we examine this hypothesis by comparing data formats, data access and heuristic methods used by both humans and computer systems in the medical decision making process. We conclude that the medical reasoning process can be significantly enhanced using emerging smart computing technologies and so-called computational intelligence. However, as humans have access to a larger spectrum of data of higher complexity and continue to perform essential components of the reasoning process more efficiently, it would be unwise to sacrifice the whole human practice of medicine to the digital world; hence a major goal is to mutually exploit the best of the two worlds: We need computational intelligence to deal with big complex data, but we nevertheless -and more than ever before -need human intelligence to interpret abstracted data and information and creatively make decisions.

IEEE Internet Computing

The recent series of innovations in deep learning (DL) have shown enormous potential to impact individuals and society, both positively and negatively. The DL models utilizing massive computing power and enormous datasets have significantly outperformed prior historical benchmarks on increasingly difficult, well-defined research tasks across technology domains such as computer vision, natural language processing, signal processing, and human-computer interactions. However, the Black-Box nature of DL models and their over-reliance on massive amounts of data condensed into labels and dense representations poses challenges for interpretability and explainability of the system. Furthermore, DLs have not yet been proven in their ability to effectively utilize relevant domain knowledge and experience critical to human understanding. This aspect is missing in early data-focused approaches and necessitated knowledge-infused learning and other strategies to incorporate computational knowledge. This article demonstrates how knowledge, provided as a knowledge graph, is incorporated into DL methods using knowledge-infused learning, which is one of the strategies. We then discuss how this makes a fundamental difference in the interpretability and explainability of current approaches, and illustrate it with examples from natural language processing for healthcare and education applications.

AI Communications

Current Directions in Biomedical Engineering

For the development, training, and validation of AIbased procedures, such as the analysis of clinical data, prediction of critical events, or planning of healthcare procedures, a lot of data is needed. In addition to this data of any origin (image data, bio-signals, health records, machine states, …) adequate supplementary information about the meaning encoded in the data is required. With this additional information - the semantic or knowledge - a tight relation between the raw data and the human-understandable concepts from the real world can be established. Nevertheless, as the amount of data needed to develop robust AI-based methods is strongly increasing, the assessment and acquisition of the related knowledge becomes more and more challenging. Within this work, an overview of currently available concepts of knowledge acquisition are described and evaluated. Four main groups of knowledge acquisition related to AI-based technologies have been identified. For image data mainly ic...

2007

We begin this article with an introduction of associative concept spaces and their properties and reiterate an earlier proposal for a definition of Medical Informatics. We focus on dynamic representations of knowledge and contrast them with ontological approaches that imply strong commitments to particular conceptualizations of reality. This focus leads naturally to unsupervised machine learning and inductive learning that include what is known in literature as unsupervised classification and grammar induction. Next, we provide an overview of hierarchical representations and propose a unifying view. Further, we review the task of grammar induction and by means of an example we show how the task of unsupervised classification is equivalent to a grammar induction procedure complemented by the use of inverted indexes. In the last section of the article we explore the implications of context-dependent, dynamic representations of knowledge for Medical Informatics and advocate an extension of the concept of evidence in biomedicine. Finally, we acknowledge the major hurdles implied by this proposal, which consist of increased complexity and of the need for privacy and confidentiality.

In this paper we define Clinical Data Intelligence as the analysis of data generated in the clinical routine with the goal of improving patient care. We define a science of a Clinical Data Intelligence as a data analysis that permits the derivation of scientific, i.e., generalizable and reliable results. We argue that a science of a Clinical Data Intelligence is sensible in the context of a Big Data analysis, i.e., with data from many patients and with complete patient information. We discuss that Clinical Data Intelligence requires the joint efforts of knowledge engineering, information extraction (from textual and other unstructured data), and statistics and statistical machine learning. We describe some of our main results as conjectures and relate them to a recently funded research project involving two major German university hospitals.

2016

When making clinical decisions, physicians often consult biomedical literatures for reference. In this case, an effective clinical decision support system, provided with a patient’s health information, should be able to generate accurate queries and return to the physicians with useful articles. Related works in the Clinical Decision Support (CDS) track of TREC 2015 demonstrated the usefulness of knowing patients’ diagnosis information for supporting more effective retrieval, but the diagnosis information is often missing in most cases. Furthermore, it is still a great challenge to perform large-scale automatic diagnosis prediction. This motivates us to propose an automatic diagnosis prediction method to enhance the retrieval in a clinical decision support system, where the evidence for the prediction is extracted from Wikipedia. Through the evaluation conducted on 2014 CDS tasks, our method reaches the best performance among all submitted runs. In the next step, graph structured ev...

AI and Society-the encompassing subject of this journal-is probably illustrated (although not exhausted) by interrogations specifically focused on AI in medicine. Justified by spectacular achievements recently facilitated through applied deep learning methodology (based on neural networks), the "Everything is possible" view dominates this new hour in the "boom and bust" curve of AI performance history. The optimistic view collides head on with the "It's not possible," ascertainments often originating in a skewed understanding of both AI and medicine. The meaning of the conflicting views-whose score reality will eventually settle-can be assessed only by addressing the nature of medicine. Specifically, by answering the question: Which part of medicine, if any, can and should be entrusted to AI-now or at some moment in the future? That both medicine and AI will continue to change goes without saying. Medicine is focused on what is needed to maintain life. It is an endeavor within the larger context of social organization of productive activity, of economic and political interaction, of culture. These and the environment, which used to be acknowledged as the dominant factor before the genetics revolution, affect medicine in all its aspects. AI in medicine is the shorthand for the meeting point between a new technology and what healthcare practitioners are or should be. Specifically:-How to identify the specific talent and dedication healthcare demands-How medical education should be conceived and carried out